My view, after years of consulting across a wide range of organisations, is pragmatic rather than idealistic. In most businesses, aligning your interventions, whatever form they take, to organisational objectives is the safer choice, and the one more likely to demonstrate value.

The first reason is measurement. Objectives, in any organisation worth its salt, come with clear and agreed measures: revenue, retention, productivity, time-to-competence, customer satisfaction; all measurable, and all open to interrogation. Values rarely have anything like the same clarity. Ask three people in the same business what “integrity” or “courage” looks like in their day-to-day work, and you’ll get three different answers, none of which will be wrong.

The second reason is buy-in. Most organisations have values because organisations are supposed to have values. They were drafted by an external agency, signed off by the board, painted on a wall, and ignored thereafter. The senior leadership team would, if pushed, jettison the whole lot tomorrow if it meant a better quarter. I don’t say that as a criticism; it’s a different set of priorities, and the pressure to perform in most sectors makes it close to inevitable.

A small minority of organisations have done the work properly; consulted their people, built values from the inside out, and aligned the top team behind them. In those places, aligning L&D to values makes complete sense. They are the exception rather than the rule, and the giveaway is almost always that the values were grown internally rather than imported from a consultancy. Culture, after all, emerges from how people work together over time; it isn’t installed.

If you’re in one of those organisations, align to the values. If you’re in the other 90%, align to the objectives, and sleep easier at night.

Leave a comment

1. Audit what you’re doing

Start with a clear-eyed look at your current portfolio. When times are tight, every active project needs a defensible line of sight to an organisational objective; this is not the moment to be carrying passion projects that nobody outside L&D cares about. Some things you’ve built, including ones with provable impact, may not survive the new priorities, and that’s fine; you can bring them back once trust is rebuilt and budgets ease. Our role is to enable performance in service of organisational objectives, so anything that doesn’t serve that aim is hard to defend at the best of times, let alone under pressure.

2. Build less content

We’ve been saying for years that L&D builds too much content, and a tight budget is the moment to do something about it. Does that 20-minute e-learning module need to exist, or could it be a sequence of emails landing across three weeks? Does empathy in the workplace need another full-day workshop, or could it be a conversation with line managers about weaving the message into their morning check-ins? Building less means the content you do produce is the content that earns its place.

3. Open the door to user-generated content

Look around your organisation for the people who know things and like to share them, then give them a way to do it: short videos, written pieces, lunchtime drop-ins, internal webinars, even an internal podcast. This is low-fi work; it doesn’t need to be polished, and the rough edges are part of the appeal. You’ll surface internal expertise that wasn’t being used, without absorbing the full content production cost yourself.

4. Audit your tech stack

New technology arrives with new price tags, but plenty of capability is already inside the tools you pay for, unused. AI is part of that picture, but so are the automations, workflows, and integrations dormant across your existing platforms. Look hard at duplication too; the classic example is Adobe Creative Cloud, where most organisations pay for seats across five or six departments who could comfortably share licences. A conversation with marketing, internal comms, or IT might shift some of that cost off your books entirely.

5. Get sharper at saying no

The other half of the audit is what you accept going forward. New requests will keep arriving, and under budget pressure the temptation is to say yes to everything to look responsive. The discipline worth building is asking, before any commitment, what problem the requester is solving and whether training is the right intervention. A short, well-handled performance conversation that ends with “I don’t think this needs a workshop” saves you the cost of building one and protects the credibility you spend every time you build something that fails to land.

None of these moves are glamorous, and none will turn into a heroic story to tell at the next industry conference; between them, however, they help a team keep doing meaningful work when the money gets harder to find. That’s the job at the moment.

Leave a comment

This isn’t a confession piece, though, or if it is, it’s one where the confession quickly becomes a question. People are tired all the time. We all know they are. The HR conversations about wellbeing, the L&D conversations about engagement, the line management conversations about productivity, they all circle this. Tiredness may be doing more to undermine how our employees work and learn than most of the things we spend our budgets trying to fix.

What tiredness is doing to the people we work with

Stay awake for seventeen hours and performance on tasks needing attention, reaction time, and basic motor control sits roughly at the level of someone with a blood alcohol concentration of 0.05 per cent. Stay awake for twenty-four hours and you’re at 0.10 per cent, which is over the drink-drive limit pretty much everywhere (Dawson and Reid, 1997; Williamson and Feyer, 2000).

Most workplaces have clear, well-policed rules about turning up drunk to work. The same workplaces will routinely schedule international travel, sandwich a 21:00 client dinner between two early-start days, or send an email at 23:00 expecting a thoughtful reply by 08:00, and call the resulting state of their employees something more flattering. Resilience, perhaps; commitment, even.

What’s going on inside a tired head is more interesting than the resigned “I’m shattered” suggests. The prefrontal cortex, the bit of brain you’re using right now to follow this sentence and to inhibit the impulse to check your phone, starts to power down. Connectivity between the prefrontal cortex and the amygdala, the small structure that handles emotional reactivity, weakens. Emotional responses become larger, top-down regulation of those responses becomes smaller, and patience for ambiguity or other people’s small errors collapses (Yoo et al., 2007).

The hippocampus, where new memories begin their life, starts to behave like a saturated sponge. A night of sleep deprivation before learning reduces the hippocampus’s ability to encode new information by something close to 40% (Yoo et al., 2007).

Tired people also make riskier choices, and, probably more worryingly, don’t notice they’re doing it. Self-rated performance stays roughly stable as objective performance drifts (Van Dongen et al., 2003). The tired version of us is convinced it’s fine; but it isn’t.

Why this should bother anyone in L&D

Sleep does three things to memory.

Sleep before learning prepares the brain to take new information in. Sleep after learning is when much of that information is consolidated and integrated with what you already know. Sleep is also where creative connections between unrelated ideas tend to get made; people who’ve slept after exposure to a problem are more than twice as likely to discover a non-obvious solution to it (Wagner et al., 2004). Pre-learning sleep loss impairs memory more than post-learning sleep loss does, but both matter (Newbury et al., 2021).

A training day delivered to participants who arrived tired loses encoding capacity. A training day followed by another short night loses consolidation. A design sprint or strategy session held across two intensive days with limited sleep in between loses the integrative work that the sleep itself was meant to do. None of these losses appear on the post-training feedback form. They show up later, as the application that didn’t happen, the skill that didn’t stick, the behaviour change that didn’t materialise, and the line manager wondering whether the training was worth the spend.

We worry endlessly about modality, engagement scores, or whether the activity was active enough or the multimedia rich enough, and most of these things matter rather less than whether the people in the room had slept. We can spend significant money optimising a training experience for an audience whose brains have been pre-degraded by the conditions of the organisation paying for the training.

Join me and the one and only Cathy Hoy as we explore why leadership development has become L&D’s No. 1 focus in 2026, according to the first SME L&D survey. We’ll also look at what great leadership development training looks like today.

Register for the free webinar today!

The workplace is doing a lot of the producing

When we take a really honest look at the situation, we see that the workplace is responsible for a great deal of the tiredness it then tries to mitigate. Long hours, mobile devices that don’t sleep, evening emails, business travel scheduled to maximise meetings and minimise recovery, and shift patterns that work against the body’s clock rather than with it.

The average UK adult gets three nights of decent sleep a week (Mental Health Foundation, 2025). The CIPD’s most recent figures put average sickness absence at the highest level in over a decade, with much of the rise attributable to stress, anxiety, and mental ill-health (Chartered Institute of Personnel and Development, 2023). RAND Europe put the annual UK economic cost of insufficient sleep at around £40B, or roughly 2% of GDP (Hafner et al., 2017).

What L&D, HR, and line managers can do about it

Most of the popular individual-level wellbeing interventions, the resilience training, the mindfulness apps, the cheery webinar at lunch, show little effect when examined rigorously (Fleming, 2024). The interventions that move sleep tend to operate higher up the system, on workload, supervision, and scheduling. The list below is roughly in the order of observed impact.

Stop scheduling training as if tiredness isn’t real

Build training around when brains can do the work. Avoid the post-lunch slump for deep cognitive work; that’s the lowest point of the circadian dip for most adults. Don’t schedule intensive new sessions for the second day of an off-site; everyone’s already running on insufficient sleep by then. Build twenty-minute breaks rather than five, and put them between conceptually demanding sessions rather than at convenient hourly intervals. Don’t introduce substantive new content after 16:00, and don’t ask people to apply complex new skills the morning after a late-finishing evening session. If your training calendar regularly violates these patterns, ask how much of your training budget is being spent on memories that won’t form.

