The problem
1Password grew from under 200 people to about 1,000 in a few years. Training had been ad hoc, run over Slack and Zoom calls. A company that size needed compliance training and a proper system for learning, so in 2023 we rolled out Docebo.
Two years later, Docebo needed more upkeep than a small L&D team could give it, and it was getting more expensive. With the contract ending, we went looking for a replacement.
Constraints
- A hard deadline. The switch had to happen on the last day of the Docebo contract. I took over the project with about six weeks to go.
- Compliance records had to survive. Completion history and certificates needed to move with the courses.
- Docebo's full data export cost too much.
- HR data doesn't follow the org chart. Our HRIS, UKG Pro, groups people by financial code, not by team and manager.
- A small team. Whatever we chose had to be easy to run.
Setting up Docebo
For the first rollout I did the administrative setup and made the integration decisions, working with IT Operations, HR Operations, and Security. I wrote the policies and structure for users, groups, courses, categories, and pages. I also mentored the HR Operations specialist who built the user export report from UKG Pro.
I was the first point of contact for technical support for most of the platform's life. In its final year a colleague took over the day-to-day work, and I handled the harder technical questions.
Choosing the replacement
We compared six platforms and got hands-on access to four. Demos only show what a vendor wants you to see, so I wrote a rubric of the tasks an instructor and an LMS admin actually do, and ran every platform through it.
That testing changed our minds. At first the team favoured another platform because its demo was stronger, but hands-on use turned up serious problems with it. I recommended Absorb. It was the fastest for building courses, had the most efficient admin setup, reported instantly, and offered phone support instead of Docebo's web tickets. It was also slightly cheaper.
Moving the data
Docebo offered a paid service to export all of our data. It cost too much, so I wrote a Python program that pulled everything through Docebo's API as JSON:
- Courses: about 170 of them, with name, code, creation date, and status.
- Enrolments: anywhere from 10 to 1,500 per course, each with status, completion date, and last access.
- Certificates, through a separate export, so compliance certificates could be rebuilt in Absorb.
- Course content for about ten important courses whose source files no longer existed anywhere else.
Not everything moved. A colleague checked with each course owner whether a course was still in use. Courses licensed from Docebo's own content library couldn't come with us. About 70 courses made the move, each with its full enrolment history.
I also helped Absorb's team understand how UKG Pro's reporting works, so the HR data feed would map correctly.
-
Before
Absorb ran early for a small group, while everyone else stayed on Docebo.
-
Switch day
Okta did the switch. Because the LMS sat behind single sign-on, the same tile opened Absorb the next morning. Employees had no new login and nothing new to bookmark.
-
After
Two months of checking that every record, certificate, and integration had come across correctly.
Onboarding learners and managers
I set up Absorb's learner interface to be much simpler than Docebo's. We made short training videos and Notion how-to pages for learners and for managers, so people could find their way around without a support ticket.
Decisions
Test the work, not the demo. A rubric of real instructor and admin tasks overturned the team's first choice.
Build the export instead of buying it. A script against Docebo's API recovered everything we needed, including courses we would otherwise have lost.
Migrate what's used. We checked every course with its owner rather than moving all 170.
Let single sign-on carry the change. The best cutover is one nobody has to think about.
What didn't work, and what I learned
Two weeks before launch, we changed how managers would reach Absorb: through the admin interface instead of the manager interface. That meant reworking the admin structure, giving managers broader permissions, and rewriting the manager handbook, all in a rush. Because UKG Pro groups people by financial code rather than by reporting line, the new setup opened a permissions gap that we then had to close.
The lesson is mine to own. When a change arrives too late to test its side effects, I should say so. Next time I'd launch on the planned manager interface, then build and test the new approach properly before switching.
Results
- Launched on time, on the last day of the Docebo contract, 1.5 months after I took over the project.
- About 1,500 learners a year on the platform.
- About 70 courses migrated with their enrolment history, and about ten courses recovered that existed nowhere else.
- $8,000 a year saved compared with renewing Docebo, plus the cost of the vendor's data export we didn't buy.
- Simpler to run: instant reporting and phone support for a small team.
What I'd do next time
- Push back on late changes. Launch the tested plan, then change it with time to test.
- Map HR data to reporting lines early. Check how the HRIS groups people before you build permissions on top of it.