BabbelB2B platform
Ops is the product
Everyone wanted more content. The data said the ceiling was operational: every course needed engineer-weeks of manual setup, batch-launched once a month. We rebuilt the operational core instead of the catalogue.
-
setup
engineer-weeks to
hours
-
teacher selection
days to
hours
-
monthly batch launch cycle eliminated
- Instead of
- deeper content
- We bet on
- experience reliability and a platform foundation.
- Outcome
- setup from engineer-weeks to hours, launch any day of the month
Growth was framed as a content problem.
I led product for Babbel's B2B live-learning business — group classes run for enterprise clients. The growth conversation when I arrived was about catalogue: more courses, more levels, more content.
The data pointed somewhere else. The ceiling wasn't what we taught, it was what it cost to run a class. Setting up a single course took around two engineer-weeks, with its own JIRA ticket and engineer-run scripts. Courses could only go live in a monthly batch, behind a one-month lead rule and a cutoff on the 14th. Placements were uploaded by hand as CSVs, and teachers were assigned through Airtable forms and chased in Slack.
So I reframed what we were optimising for: not content depth, but experience reliability. That made operations the product.
One course lifecycle, before and after.
This is the full life of a course, from prep through to wrap-up. How it ran on the left, what replaced it on the right.
The rebuild had one organising idea: make a course a real object in the platform, with a central event system underneath it, instead of a row in a sheet. Once that existed, setup dropped from engineer-weeks to hours with no dedicated engineering, teacher–course matching could be pre-mapped rather than negotiated per course, and the monthly batch disappeared — a course could start whenever the client needed it. The same matching cut teacher replacement mid-course from days to hours.
Prep
AS-IS
- ~2 engineer-weeks of setup, with a JIRA ticket and engineer-run scripts per course
- One-month lead rule and a 14th-of-month batch cutoff
- Manual CSV placement uploads
- Teacher assignment via Airtable forms and Slack chasing
TO-BE
- Unified course object — setup in hours, with no dedicated engineering
- Batch cycle eliminated — courses start any time
- Pre-mapped teacher–course matching
In Progress
AS-IS
- Spreadsheet- and Airtable-driven operations
- Teacher replacement taking days
TO-BE
- Central event system driving sessions and attendance
- Control Panel sync partial
- Teacher replacement in hours, via the same matching
What actually landed.
Five foundations went into the rebuild. Three shipped and were fully adopted. Two shipped but were still ramping when the business decision landed.
- Unified course object and manual-process removal Shipped and adopted
- Central event system Shipped and adopted
- Spreadsheet and Airtable elimination Shipped and adopted
- Control Panel sync Shipped, still ramping
- OTT Dashboard Shipped, still ramping
3 of 5 fully adopted at shutdown.
Mentoring, and the C-level conversation.
This was a contract role, so the team relationship was a mentoring one rather than a reporting one — I mentored three PMs across product domains.
Growth prioritisation ran as a partnership with the C-level.