Treat travel and shift handovers as training disqualifiers

If somebody has flown in the night before, has come off a night shift, has been on the road three days running, they shouldn’t be in your training room the next morning. Make it acceptable, and expected, to reschedule. Build that flexibility into your booking system; a same-week reschedule with no penalty is the operational design that works. The HSE’s HSG256 guidance on shift work is the foundation document for getting shift design less wrong, and most organisations running shifts have never read it (Health and Safety Executive, 2006). Read it. Apply it. Audit your current rotas against it. Where you find permanent night shifts, backward rotations, sub-07:00 starts that aren’t operationally required, or runs of seven or more consecutive shifts, fix them in that order.

Audit your meeting and email norms before you audit your training

A norm of 22:00 emails is a norm of next-day cognitive impairment, and no amount of clever instructional design will compensate for an organisation operating at the equivalent of mild intoxication. Look at the hours during which your senior leaders send messages, whether replies are expected by morning, and whether anyone has ever been promoted for being seen to work late. The single most useful thing many L&D and HR teams could do this year is sit down with their leadership team and have an awkward conversation about what their email behaviour is teaching everyone else. If that conversation can’t happen, the next-best move is a written norm: no internal email outside contracted hours with delay-send used by default; meeting-free blocks during the working day; and a clear statement that out-of-hours messages don’t require a same-day reply.

Train line managers to protect sleep, not to recommend yoga

Supervisor support for sleep, the willingness of a manager to say “push the deliverable, you’ve had three brutal weeks” rather than “have you tried the meditation app”, produces measurable improvements in employee sleep and wellbeing (Hammer et al., 2021). This is a skill, it can be taught, and it costs less than the wellbeing platform subscriptions most organisations are paying for. Equip managers with three things: the vocabulary to ask about sleep without it sounding intrusive, the authority to redistribute work when somebody is running on empty, and a clear set of organisational permissions that back them up when they do.

Take sleep disorders seriously

A sizable number of the people who turn up to work tired aren’t tired because they’re staying up watching Netflix. They have obstructive sleep apnoea or another underdiagnosed sleep disorder, and most don’t know it. For safety-critical roles in particular, this is a conversation to have with your occupational health provider; for everyone else, the simplest first step is including a validated screening tool such as the STOP-BANG questionnaire in your routine health checks, with clear referral pathways to a sleep clinic for those who score high. Treatment of sleep apnoea pays for itself comfortably in recovered productivity, and that’s before counting the safety benefit.

Stop framing tiredness as a personal failing

Change the language. Tiredness in your workforce is not, on the whole, a character flaw, a discipline gap, or an opportunity for a sleep-hygiene leaflet. It’s a signal about workload, scheduling, and leadership behaviour. When somebody mentions they’re exhausted, the first response from a manager or HR business partner shouldn’t be “have you tried” anything; it should be “tell me what’s making you tired”. That conversation produces useful information.

A short note before I shut the laptop

The strange thing about writing this piece is that I knew most of it before I went to Karlsruhe. I’ve read this material and quote some of it in workshops. I still managed to lose 4 days of writing to a conference schedule.

This is the trap most of our employees are in. They know they’re tired, and they know it’s affecting them. They’re not, on the whole, choosing it freely; they’re navigating workplace expectations that make adequate sleep difficult, and then turning up to the training we put on to help them perform better. Whether that training works or not is decided long before they walk in.

Tiredness is a performance problem, a learning problem, a safety problem, and a leadership problem, and it sits much closer to the centre of our work than we’ve tended to admit. It deserves a place on the agenda alongside skills, capability, and culture.

Leave a comment

References

Chartered Institute of Personnel and Development (2023) Health and wellbeing at work 2023. London: CIPD.

Dawson, D. and Reid, K. (1997) ‘Fatigue, alcohol and performance impairment’, Nature, 388(6639), p. 235.

Fleming, W.J. (2024) ‘Employee well-being outcomes from individual-level mental health interventions: cross-sectional evidence from the United Kingdom’, Industrial Relations Journal, 55(2), pp. 162-182.

Hafner, M., Stepanek, M., Taylor, J., Troxel, W.M. and Van Stolk, C. (2017) ‘Why sleep matters: the economic costs of insufficient sleep: a cross-country comparative analysis’, RAND Health Quarterly, 6(4), p. 11.

Hammer, L.B., Brady, J.M., Brossoit, R.M., Mohr, C.D., Bodner, T.E., Crain, T.L. and Brockwood, K.J. (2021) ‘Effects of a Total Worker Health leadership intervention on employee well-being and functional impairment’, Journal of Occupational Health Psychology, 26(6), pp. 582-598.

Health and Safety Executive (2006) Managing shift work: health and safety guidance (HSG256). London: HSE Books.

Mental Health Foundation (2025) Restless Britain: average UK adult has only three days a week with good quality sleep.

Newbury, C.R., Crowley, R., Rastle, K. and Tamminen, J. (2021) ‘Sleep deprivation and memory: meta-analytic reviews of studies on sleep deprivation before and after learning’, Psychological Bulletin, 147(11), pp. 1215-1240.

Van Dongen, H.P.A., Maislin, G., Mullington, J.M. and Dinges, D.F. (2003) ‘The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation’, Sleep, 26(2), pp. 117-126.

Wagner, U., Gais, S., Haider, H., Verleger, R. and Born, J. (2004) ‘Sleep inspires insight’, Nature, 427(6972), pp. 352-355.

Williamson, A.M. and Feyer, A.-M. (2000) ‘Moderate sleep deprivation produces impairments in cognitive and motor performance equivalent to legally prescribed levels of alcohol intoxication’, Occupational and Environmental Medicine, 57(10), pp. 649-655.

Yoo, S.-S., Hu, P.T., Gujar, N., Jolesz, F.A. and Walker, M.P. (2007a) ‘A deficit in the ability to form new human memories without sleep’, Nature Neuroscience, 10(3), pp. 385-392.

Yoo, S.-S., Gujar, N., Hu, P., Jolesz, F.A. and Walker, M.P. (2007b) ‘The human emotional brain without sleep: a prefrontal amygdala disconnect’, Current Biology, 17(20), pp. R877-R878.

If we are going to make skills the organising language of talent, learning, and performance, the word has to mean something specific. When it covers everything, it carries no useful information. We end up with skills frameworks that contain personality traits, character qualities, physical attributes, and learned task abilities all piled together as though they were the same kind of thing, and then we wonder why the resulting strategy never delivers.

The same loose usage shows up in training programmes that promise to “build the skill of resilience,” in job adverts asking for “the skill of being a self-starter,” and in performance conversations that confuse capacity with competence. Enthusiasm for skills work is plentiful; shared vocabulary is in short supply.

Here is one workable set of definitions, offered as a starting point rather than as doctrine.

Knowledge is what a person understands or holds in their head: facts, concepts, principles, procedures, and context. It can be assessed through recall, recognition, explanation, or application to a familiar problem.

Skill is a learned ability to perform a specific task to a defined standard, developed through practice and observable in performance. It is task-bounded and improvable through deliberate effort. Facilitating a difficult conversation is a skill; writing a clear technical brief is a skill; running a structured root cause analysis is a skill.

Capability is the broader capacity that enables performance, combining knowledge and skill with personal attributes, physical and cognitive capacities, motivation, and the conditions in which a person works. Capability is what an organisation builds when it supports the development and deployment of skills in real contexts.

Within this framing, strength and endurance belong under physical capacities. Curiosity and conscientiousness belong under personal attributes. Resilience, however we choose to describe it, belongs closer to a capacity shaped by context and experience than to a discrete learned ability that can be trained in a half-day workshop.

What you call these things matters less than the fact that you call them the same things every time. The specific definitions above are not the point; you may have better ones, and most organisations will adapt them to fit how work is talked about locally. The point is that your organisation has working definitions, agrees on them across L&D, HR, talent, and operations, and uses them consistently in conference talks, training programmes, performance reviews, and day-to-day conversations about who can do what. Without that shared language, every discussion about skills is a discussion about different things being called the same name.

Leave a comment

On panel after panel, speakers talked about agents doing “entire jobs” as if that were an unremarkable thing to say in a room full of people whose work involves other people. Teams of the future, we were told, will be made up of humans and AI agents working alongside one another, and this is described as inevitable. Inevitability is the wrong word for any of this; what is being described is a choice, or rather a long series of choices, made by leaders, by procurement teams, by boards, and by the vendors actively shaping the language we use to describe our own future. Repetition does not constitute destiny.

What I did not hear, on any panel, in any session, in any of the conversations I sat in on, was a serious word about the work that falls to us once those choices have been made. If we are going to commit to teams of humans and agents, we are also committing to managing the displacement that decision creates; to re-skilling, in some cases, entire generations of people whose roles have evaporated under them; to rebuilding the entry points to careers that we are now so casually proposing to automate away; to working out, somehow, how an organisation produces senior employees in five or ten years’ time when it has decided it no longer needs juniors today.

