Most people learn their job by being near someone who already knows it. You overhear a call and pick up how a difficult client gets handled. You watch someone open a spreadsheet and notice the shortcut you’d never have found in a manual. None of it appears in a training plan, and for decades none of it needed to.
Then teams stopped sitting together every day, and that quiet transfer of knowledge thinned out without anyone deciding it should. The result isn’t a training gap that shows up in a report. It shows up as a junior colleague who is slower than they should be, and nobody can say exactly why.
The tools nobody says they can’t use
Count the platforms your organisation has introduced in the last two years, then consider how many of your people use more than a fraction of any of them. Some adapt in an afternoon. Others quietly work around the thing by exporting to a spreadsheet, keeping a parallel system, asking a colleague to do the bit they never learned. When everyone shared a floor this was self-correcting: someone leaned over, noticed, and fixed it in thirty seconds. Across offices and home desks, the workaround just persists.
The evidence for who’s stuck already exists. Most admin consoles report feature-level usage per seat, so dormant licences and untouched functions will tell you where to aim. Repeat helpdesk tickets are the other signal. The same question from different people is a training gap, not user error. And asking people what the slowest part of their week is will usually surface something a fifteen-minute session would fix.
Then train narrowly. Fifteen minutes on one feature, four weeks after onboarding rather than during it, when people have hit the problem it solves. Two or three specific recordings linked from where the question arises, not a library nobody opens. And a named owner for each tool whose job includes noticing when a system isn’t landing.
Rebuilding development when people aren’t co-located
If in an organisation capable new hires taking far longer to find their feet than they used to, then the fix isn’t a bigger learning budget. Most of what was lost was informal, and informal things have to be rebuilt deliberately. These are the mechanisms that tend to hold.
Make the reasoning audible.
Early-career people rarely lack information; they lack access to judgement. They can read the decision but never heard the argument behind it. The cheapest correction is to spend the last five minutes of any recurring review on why we chose that, with someone junior in the room. Record it if the meeting is already being recorded.
Pair on real work, not on abstractions.
Standing weekly hour, same slot, screen shared, junior driving their own live task with a senior watching. It works because it’s specific and it’s scheduled. General offers to help never get taken up.
Cut the cost of asking.
Questions that used to be a turn of the head are now a direct message someone has to work up to sending. Move them into a shared channel where answering once serves ten people, and establish the norm by having senior people visibly ask questions there themselves. That single behaviour does more than any policy statement.
Bring people in as a cohort where you can.
A graduate joining alone has nobody to compare notes with. Three joining the same week teach each other constantly, and the effect lasts years. If your hiring doesn’t work in cohorts, group new starters across departments instead, the peer relationship matters more than the shared discipline.
Write down what used to be passed on verbally.
Distributed teams that work well are almost always the ones that document decisions and how-tos properly. The test is simple: can a new starter answer their own question in under five minutes without interrupting anyone? If not, the gap is retrievable knowledge, not training.
Reverse the direction occasionally.
Have junior people present work and senior people respond. It surfaces misunderstandings that would otherwise sit undetected for months, and it gives quieter staff a reason to be visible.
Design the first ninety days on purpose.
A named buddy with a weekly slot, real work inside the first fortnight rather than a month of reading, and a scheduled check at thirty, sixty and ninety days that asks what’s still unclear rather than how it’s going.
Knowing whether any of it is working
Course completion rates tell you nothing. A handful of measures that do:
- Time to first independent delivery. Track how long it takes a new starter to complete something end to end without supervision, and compare cohorts. If that number is drifting upwards, development is eroding regardless of what your engagement survey says.
- The shape of question volume. Questions rising then falling over a few months is healthy. Flat and low from week one is not. Silence usually means someone has stopped asking, not that they’ve understood.
- Who gets asked. If one person answers everything, you have a bottleneck and a single point of failure. Spreading that load is itself a development exercise for the people who take some of it on.
- Why early-career people leave. Their exit conversations are the most direct evidence you’ll get, and the one most organisations don’t separate out from the general picture.
Who is actually learning from whom
There’s a question underneath all of this that most organisations can’t answer: when your people are in, who are they actually with?
If your graduate intake comes in on Tuesdays and the senior people they should be learning from come in on Thursdays, no amount of learning budget will fix it. The two groups are running a distributed working pattern past each other. That’s not a training problem, it’s a scheduling one, and it’s visible in your occupancy data if you look.
Four things worth checking:
- Overlap between seniority bands, not overall attendance. A floor at 70% occupancy can still have almost no contact between the people who need each other.
- Team co-presence. What proportion of days is a full team in together, rather than a rotating half of it?
- Whether anchor days collide. Teams that depend on each other choosing different in-office days is one of the most common and most correctable problems in the data.
- Where people sit when they’re in. Grouping new starters together is good for peer support and bad for absorbing seniority. Most organisations need both, arranged deliberately rather than by whoever booked first.
This is where workplace data earns its place. Capella is workplace management software with built-in occupancy analytics, which means the overlap between teams, seniorities and sites is something you can look at rather than assume. Anecdote will tell you the office feels busy. Data will tell you whether the people who need each other are in it at the same time.