How Docebo used the Companion browser extension to close its own last-mile problem.

Training almost never fails in the room. It fails three weeks later, at 4pm on a Tuesday, when someone opens a deal in Salesforce and blanks on the discount tier they learned in a workshop they can barely remember attending.

That’s the moment nobody owns. Completion rates said the training worked. Evaluation scores said people got it. Everyone nodded along in the room.

Then everyone went back to their desks, and the actual test started.

That’s where things get shaky. People remember the concepts, but executing in real time is harder than it looked in the training deck. So they improvise. They ask a coworker. They fall back on the old way, just this once.

What that gap actually costs you

This is a delivery gap. And it’s expensive. That rep doesn’t stop the deal to relearn the discount tiers. They guess, or they ping a coworker, or they dig through a shared drive for a pricing sheet that’s three versions old. Now multiply that by every rep, every question, every day: employees spend 20% of their work week just searching for internal information. Another 5.3 hours a week goes to waiting for or recreating knowledge someone already learned. People get interrupted every two minutes, so that one question doesn’t just cost the person asking. It costs whoever’s shoulder they tapped too.

When someone can’t find what they need in the moment, they don’t go looking in a course library. They tap the shoulder next to them.

So we asked a different question: what if the answer just showed up, exactly where the work already was, not a link back to a platform someone has to remember to visit, but the actual guidance, right on the page?

Turn the browser into the answer

That’s what Docebo Companion does. It turns the web tools people already work in into a learning surface: Salesforce, an intranet page, a support ticket, anywhere. No new destination to remember. No redirect back to the LMS. And it’s already included with Docebo. No separate SKU, no additional budget required.

We didn’t just build it. We turned it on ourselves, not to test a feature, but to actually live the problem we ask customers to solve.

Where we started

We didn’t roll this out company-wide on day one. We picked a handful of high-stakes moments in our own internal tools and started there.

Sales enablement. Our reps building out opportunities or negotiating a contract inside our own Salesforce instance could pull up guidance on pricing, discounting, or deal structure without leaving the deal itself.

Leadership and performance support. Our managers setting OKRs or running performance reviews inside our own performance management tool (Dayforce) could get support on how to have that conversation, right as they were building the review, not three weeks earlier in a training session they’d already half-forgotten.

And for everything else, we built in a fallback. When someone opens Companion on a page that doesn’t have specific content configured, it doesn’t come up empty. It surfaces company news and updates, so there’s still something useful there even outside a mapped moment.

That fallback is also where the simpler stuff lives. Questions like “when do I get paid” or “what’s the expense policy” used to mean pinging a coworker or digging through the intranet. Now they’re a two-second search inside the same tool someone’s already in.

Three things that made it stick

  1. We chased specific moments, not full coverage. The instinct is to load in everything, all your content, every site, day one. We resisted, and asked one question instead: where do people actually get stuck? Three answers surfaced: onboarding, manager conversations, and the same handful of questions people asked over and over.
  2. We killed friction before launch, not after. We took out the last step people would’ve had to do themselves, so the day we went live, the extension was already sitting in everyone’s browser. Then we made noise about it: an email, a mention at the company town hall, reminders in Slack, even a small feedback contest to keep people trying it.
  3. We chose focus over volume. The tighter your content maps to real moments, the more it gets used. We didn’t start by auditing every piece of content we owned. We started with the workflows that mattered, then worked backward to what would actually help.

31% more engagement — and not one new course

Two months in, engagement was up 31%. Docebo didn’t add a course, launch a campaign, or ask anyone to log in more. Reps still needed a discount tier. Managers still needed a way to open a hard conversation. What changed was when they got the answer: right inside the tool where the moment was actually happening.

The one thing that actually changed

You don’t need to map every moment before you start. Docebo found three: reps freezing mid-deal, managers walking into a review they hadn’t prepared for, and the same handful of questions landing on the same coworker every week. Find your own versions of those. Remove the friction to reach people there. Talk about it more than once. Then measure what shifts.

Somewhere in your company right now, a rep is about to open a deal and blank on a number they learned three weeks ago. Whether they guess, ping a coworker, or get the answer instantly is the difference between training that looked good in the room and training that actually holds up.