That last point is the one that troubles me most. The far more probable outcome of all this looks nothing like the elegant hybrid team the keynote slides suggest; it looks like a workforce hollowed out at the bottom by businesses convinced they have found a shortcut, then surprised to find no experienced talent pool waiting for them when they look up. We will be the ones asked to fix that, of course, on shorter timelines and with less budget than the AI investments that caused it.

If we are serious about maturing this conversation, we have to be willing to have the harder one. We have to be honest about the human cost of the path we are being sold, about the work that will land on people professionals as a direct consequence, and about the fact that “inevitable” is a word vendors use when they want us to stop asking whether something is a good idea. The mature version of this discussion is an unsentimental account of what we are about to ask of the people we are paid to support, and the work it will create for those of us responsible for them; we have not had it yet, and the longer we delay it, the more difficult that work becomes.

Note: This is one of those topics where everyone has a perspective, and no one really knows the full picture. It’s funny how this is only really acknowledged by the most educated and least knowledgeable people. There’s a lot of claimed expertise in the middle ground, but it’s propped up by very little (if any) real-world evidence.

I have no special insight into how close we are to useful workplace agents, and I do not know what the timeline will look like for regularly seeing agents in the workplace. I am not trying to assess the accuracy of vendor claims on those points.

That said, this does not stop me from thinking about the potential human and operational impacts of this kind of change. It is not without precedent. There have been massive automation-driven displacements before, and most of them have been poorly managed, causing significant societal and economic damage.

My argument is not for or against AI in the workplace or AI agents. It’s for careful planning as we work through this disruption and for deep consideration of the human effects when someone says they are building a team of agents, or that in the next three to four years they expect 80% of their workforce to be AI agents. We should acknowledge that means they expect to lay off 80% of their employees. The work of ensuring that those people can either be relocated within an organisation or have meaningful opportunities outside of it lies with us, including our ability to upskill, reskill, and support people as they may need to pivot their careers.

Passion

Passion in a talk is the audible and visible evidence that the speaker cares about what they’re saying. It shows up in vocal warmth, in raised energy through specific phrases, in eye contact that doesn’t drift to the back wall, in word choice that sounds chosen rather than recited; an audience knows very quickly whether the person on stage is invested in the material or merely getting through it.

Audiences match their attention to the speaker’s commitment. Sound invested, and they are more likely listen carefully; sound bored, and they’ll be bored too. No amount of slide design will recover the situation. Passion is the unspoken contract between speaker and room: I care about this, so it’s worth your care.

The control problem comes when passion runs unchecked across the whole talk. A speaker who delivers every sentence at the same elevated emotional intensity fatigues the room; the peaks flatten into a plateau, and the moments that should land hardest land softly because nothing has been kept in reserve. Sustained intensity also crowds out nuance, which is where most useful ideas live, and it tends to push audiences into a defensive posture where they stop processing and start waiting for the energy to drop.

Practically speaking, identify the two or three points in your talk that carry the most weight, and to commit your highest emotional intensity to those alone. Everywhere else, dial it back to engaged and warm. Speak the connecting material with interest; reserve fervour for the ideas you most need the audience to leave with. Audiences read contrast more readily than absolute volume, so the variation between warm engagement and full conviction is what makes the high-intensity moments register at all. A talk delivered at one emotional pitch from beginning to end has no peaks, regardless of how passionate that pitch happens to be.

If you want to up your learning science game, join us on the 29th of May 2026 in Birmingham for IDTX Evidence-Informed practice Conference where research meets practice.

As a reader of the Instructional Design Tips Substack, you can get 25% off your ticket using code CPDW25 at checkout.

Book your ticket today!

Pace

Pace is how quickly your words leave your mouth, and it works as a control on what the audience does with what they’re hearing. Faster pace creates urgency and forward momentum; slower pace creates weight and the space for the audience to think.

Information delivered fast tends to wash over a listener as a wave of impression they can’t easily examine; information delivered slowly invites scrutiny, which is what you want when the point is one the audience needs to take seriously. Used well, pace tells the audience how to weight what they’re hearing without you ever having to announce “this bit matters.” Used badly, it produces either a breathless rush that nobody retains or a uniform plod that nobody cares to listen to.

A talk that maintains one moderate pace throughout will read as professional and unmemorable; a talk that speeds up through stories and energy and slows noticeably for substance will feel dynamic, and the audience will remember the slow bits. This is largely why the speakers who feel commanding tend to use less speech than the ones who feel rushed; slower delivery creates more weight per word, and it’s this percieved weight that people tend to retain after they leave the room.

I like to map my talks before delivering them. Deciding where I want the audience leaning forward versus where I want them taking notes. Stories and energetic transitions can run quickly; the audience knows these are the colour around the substance and processes them accordingly. Definitions, central claims, and the ideas you most want the audience to remember should be delivered noticeably more slowly than the surrounding material, with deliberate articulation. The contrast does the work of emphasis without you having to over-rely on volume or repetition.

Pauses

Pauses are silences placed deliberately, and they are the most underused tool in speaking. A pause after a substantive claim gives the audience time to register and consider what’s been said; a pause before a key sentence builds anticipation and pulls attention forward; a pause in place of a filler word like “um” or “so” removes verbal clutter and signals composure. Most people fear silence and fill it; the speakers who command rooms have learned to be comfortable in it.

Note: One of my favourite things to do is step on to stage and just wait. the room will rimble down to a ready silence, hanging on your first words. Silence is radical, when the world is making so much noise.

Comprehension and retention require processing time, which continuous speech denies the audience. Make a point and immediately move to the next sentence, and the listener hasn’t had time to absorb the first idea before the second arrives; the points land on top of one another and neither registers properly. Silence is where the listener does the work; without it, you’re delivering a monologue to people, not communicating with them.

The fear most speakers carry is that silence will read as faltering or as having lost their place. This is a calibration problem about how the silence is held, not a problem with silence itself. A pause accompanied by relaxed posture and steady eye contact reads as deliberate; the same silence accompanied by darting eyes and a glance at notes reads as panic. The pause is the same; the body language around it tells the audience how to interpret what’s happening.

Mark two or three points in your script or notes where you want a deliberate pause of around three seconds, then hold the silence with a steady gaze and resist the urge to fill it. After a key claim, count an internal beat before continuing. The first time you do this, three seconds will feel like an eternity to you and read as entirely normal to the audience; speakers consistently underestimate how short their pauses sound to listeners, which is why most talks have far too few of them. A pause that feels uncomfortably long from the stage is, more often than not, the precise length the audience needed to follow you.

Passion gives the audience reason to care; pace tells them what to weight most heavily; pauses give them the space to absorb it. None of these techniques require natural charisma or any extraordinary measure of stage presence. They require preparation and the willingness to shape how you sound, rather than letting default delivery carry a talk that deserves better.

Leave a comment

Stories on stage do two jobs. They give the audience something concrete to picture while the abstract idea is still settling, and they create an emotional shift that makes whatever comes next more memorable than it would have been on its own. The work of being a good speaker on this front splits cleanly into collection and delivery.

I think of my own story collection as a compost heap, because that’s how it behaves. Stories go in half-formed and apparently unimpressive, sit there for months or years, and at some point they become exactly the example I need for a session I had not yet been asked to deliver. The act of collecting matters more than the act of choosing in the moment, because we cannot predict in advance which story will turn out to be the one that lands.

Some practical ways to build the heap:

Collect everything that surprises you. A passing remark in a meeting you have not been able to stop thinking about, or a moment in a workshop where the room shifted; if it caught your attention, it is a candidate. Capture it the same day, with enough detail to reconstruct it months later when you need it.

Stop hunting for dramatic stories. The project rescue and the high-stakes failure tend to overshadow the point you are trying to make, because the audience remembers the helicopter and not the principle behind it. A small story, fitted closely to what you are explaining, does more work and is far easier to find in the first place.

Capture the texture, not the summary. The exact phrase someone used, or the specific number that surprised you, will make a story feel real on stage. A summary of what happened is a report, and reports are forgettable.

Revisit the heap before every talk. Most speakers prepare a session and then go searching for stories to slot in; the work flows better the other way around. Read back through the heap before you start writing, and let the stories suggest the structure of the talk to you.

Telling the story well comes down to a structure that holds across nearly every story worth using: setup, challenge, resolve.

Setup gives the audience just enough context to care, and almost nothing more. Trim it back further than feels comfortable, and trust the audience to fill in the rest; long setups lose the room before the story has properly begun.

Challenge is the tension at the heart of the story, the thing that did not go to plan or the question nobody could answer. If you cannot identify the challenge in something you are about to tell, the story is not yet ready to use, and rehearsing the telling will not save it.

Resolve closes the story, though it does not need to be tidy or triumphant; sometimes the resolve is that nothing was solved, and that becomes the lesson. What matters is that the place the story lands is the same place your point begins.

Stories that fall flat are almost always missing one of these components, or have all three but in the wrong proportions. A long setup with no challenge becomes background noise; a challenge with no resolve leaves the audience confused and slightly irritated.

