Software training isn't a content problem. It's a location problem.

Every company I talk to has the same complaint, phrased the same way: "We have all this documentation and nobody reads it."

The instinctive diagnosis is a content problem. The docs must be stale, or badly written, or hard to search — so the fix is a better wiki, a video library, a fresh round of documentation sprints. Companies have been running this loop for a decade. The docs get better. The Slack channel still fills up with "hey, quick question — how do I…"

Here's the diagnosis I believe instead: the content is usually fine. It's just in the wrong place. Help lives in one application — the wiki, the LMS, the shared drive — and work happens in another. And that gap, the alt-tab between where you're stuck and where the answer is, is wider than it looks.

The gap, itemized

Think about what actually happens when someone gets stuck mid-task. They're in the CRM, halfway through logging a deal, and they can't find where the discount approval goes. Their options:

1. Open a new tab, find the wiki, guess the search term the doc's author used, open three candidate pages, map a screenshot of last year's UI onto this year's, alt-tab back, try to remember which of the seven steps they were on.
2. Ask the person next to them, who answers in nine seconds.

This isn't a close call, and no amount of content quality makes it one. The wiki is competing with a colleague on distance, and it loses every time. That's why "nobody reads the docs" — not laziness, not bad writing. The docs demand that a person mid-task abandon their context, and a person mid-task will do nearly anything to avoid abandoning their context.

The nine-second answer, of course, has a hidden invoice. It costs the answerer their focus, it doesn't scale past a few repeat questions, it silently concentrates institutional knowledge in whoever's been around longest, and it stops working entirely when the team is remote, or across time zones, or when Sarah — it's always a Sarah — finally goes on vacation and half the org discovers they never actually learned the tool, they learned to ask Sarah.

So the org's real training system isn't the LMS anyone bought. It's an unofficial mesh network of Sarahs. The wiki is a monument built next to it.

The enterprise version of this insight costs $200,000

Here's the thing: the software industry actually figured this out years ago. The category is called digital adoption platforms — WalkMe, Whatfix — and the core idea is exactly the location fix: put the guidance inside the application, on the page, at the moment of confusion. The idea is correct. I'll go further: it's the only idea in corporate software training from the last decade that attacks the actual problem.

But look at how it shipped: enterprise contracts, per-application implementation projects, selector maintenance, admin certification programs. A WalkMe deployment is a six-figure line item with an implementation partner attached. The fix for "help is too far from the work" became available strictly to companies that could fund an integration project per app — and the guidance still had to be hand-built and hand-maintained, screen by screen, so it aged like the wiki it replaced.

Which produced today's strange equilibrium: the thousand-person company gets in-app guidance for the three apps IT blessed with a deployment budget. Everyone else — the 40-person company, the ops team on eleven SaaS tools, the department whose app wasn't in scope — gets the wiki. The location problem was solved, then priced as a luxury good.

What changed

Two things, recently.

Browser extensions matured to the point where "inside the app" no longer requires integrating with the app. An extension sees the same page the user sees, on any web tool, with zero involvement from the tool's vendor. The per-app implementation project — the thing that made the enterprise model expensive — stops being necessary at all.

And AI got good enough to read an interface and answer "how do I do this here?" against the live page, rather than replaying steps a human recorded months ago. That kills the maintenance treadmill, which was the other thing the six-figure contracts were paying for. (We learned this the hard way — we built the recording approach first, and shelved it.)

Put those together and the location fix stops being an enterprise procurement event and becomes, roughly, an install. I'm obviously not neutral here — this is precisely the bet Guidely is built on, so read this paragraph as a founder's thesis, not a market report. But the thesis doesn't depend on my product: someone was always going to collapse "in-app guidance" from a deployment into an extension, because the expensive parts were accidents of implementation, not laws of nature.

The test

If you run a team and you're weighing another documentation push, try one diagnostic first. Take the five questions your team asks most, and for each one, honestly trace the path from the moment of confusion to the answer — every tab, every search, every guess. Count the steps.

If the path is long, you don't have a content problem, and writing more content will do what it's always done. Fix the distance instead: put the answer where the question happens, whatever tool you use to get it there. Even a pinned checklist inside the CRM via a humble browser bookmark beats a beautiful wiki nobody opens.

Content answers questions. Location determines whether the question ever reaches the content. For ten years we've optimized the first thing and wondered why the Slack channel never got quieter.