Build the heap, capture small things, and tell them with setup, challenge, and resolve.

Leave a comment

Note: to be fair, we have 101 things going on, and this is not exactly part of the core job role, so I understand why it’s happening, and it’s not just in our industry.

I’ve noticed the same thing across most of the technology space. There have been some standout public speakers in recent years, but overall people have become much worse at this. I think, globally, there isn’t enough time and money invested in developing this skill, because it’s seen as something only very advanced people need, or that it’s not important anymore. Domain knowledge, expertise, and respect alone are enough to get people to pay attention to you.

The reality is that people will stay and listen, and people will clap, but people are more likely to take on ideas when they’re well presented. I think the main reason we see so many bad ideas scaled across organisations, so many bad ideas invested in, and so many bad people gaining investment is that everyone else lacks communication and public speaking skills.

The opening is where it falls apart most often. People walk onto a stage, give their name, tell you the company they work for, tell you how long they’ve been there, tell you what their department is responsible for, and then read out bullet points listing what today’s session will cover. You are three or four minutes in before the subject is touched. It is the human version of those late-1990s e-learning courses that opened with an author credit screen.

If people have shown up to your session, they have already bought into your credentials. Don’t bore them. Get on with it.

What you should be doing instead is creating a moment. The first sixty seconds of any talk are the highest-attention part of the entire session, and they should be treated that way. Dr. James Whittaker has spoken on this idea, that a speaker is closer to a performer than a presenter, and that the opening is where you earn the right to keep the audience’s focus for the next forty minutes.

Some ways to do this:

Open with a story. Drop the audience into the middle of a real situation, in the first person, before they know where you are going. Skip the setup.

Present a hard-hitting statistic. Not a vanity number; a figure that reframes how the audience thinks about the problem in the first ten seconds.

Be deliberately provocative. State a position you know a portion of the room will disagree with, and let the tension sit there for a second before you keep going.

Make them laugh. A moment of humour, ideally observational and connected to the topic, signals that the next forty minutes will not be a chore.

Use silence. Walk to the centre of the stage, look around the room, hold a beat, and then begin. Confident silence is one of the most underused opening tools available.

Start with a single object or image. Hold something up, put a single image on the screen, and talk to it. No title slide, no agenda, no logo.

State the thesis. Deliver the central argument of your talk in one sentence, before anyone has had a chance to settle in. Spend the rest of the session defending it.

Any of these can work. What unites them is the same underlying move: you are stepping onto the stage and taking the room, creating the moment before anyone has had a chance to disengage. There are people in this profession with extraordinary expertise we do not hear from enough, because they have not yet developed the speaking skills to stand out. Starting strong is the first habit worth building.

Leave a comment

One principle runs through all of this; digital components belong in a training experience when they do something the room, the facilitator, and the participants can’t. If a digital component can’t pass that check, it’s taking up space something else should be doing.

Get 5% off forever

A few thing to consider…

When Harvard students received identical physics content either by active problem-based instruction or by passive lecture, the active-learning group scored significantly higher on objective tests but also rated their own experience lower than the passive group (Deslauriers et al., 2019). Feeling engaged and being engaged are different states, and the first often runs inversely to the second. Digital tools are effective at producing the feeling of engagement; polls, word clouds, emoji reactions, and dot-voting all produce a lot of visible behaviour, and that behaviour reads well in the room and in post-session feedback. Whether it corresponds to the thinking we need participants to do is a separate question.

Chi and Wylie (2014) give us a more precise vocabulary through the ICAP framework: participants can be passive, active, constructive, or interactive in the thinking they’re doing. Clicking a multiple-choice option is active. Writing out an explanation for why you clicked it is constructive. Arguing with someone else about whose explanation is stronger is interactive. Most in-room digital tools produce the first of those three without moving people up to the next. The test for any digital component is whether it helps shift participants up that ladder.

Pre-work

The place where digital components consistently earn their place is before participants arrive. When pre-work is designed well, the in-person session doesn’t need to cover everything from scratch; it can jump to the application, the discussion, and the practice the room does better than any digital channel.

This is where the duplication problem most commonly creeps in. A team builds an effective e-Learning module that covers the foundational content, sends it out three days before the workshop, and then, because they’re not confident that everyone has done it, covers the same material again for the first forty minutes of the in-person session. The people who did the pre-work sit through a recap of what they already know; the people who didn’t do it are effectively rewarded for not doing it.

The fix is to commit. If we’re going to send digital pre-work, we design the in-person session on the assumption that people have done it, and we use the opening of the workshop to check for comprehension before moving quickly into the application work that the room does best. If we don’t trust that people will have done the pre-work, we make it shorter, more useful, or more tightly linked to something they need to bring with them. Hedging the design against the possibility of non-completion, by duplicating the content live, undermines the pre-work and bores the participants who did it.

Secret Cinema’s design approach is a great example. Attendees receive pre-event dossiers, in-character emails, and app-based briefings that prime them for the physical experience without giving away what will happen. By the time they walk through the door, their attention is already oriented towards the event, and the live experience can do what it does best without having to establish context from scratch. Most workshop pre-work is built to transfer information; the stronger use is to prime attention, establish shared vocabulary, and surface the questions participants are bringing into the room.

Note: I know many people will struggle with the idea of not re-covering everything just in case, but if you keep doing this, then there really is no point in issuing pre-work.

You need to communicate clearly and directly that pre-work is not optional. You need to agree with whatever planning authority or leadership authority exists in your organisation that time will be dedicated to doing this, and that it is an essential part of the training they are agreeing to invest in by running the workshop.

Explain to every individual participant, as directly as possible by email or a conversation, that failure to do the pre-work will prohibit them from completing the course. We are far too afraid, I think, in the world of L&D of the idea of people failing a course. If attendees cannot even complete basic pre-work, why on Earth would we consider them as having passed or completed a course?

I won’t go into this in further detail in this article, as it’s another whole set of considerations, but this is an area that if we continue as we are, we’re likely to hamstring ourselves.

If you’d like to explore how you can use research from psychology, neuroscience, the cognitive sciences and behavioural sciences in your work in L&D, HR, performance enablement or leadership, consider attending this year’s Evidence Informed Practice Conference.

As a reader of this Substack, you can get 25% off your ticket using code CPDW25 at checkout.

Learn more and book your tickets today!

In the room

Now let’s take a look at the considerations for using technology in the room during a live session. Sana, Weston, and Cepeda (2013) found that participants within visual range of a peer using a device for non-task purposes scored seventeen per cent lower on comprehension tests, even when they weren’t using a device themselves. Distraction is contagious, which means BYOD workshops can introduce risk of distraction not only to those using a device but to those around them as well.

When Carter, Greenberg, and Walker (2017) permitted laptops and tablets in classrooms, final exam scores dropped by around 0.18 standard deviations, even when tablets were restricted to flat-on-desk positions, generally considered a way to ensure they were only being used for what they were meant to be used for. If we’re running workshops where retention and application a month later are what we care about, handing everyone a second screen comes with a measurable cost.

Three use cases are supported well by the evidence, and they’re the ones I consistently keep in my designs.

The first is anonymous contribution. Hunsu, Adesope, and Bayly (2016) synthesised fifty-three studies on audience response systems and found small but reliable positive effects on both engagement and learning, the reason being anonymity. In a room where some participants are more senior than others, or where the topic is sensitive, a digital poll lets people contribute an opinion without having to say it out loud.

The second is parallel contribution at scale. In a workshop of thirty people, a shared canvas lets everyone put their ideas up at once, without having to wait their turn, which is something the physical room cannot manage without considerable facilitation effort. Rogers and Lindley (2004) explored “one-driver” dynamics where one person controls and others spectate. They found that individual device entry points distribute contribution more evenly. When we use a Mural or a Miro in the room, we can give everyone their own entry point into the shared canvas, so that no-one in the room becomes a passive observer of someone else’s screen.

The third is timing, calculation, or hidden state. Timers, automated grouping, rotating facilitator prompts that need to appear on cue, content that’s randomised or personalised to role or team; these are jobs digital tools do better than a flip chart and a stopwatch, and they free the facilitator up.

The big question here about attention. Weiser and Brown (1996) described calm technology as technology that moves smoothly between the periphery and centre of attention, and Bakker and Niemantsverdriet (2016) extended this into a continuum of focused, peripheral, and implicit attention. The useful principle for workshop design is that every digital component should sit at the lowest level of attention that still lets it do its job. A silent, glanceable timer is better than a beeping countdown; an ambient progress indicator is better than a modal alert; a chat that’s checked at natural pauses is better than notifications that interrupt the facilitator. The more of a participant’s attention the digital layer demands, the more it competes with the in-room conversation, and the less reason we have to bring it into the room at all.

Post-work

If I were allowed to keep only one digital component in my workshop design, it would be what happens after the session ends.

Let’s start with some spaced practice. Three or four well-designed retrieval questions, delivered by email or a basic app at increasing intervals, do more for long-term retention than almost anything we can add to the live session, and they cost a fraction of the development time most of us spend building in-room digital activities.

As always, reminding people of information is not enough. To strengthen the memory trace, we need to help them recall and apply that information in different, realistic contexts.

Reflection is up next. A well-timed digital reflection prompt does two things:

1. it nudges participants to try applying what they covered in the session,

2. it gives us a signal on transfer that we’d otherwise have no visibility on.

Asking participants what they’ve tried, what worked, and what got in the way gives us data we can use to design the next intervention, and it keeps the workshop alive in their working lives beyond the day it ran.

The harder case is asynchronous peer conversation. The research is mixed, and most attempts fail not because the idea is wrong but because the digital component gets designed as if it were a forum when what it needs to be is a moderated, purposeful set of prompts with a clear reason for participants to come back. If we can’t explain, in one sentence, why someone would log in on a Tuesday to contribute, the community won’t sustain, and we’ll be looking at an empty channel within a fortnight.

A simple diagnostic

When I’m working with a team on the digital elements of an in-person programme, I ask the same six questions in the same order. They won’t give us a perfect answer, but they’ll filter out most of the digital components we reach for out of habit.

What job is this doing that the room cannot do itself?

If the room can do it, the room should. Anonymity, parallel contribution at scale, timing, hidden state, personalisation, asynchronous reach; these are the jobs that justify a digital component. Everything else is a job the facilitator and the participants can do without needing a screen in the way.

What level of attention does it demand?

If a participant has to look away from the conversation to engage with it, the digital component is competing with the in-room work. If it can sit at the periphery of attention, checked briefly or glanced at, it supports the room.

What kind of thinking does it produce?

Mapped against the ICAP framework, does the digital activity produce passive absorption, visible behaviour, generative thought, or dialogue? We’re aiming for the second two wherever we can.

Could this live before or after the session, and work better there?

Most components we reach for in the room would do more for retention and transfer if they ran after the room. Spaced retrieval, reflection, and follow-up prompts are where the evidence is strongest, and where most of us underinvest.

What happens when it fails?

If the workshop collapses when the wi-fi drops, the digital layer is holding you back rather than supporting you. Every digital component should be designed so the session still works without it.

Note: This assumes the digital item isn’t what you’re actually training people on. If you’re doing software training, you’ll need the software itself to train people.

Who is excluded?

Every digital component expands the accessibility and equity surface of the workshop. Participants with older devices, screen readers, low data allowances, or limited comfort with the tool have to be designed for, not discovered at the moment they can’t join in. The Web Content Accessibility Guidelines (W3C, 2023) give us the free, public framework for auditing most of the tools we’re using.

Brining it together

The change I’d commend to anyone asking me this question is to stop asking how to bring more digital into the workshop and start asking what problems you’re struggling to address in your workshops. Only then can you ask, “Is there a technology that can support me in this?”

Digital in the room should be disciplined, purposeful, and peripheral to the conversation, turning up only when it does something the room itself cannot. Digital before the room should be committed to, with the in-person session designed on the assumption that pre-work has been done and the opening used to check for comprehension before moving into application. Digital after the room should be where most of our investment goes, because it’s where the retention and transfer research is strongest, and because it’s the phase where digital has a hard-to-replicate advantage.

Leave a comment

References

Bakker, S. and Niemantsverdriet, K. (2016) ‘The interaction-attention continuum: Considering various levels of human attention in interaction design’, International Journal of Design, 10(2), pp. 1-14.

Carter, S.P., Greenberg, K. and Walker, M.S. (2017) ‘The impact of computer usage on academic performance: Evidence from a randomized trial at the United States Military Academy’, Economics of Education Review, 56, pp. 118-132.

Cepeda, N.J., Pashler, H., Vul, E., Wixted, J.T. and Rohrer, D. (2006) ‘Distributed practice in verbal recall tasks: A review and quantitative synthesis’, Psychological Bulletin, 132(3), pp. 354-380.

Chi, M.T.H. and Wylie, R. (2014) ‘The ICAP framework: Linking cognitive engagement to active learning outcomes’, Educational Psychologist, 49(4), pp. 219-243.

Deslauriers, L., McCarty, L.S., Miller, K., Callaghan, K. and Kestin, G. (2019) ‘Measuring actual learning versus feeling of learning in response to being actively engaged in the classroom’, Proceedings of the National Academy of Sciences, 116(39), pp. 19251-19257.

Hunsu, N.J., Adesope, O. and Bayly, D.J. (2016) ‘A meta-analysis of the effects of audience response systems (clicker-based technologies) on cognition and affect’, Computers & Education, 94, pp. 102-119. (Availble on request)

Mawson, C. and Kang, S.H.K. (2025) ‘The distributed practice effect on classroom learning: A meta-analytic review of applied research’.

Rogers, Y. and Lindley, S. (2004) ‘Collaborating around vertical and horizontal large interactive displays: Which way is best?’, Interacting with Computers, 16(6), pp. 1133-1152.

Sana, F., Weston, T. and Cepeda, N.J. (2013) ‘Laptop multitasking hinders classroom learning for both users and nearby peers’, Computers & Education, 62, pp. 24-31.

W3C (2023) Web Content Accessibility Guidelines (WCAG) 2.2.

Weiser, M. and Brown, J.S. (1996) ‘The coming age of calm technology’.

As always, I want to start by thanking you for being a paid subscriber to the Instructional Design Tips Substack. Your support lets me keep writing here five days a week, and I’m also publishing this omnibus edition as a thank you to you.

If there are topics you’d like to see covered in a future Members Only article or in any of the daily articles, please let me know in the comments or in the chat. I’ll always prioritise talking about the things you’re interested in over the ideas floating around in my head when I’m writing.

In this edition:

• When Fear Sets Your Goals For You

• Quick Definition: MCP Server

• Reader Question: The SME Who Can’t Simplify

• Reader Question: What Kit Do You Use to Run Virtual Workshops?

Read more

Part one: conferences and events

With conference season running roughly May through November, both hosts share how they now navigate the schedule. The core observation is that cramming in every session rarely pays off; the conversations outside the sessions, the fringe events, and the smaller niche gatherings often deliver more than the headline acts. Tom and Heidi cover the differences between US, UK, and mainland European event cultures; the commercial pressures that shape speaker line-ups; and why the phrase “conference junk food” captures something important about how our attendance patterns end up shaping future programming.

The second half turns to the question that always comes up: how do I get onto a conference stage? Tom lays out the three things he looks for when selecting speakers, including why public speaking is a distinct skill from training or facilitation, and why a video or podcast track record matters more than a strong CV when someone is making a speaker selection call.

Part two: MCP and the language of L&D tech

Following Docebo’s recent announcement, Heidi noticed the acronym MCP, Model Context Protocol, appearing in multiple vendor pitches within a single week. Tom explains what MCP is, where it came from (Anthropic, 2024), and why its status as an open standard mattered. The broader thread is about how L&D vendors dress up table-stakes technology in impressive-sounding language, and the hosts close on why human judgement remains essential for testing whether your ideas are any good; AI tools, helpful as they are for many things, will not look you in the eye and tell you you are wrong.

Chapter markers

• (01:30) Welcome; first time live across three platforms

• (02:48) This month’s topics: events and tech jargon

• (04:35) Why three days of back-to-back direct instruction misses the point

• (06:17) Being selective with the schedule and using the gaps well

• (08:44) The conversations outside the sessions

• (09:22) Fringe events, and why Tom is skipping Learning Technologies this year

• (10:28) Lauren’s point on the cost of attending

• (11:38) Small, niche events: Andrew Jacobs’s podcasting-for-learning day

• (13:58) LearnTech Karlsruhe and the European take on environment and learning

• (19:49) Behind the curtain: speaker selection, gender mix, and committee politics

• (24:49) Conference junk food: the gap between what people say and where they sit down

• (28:27) Speaking at events, part one: you are unlikely to be paid

• (32:19) Topic is almost irrelevant; teach what you know, and have a story

• (35:14) Public speaking as a distinct skill; The Art of Stage Presence

• (40:32) How selectors look for evidence of presentation ability

• (44:52) Vanilla vs opinionated speakers, and choosing the right stage

• (48:52) When the answer is to build your own stage

• (49:24) Perspective: nobody outside the L&D bubble knows who any of us are

• (52:28) Tech chat: what is MCP, and why is everyone suddenly saying it?

• (56:23) MCP as marketing language and the AI-in-disguise problem

• (58:43) Why human challenge beats AI validation every time

• (1:01:09) Sign-off; no May episode

Mentioned on the show

• Lauren Waldman’s comment on the cost of attending conferences

• Andrew Jacobs’s podcasting-for-learning event (London, early 2026)

• LN Connect, London

• LearnTech, Karlsruhe, Germany (IKEA and Herman Miller on environment and outcomes)

• Learning Technologies, World of Learning, DevLearn, and ATD (referenced)

• The Business of Training Conference (UK)

• The Art of Stage Presence (YouTube) — Tom’s go-to recommendation for anyone wanting to present

• Docebo’s recent LMS, HR, and AI integration announcement

• Model Context Protocol (MCP), released by Anthropic

Coming up

No May episode; the ID Update returns in June. Tom will be at IDTX on 29th May 2026 at Eastside Rooms, Birmingham, the inaugural in-person Evidence Informed Practice conference. Heidi will be on LinkedIn and Substack throughout May.

Get more from Tom McDowall in the Substack app

Available for iOS and Android

Get the app

A lot of what we do in L&D has the same shape, if we stand back far enough to see it. We design a programme, launch it, and treat that launch as the end of the project; the work gets filed as done, and our attention turns to whatever’s next. What we’ve done is plant; what’s left is every other act of gardening, and it’s those other acts that are where the value comes from, or doesn’t.

Donella Meadows, one of the most readable thinkers on complex systems, wrote that she’d learned how to work with systems partly from gardening, along with whitewater kayaking and playing music; the common thread was that you can’t control the system you’re participating in, but you can stay awake and respond to what it gives back (Meadows, 2001). She came to think of self-organising, non-linear, feedback-driven systems as inherently unpredictable, and the sensible response, she said, is to change how we stand in relation to them: from trying to master the system to dancing with it. The gardener’s stance, in other words.

Karel Čapek, the Czech writer whose 1929 book The Gardener’s Year is one of the most honest accounts of the trade, put it plainly: a garden is never finished, and in that sense it resembles every other human undertaking (Čapek, 1929). Gardens don’t finish, and programmes don’t finish; the language of completion misleads us from the start. A programme we launched six months ago isn’t “done”; it’s in its second season, and it wants attention. A capability framework we wrote two years ago is a border that’s grown into itself, with some plants thriving and others crowding out their neighbours; it wants someone standing in it often enough to notice the difference.

There’s a kind of freedom in accepting all of this. It lets us stop treating every unintended outcome as a failure of planning, which it usually isn’t; the organisation is a system more complex than any of us can fully see, and the best we can do is design with care, and stay involved long enough to respond to what shows up. Some of what shows up will be welcome in ways we didn’t see coming; some will need dividing or removing. The work is the tending, either way.

Plant thoughtfully, expect to be surprised, and stay in the garden.

Leave a comment

References

Čapek, K. (1929) The Gardener’s Year. London: Allen & Unwin.(PAID)

Meadows, D. H. (2001) ‘Dancing with Systems’, Whole Earth, Winter 2001.
(The link here is to a 2002 reprint in the Systems Thinker.)

Avatars embedded into learning content have largely failed to deliver on their early promise; meta-analytic work on pedagogical agents shows only marginal effects on cognitive load, with design choices and content type moderating outcomes heavily (Li et al., 2025). What I think deserves more attention, is the potential for AI-driven role plays, where voice and potentially video avatars, take the part of a conversational partner and let people practise interactions privately.

Role plays, done well, are among the most useful interventions we have for developing interpersonal skill; the difficulty has always been getting people to do them. The awkwardness of being observed and the fear of getting it wrong in public mean role plays in group training sessions are often half-hearted at best. A private practice environment removes most of that social friction while preserving the repetition and rehearsal the skill requires; research in language-learning contexts shows that AI conversation bots can improve speaking performance while reducing the anxiety that typically blocks participation (Ding and Yusof, 2025).

One of the big drivers in my thinking here is that the people now entering the workforce are doing so with a significant gap in the kinds of interpersonal skills we used to assume would be picked up informally. In the UK, research commissioned by Inspiring Learning found employers rated over a third of their Gen Z employees as lacking basic communication skills (Gentle, 2024). Work from the University of Minnesota with emerging adults during the pandemic documented reduced in-person communication and increased psychosocial distress during the years when these skills would typically be building (Alexander et al., 2021). Calling this a generational failing, as some do, helps nobody; it is the predictable consequence of young people growing up under socio-economic conditions most of us would find intolerable, and then losing their developmental years to lockdown.

The usual L&D response is to hand-wave this with generational stereotypes, which gets us nowhere. A more useful response is to accept that foundational interpersonal skills work now needs to happen much earlier in people’s careers than we used to assume; the ability to have a difficult conversation, or to push back on a colleague without the wheels coming off, can no longer be taken as given. If AI roleplays can scale that kind of practice in a way that respects how uncomfortable it is to learn these skills, they have a meaningful contribution to make.

I remain cautious about most of what AI is being sold into L&D to do; avatar-driven video and mass content generation are largely solving the wrong problems. Private roleplay, with decent feedback, at the point where people most need it, is one of the first applications I have come across where I think the technology could make a meaningful contribution to people’s working lives.

Leave a comment

References

Alexander, T., Larson, N., Berge, J. and Neumark-Sztainer, D. (2021) Changes in interpersonal relationships of emerging adults during the COVID-19 pandemic: adaptations and implications for psychosocial distress. University of Minnesota School of Public Health.

Ding, D. and Yusof, A.M.B. (2025) ‘Investigating the role of AI-powered conversation bots in enhancing L2 speaking skills and reducing speaking anxiety: a mixed methods study’, Humanities and Social Sciences Communications, 12, 1223.

Gentle, S. (2024) ‘New research shows young people entering workforce with a lack of soft skills including communication and resilience’, Onrec, 17 September.

Li, H., Wang, Z., Ding, L., Zhang, J. and Wang, G. (2025) ‘The facts about the effects of pedagogical agents on learners’ cognitive load: a meta-analysis based on 24 studies’, Frontiers in Psychology, 16, 1635465.

Recognising that is the point at which the work gets harder. We’re told to treat training as the last resort, and we should, but there’s a risk that we become the function known for saying no, for shrugging, for drawing a big red circle around a problem and walking away from it. Nobody wants to work with the team that does that, and for good reason.

The work, then, is learning to hand off well. A few approaches have proved reliable.

Show your working. When you’ve done the interviews, reviewed the data, mapped the process, and arrived at the view that the problem sits outside training’s remit, bring all of that with you. Don’t just tell the head of operations that the issue is their tooling; walk them through how you got there, what you looked at, who you spoke with, and what the patterns told you. A well-evidenced handover is respected; a bald assertion is dismissed.

Offer options. You don’t run their function, and you know less about its constraints than they do. Present what you’ve seen and a few directions the evidence points toward, while making clear that the decisions about what’s feasible and what fits are theirs. That framing keeps ownership where it belongs, while still giving them something useful to act on.

Reduce the friction of acting on it. You’re asking someone else to pick up work, so keep the ask clean. Summarise neatly, package the evidence so it can be referenced later, and offer to support the conversation with their own team if that would help. The quality of a handover shows in how easily it can be picked up.

Stay involved where it makes sense. A handover isn’t a goodbye. Offer to help test an assumption through a small experiment, and check in once a change has been made to see whether the performance indicators have shifted. That’s what separates collaboration from cooperation, and both from the more corrosive pattern of working around each other, or worse, competing for credit.

We can’t own everything, and we shouldn’t try to, but the people who do good performance work influence change they don’t control. That’s where L&D’s influence lives: in the quality of the handovers we create.

It also made me think about curiosity, and about how impoverished our version of it has become.

In the world of work, we talk about curiosity a lot; it appears in competency frameworks, in leadership development programmes, in culture audits, and in the language of L&D strategies that promise to create ‘curious learners.’ What we actually mean, when we examine it, is something much narrower: we want people to be curious enough to engage with the training we’ve designed, curious enough to upskill in the areas we’ve identified as priorities, curious enough to explore, within a fairly tight radius, the things we’ve already decided they need to know. That is a legitimate and understandable constraint, given that organisations have short-term objectives and finite budgets and real performance pressures, but the gap between what we call curiosity and what curiosity has historically meant for our species is enormous.

What we watched launch into space last week was the product of curiosity in its most unreduced form. Before Artemis, before Apollo, before the Wright Brothers and Kitty Hawk, before the first fire, before every one of those civilisational leaps that we now teach as history: at no point were we ready. At no point had we surveyed the workforce and determined that fire-making or flight or orbital mechanics were the skills we’d need for the next financial year. We did those things because we looked at the world and thought, with that peculiar human stubbornness, that there was something further out there worth reaching for. We did them because we were curious.

This matters to me when I think about talent development and retention, because the version of skills planning we’ve settled into as a profession is, to put it charitably, optimistic in its claims and poor in its outcomes. We cannot predict the skills we’ll need in three years, and the annual skills surveys that circulate with great confidence haven’t improved in accuracy simply by accumulating; they’ve just accumulated. The World Economic Forum’s lists, however well-intentioned, have a habit of describing yesterday’s anxieties in tomorrow’s language. At some point, we just have to get comfortable with not knowing, and design talent ecosystems based on this.

What that might look like, in practical terms, is giving people time, flexibility, and resource to explore skills and ideas that don’t yet have a clear destination. Organisations that enrich their talent pools through broader development, rather than narrower skills targeting create the conditions for people to be prepared for roles and opportunities that don’t exist yet. The average employee is not going to the moon. But if we, with all our flaws and contradictions and geopolitical disorder, can still gather around a live stream and be awed together by a view of the moon we’ve never seen before, then surely we can afford to build a little more genuine exploration into our long-term talent strategies.

Curiosity doesn’t need to be injected into people; it needs to be given room.

Leave a comment

Note: I can say this from personal experience. Early in my career, I was like many others who thought xAPI (or perhaps CMI5) would supplant SCORM, but it quickly became clear that was not going to happen. Even after that, it is incredibly easy to get drawn into exaggerated, hyperbolic language, so it is vital that we do not necessarily buy into the hype.

The trouble is that making this statement with any confidence, let alone any accuracy, requires you to cover your eyes, plug your ears, and shout la-la-la at all the evidence that exists around you. The most concentrated piece of that evidence is about to present itself at Learning Technologies London, where thousands of people will walk through an expo hall populated almost entirely by LMSs, LXPs, and whatever other combination of letters vendors have decided sounds more contemporary this year. At their core, all of these platforms do the same thing: they are places where people go to complete learning activities and where the data from those activities is stored and managed. The branding shifts; the fundamental function doesn’t.

Get 10% off forever

I’ve always been a little puzzled by the glee with which people announce the death of the LMS. The LMS isn’t broken as a concept; it does its job. A learning management system is, at its most fundamental, a repository for the data outputs from digital learning, and that has always been its purpose. If you’ve created content that can’t be tracked, that is a design decision you made. If you’re tracking the wrong information, that is a scoping problem. If the platform you selected doesn’t suit your context, the selection process let you down.

Platforms are the butt end of many complaints in our profession, and I can’t even begin to count the number of people who’ve told me their engagement problems stem from a terrible LMS interface; and yes, some platforms are not especially inspiring to look at, but that has never been their primary purpose.

If you need total control over the experience your people have, a headless LMS gives you that. If you want something that works reliably straight out of the box, the established enterprise players like Docebo and Cornerstone are there. If you want flexibility at a lower cost point and you’re prepared to invest time rather than budget, Moodle and Canvas have been serving that need for years. In the same way there is no universally best author, there is no universally worst LMS; there is only the wrong LMS for a context, and all usefulness is contextual.

The idea that we can meaningfully separate the LMS from the LXP from every other L-prefixed platform is, to a large extent, a story the vendor market tells itself. Fosway Group, whose 9-Grid for Learning Systems is the most rigorous independent analysis of the European learning platform market, dropped the LMS and LXP classifications entirely in 2020, reclassifying everything as Suites or Specialists, precisely because the distinctions had become more marketing than substance (Fosway Group, 2020). Their 2026 edition of the same research describes a market that is “highly competitive and still growing” (Fosway Group, 2026). Grand View Research estimates the global LMS market at approximately USD $28.58 billion in 2025, with projections reaching $70.83 billion by 2030 (Research.com, 2026). A dying category does not post those numbers.

The parallel with SCORM is instructive. xAPI and CMI5 were both developed as explicit successors to a specification that has been embedded in the industry since 2001; neither has come anywhere close to displacing it in day-to-day organisational use. Technologies that embed themselves in compliance frameworks, procurement cycles, and institutional infrastructure do not vanish because superior options emerge; they persist because switching costs are real, and “working reliably” has a strong competitive advantage over “architecturally elegant but requiring a full migration.”

The LMS will change, as it already has; every new feature release, every AI integration, every UX overhaul shifts the landscape incrementally. That is evolution, and confusing evolution with death does the profession no favours.

Share

Leave a comment

References

Fosway Group (2020) The LMS, NGLE and LXP are dead!, Fosway Group [online].

Fosway Group (2026) Learning Systems Comparison, Fosway Group [online].

Research.com (2026) 51 LMS Statistics: 2026 Data, Trends & Predictions, Research.com [online]. [Market size figures attributed to Grand View Research, 2024.]

This is the fifth article in the UX for L&D series, and it is the one most directly focused on digital learning design. The principle applies broadly, as every article in this series has tried to show, but the research here comes from web usability, and the most direct application is to the screens we build in e-learning tools, LMS environments, and digital performance support. If you design content that people engage with on a screen, the findings below are worth taking seriously.

Eye-Tracking Research

The Nielsen Norman Group first published findings on how people read web content in 2006, based on a study tracking the eye movements of 232 participants across thousands of web pages (Nielsen, 2006). The headline finding was: in the absence of good visual design, people read in a shape that roughly resembles the letter F. They read across the top of the content, move down and read across a shorter second line, and then scan vertically down the left side of the page, with decreasing attention as they go further down.

A follow-up study eleven years later found the same patterns holding across both desktop and mobile devices (Pernice, 2017). The F-pattern, it turns out, is not a product of early web design habits that people would grow out of; it reflects something more fundamental about how people navigate unfamiliar text when they have limited time and motivation to read everything.

The F-pattern is that it is not desirable behaviour. It is what happens when design gives people no better option. When a screen presents a wall of text with no clear hierarchy, no meaningful headings, and no visual signals about where the important content sits, people default to the F-pattern because it is the path of least effort. They read what is early, they read what is left-aligned, and they skip the rest.

The 2019 NN/g research expanded the picture considerably by identifying three other scanning patterns that appear alongside the F-pattern depending on how content is structured and how motivated people are to engage with it (Pernice and Whitenton, 2019).

The layer-cake pattern occurs when meaningful headings and subheadings are present and visually distinct. People skip between headings, assess whether the section beneath is relevant to them, and read more carefully when they find something that matters. This is the most efficient scanning pattern for information-seeking tasks, and it is achievable through good heading design.

The spotted pattern occurs when someone is looking for something specific, such as a number, a link, or a keyword. They skip large sections of content entirely in search of the target. This pattern tends to reveal poor content architecture: if people are spotted-scanning an e-learning module, it is usually because the thing they need is buried rather than surfaced.

The commitment pattern is the closest thing to the linear reading most e-learning designs assume. It occurs when someone is highly motivated to read carefully, either because the content is directly relevant to something they care about, or because they know they will be assessed on it. Importantly, even the commitment pattern benefits significantly from good content structure; chunked, well-headed content produces better comprehension than walls of text even when people are genuinely trying to read everything (Pernice and Whitenton, 2019).

What This Means for How We Design Screens

Content placement is not neutral

The top of the screen receives the most attention; the left side retains more attention than the right as the eye moves down the page. Content placed in the lower-right area of a screen is, in the absence of strong visual signals directing attention there, largely invisible to most people. This matters enormously for where we place key information, critical instructions, and anything that people must understand before they proceed.

The first words of every line carry disproportionate weight

In the F-pattern, people fixate on the beginning of lines and then make a rapid decision about whether to continue across or drop to the next. If the first few words of a paragraph do not signal relevance, the paragraph will be skipped. This argues strongly for front-loading: the most important word or phrase in any sentence should appear as early in that sentence as possible, and the most important sentence in any paragraph should appear first.

Headings do more work than most e-learning designers give them credit for

In the layer-cake pattern, headings are the navigation system. People use them to decide what to read and what to skip. A heading like “Section 3” or “Key Points” tells someone nothing about whether the content below matters to them; a heading like “What to Do When a Customer Escalates a Complaint” tells them immediately. Meaningful, specific headings are the single highest-leverage design change available for improving how people move through digital content.

Walls of text are not read

This sounds obvious, but the evidence is that they are not even scanned efficiently; they produce the F-pattern, which means that anything more than a few lines into a block of unbroken text will receive little or no attention. Chunking content into shorter paragraphs, separating sections with visual whitespace, and using formatting to signal structure are all practical responses to the way people actually scan.

The right side of the screen is largely unused for text content

This does not mean the right side of the screen is useless; images, diagrams, and visual elements placed there can attract and hold attention effectively. But columns of text on the right side of a two-column layout will typically receive far less attention than the equivalent content on the left.

If you want to up your learning science game, join us on the 29th of May 2026 in Birmingham for IDTX Evidence-Informed practice Conference where research meets practice.

As a reader of the Instructional Design Tips Substack, you can get 25% off your ticket using code CPDW25 at checkout.

Book your ticket today!

Learning Objectives

A great deal of existing e-learning begins with a list of learning objectives on the first or second screen. This has been common practice for decades, grounded in the reasonable idea that people should know what they are going to cover before they cover it. The problem is that, in practice, these screens are almost universally formatted in ways that guarantee they will be scanned in the F-pattern and largely ignored.

A bulleted list of objectives that begins “By the end of this module, you will be able to...” presents the least important words first in every line, places the most specific content towards the end of each bullet, and provides no visual hierarchy to signal which objectives matter most. People read the first bullet partially, skim the beginning of the next two or three, and click next.

If objectives screens are going to be included, the research on scanning behaviour suggests they should be written with the most meaningful, specific words at the start of each objective, kept to three at most, and given enough visual separation that each one reads as a distinct item rather than a list that blurs together. If there are more than three objectives, the question worth asking is whether the module scope is too broad, not how to format a longer list.

Note: The above does not represent well-designed learning objective screens, only how people are likely to interact with the text. I do not recommend starting digital learning with a list of learning objectives, but since this practice is still common across much of the industry, if you do use one, this should be taken into account.

Designing for the Commitment Pattern

The commitment pattern, in which someone reads carefully and thoroughly, is achievable in digital training contexts, but it requires two things that are often absent: content the person believes is directly relevant to something they care about, and design that does not punish careful reading.

On the first of these, eye-tracking research is clear that motivation is the primary driver of the commitment pattern (Nielsen, 2006; Pernice and Whitenton, 2019). People read carefully when they believe the content will help them do something they want or need to do. So, if the programme is not perceived as relevant to real work, no amount of good layout will produce careful reading.

On the second, the research shows that even committed readers, those in the commitment pattern, comprehend more when content is structured with clear headings, shorter paragraphs, and deliberate whitespace. Good design does not interrupt careful reading; it supports it. The argument sometimes made that formatting and chunking are only necessary for low-engagement content is not supported by the evidence.

A useful exercise is to take any screen you have built and ask two questions:

• where would someone’s eye go first if they arrived at this screen with moderate motivation?

• would the content that receives the most early attention be the content that matters most?

If the answer to the second question is no, the screen needs redesigning before it needs editing.

A Practical Review Process

Incorporating scanning behaviour into design review does not require eye-tracking equipment or specialist software, though both are available and informative if you have access to them. A structured walk-through of any material using the following questions gives a reasonable approximation of what the research predicts.

• Cover the lower-right quadrant of the screen and assess how much of the critical content remains visible. If the answer is very little, the layout needs reviewing.

• Read only the headings, ignoring all body text, and ask whether someone who read only those headings would have a clear sense of what the screen covers and why it matters. If they would not, the headings need rewriting.

• Ask whether a motivated person who wanted to read carefully would be helped or hindered by the current structure. If the structure imposes cognitive work on someone who is trying to engage, it is working against the content.

These questions do not replace user testing, which remains the most reliable way to understand how people engage with a given piece of content, but they make scanning-behaviour research actionable in everyday design practice without requiring significant additional time or resource.

Designing for how people actually read is a straightforward act of respect: if we understand that people scan before they read, and that scanning behaviour is shaped almost entirely by the design decisions we make, then building screens that ignore those patterns is a choice to prioritise our convenience over their experience, and the evidence suggests performance suffers accordingly.

Leave a comment

References

Nielsen, J. (2006) ‘F-shaped pattern for reading web content’, Nielsen Norman Group.

Pernice, K. (2017) ‘F-shaped pattern of reading on the web: Misunderstood, but still relevant’, Nielsen Norman Group.

Pernice, K. and Whitenton, K. (2019) ‘Text scanning patterns: Eyetracking evidence’, Nielsen Norman Group.

Nielsen Norman Group (2022) ‘The layer-cake pattern of scanning content on the web’.

As always, I want to start by thanking you for being a paid subscriber to the Instructional Design Tips Substack. Your support lets me keep writing here five days a week, and I’m also publishing this omnibus edition as a thank you to you.

If there are topics you’d like to see covered in a future Members Only article or in any of the daily articles, please let me know in the comments or in the chat. I’ll always prioritise talking about the things you’re interested in over the ideas floating around in my head when I’m writing.

In this edition:

• My Infographic Problem

• Quick Definition: Neuroplasticity

• The Questions I Ask Before I Book a Conference

Read more

The instinct in this situation is usually to reduce the ask: send shorter emails, use async tools, drop a review link and hope they find ten minutes before your deadline. I’ve tried all of these, and what I’ve found is that they tend to make the problem worse, because you’re asking a busy person to fit your work into the gaps between their actual work, and there rarely are any.

The more effective approach, and I appreciate it sounds counterintuitive, is to book more time, not less.

Why synchronous time works

When someone’s diary is already under pressure, asynchronous tasks are the first casualties. A document waiting for comments, a survey left half-finished, an email thread requiring a considered reply: all of these compete with everything else demanding attention in the moment, and they frequently lose. A blocked hour in the diary is protected from that competition; your SME knows that between 10am and 11am on Thursday, they’re working on this, and everything else waits.

The key then is making sure that protected time is used well, which means arriving with a clear purpose. Depending on where you are in the project, that might mean:

• Answering scoping questions: What are we trying to achieve? What do we need people to be able to do? What are the blockers?

• Reviewing content or drafts: Walk me through this section; tell me what’s missing, wrong, or unclear.

• Structured design activities: Card sorting to sequence content, assumption mapping to surface what we think we know and what we need to verify.

Booking the review into the build

One approach I’ve found particularly useful is to schedule SME reviews directly into the project timeline at the point of planning, rather than arranging them reactively when something is ready. If you know you’ll need a technical review of your alpha e-learning build in week four, book the call in week one. On the day, have your SME work through the module while sharing their screen, commenting live, with you listening, noting, and probing anything that isn’t clear. Record the call.

At the end of an hour, you have your SME review. You haven’t chased anyone. You haven’t sent three reminder emails. And your SME spent one bounded, focused hour rather than five fragmented ones.

This approach may require you to be available at times that suit them rather than you, and that’s a reasonable trade. Processes should be designed around what produces the best outcome in the least time for everyone involved, not around what’s most convenient for the person organising them.

Synchronous time, used well, protects the collaboration that async time leaves to chance.

Leave a comment

Senior leaders who attend these things tend to come back energised, which is great news, but the instinct inside many teams is to treat a returning CEO armed with a conference souvenir idea as a threat to be managed. The more useful perspective is to recognise that a leader who is actively seeking out ideas, attending events, and thinking about how to improve the organisation is an asset. The problem isn’t that they’ve heard an idea; the problem is that not every idea travels well from a conference stage to a specific organisation’s context, and our job is to help them tell the difference.

That job starts with assessment. Before any conversation about why something might be a poor fit, we need to have examined the idea on its merits: what is it based on, what does the evidence say, and where has it worked or failed? From that foundation, the conversation with a senior leader shifts from “I don’t think that’s a good idea” to “I’ve looked at this carefully, and here are the risks I’m concerned about.”

Framing concerns as risks rather than objections changes the dynamic of the conversation. Business leaders understand risk; they weigh it against reward every day. Presenting your assessment in terms of short, medium, and long-term risk, tied directly to the things the organisation cares about, is far more likely to be heard than a general expression of scepticism.

In these conversations, it’s important that you can articulate the short-, medium-, and long-term risks you foresee in a potential intervention. The specific risks will vary based on the organisation or intervention you’re considering, but here are some common short, medium, and long-term risks.

Short-term risks

• Immediate morale impact if employees perceive the initiative as poorly considered or inconsistent with the organisation’s direction

• PR exposure if the programme requires external communication before it’s ready, particularly if it invites public scrutiny

• Recruitment pipeline damage if the initiative signals instability or a culture at odds with what candidates are seeking

Medium-term risks

• Financial cost of reversing an initiative that hasn’t worked, including the resource required to rebuild what was disrupted

• Increased employee attrition if confidence in senior decision-making erodes, making recruitment more expensive and competitive positioning harder

• Reduction in L&D credibility, which makes it harder to gain buy-in for initiatives that do have a strong evidence base

Long-term risks

• Sustained revenue impact if the initiative affects customer-facing capability, service quality, or brand perception

• Regulatory or audit exposure if the programme touches areas governed by compliance requirements

• Competitive disadvantage if the organisation signals to the market that it follows trends rather than strategy, making it harder to attract senior talent and retain key clients

The ability to have this conversation well depends almost entirely on preparation that predates it. If you already understand your organisation’s position through tools like Porter’s competitive framework (Porter, 2008), maintain a live analysis of organisational strengths, weaknesses, opportunities, and threats, and have mapped the metric chains that connect people decisions to business outcomes, you will have the evidence to speak credibly about risk in your specific context, rather than in general terms that any leader can dismiss.

Leave a comment

References

Porter, M.E. (2008) ‘The Five Competitive Forces That Shape Strategy’, Harvard Business Review, 86(1), pp. 78–93.