How The Well automated client reporting across 50+ teams
The Well built every client report by hand from downloaded programme data. On top of its Notion operating system, we built two AI Skills that turn the daily synced data into a reviewed, client-ready report and PDF, then refined them with the team until every report came out the same way.
50+ teams reported on from one set of live data
Company reports built by AI Skills from data synced every morning, and checked by a person before they go out
The Well is a South African team culture business that runs monthly facilitated conversations for corporate teams through WhatsApp. Once its Notion operating system was live, with program data syncing in from Airtable every morning, the next job was reporting. How we built the operating system
What was going wrong
Every client report was built by hand. Aret downloaded the program data from Airtable and assembled each report piece by piece. The Well needed several kinds: a quarterly report for each client company, an update after each round of sessions, and a quick snapshot when a client's leadership asked how a team was doing. With about 50 active teams and more on the way, that wasn't going to scale.
How we built it
Matt led the work over two phases, building the reports on top of the data the operating system already held.
Centralisation. Every client company, team, member and response syncs from Airtable into Notion each morning. The sync runs on a server rather than anyone's laptop, and can be rerun with one button when something changes.
Collaboration. We agreed with the team what a company report should show: headline stats against the previous quarter, personality profiles, feedback themes, and how teams engage between sessions. Reports stay at company level and are filtered by team when a leader asks, so 50+ teams don't mean 50+ reports to maintain.
Personalisation. Dina asks her own questions of the engagement data with Notion AI, instead of waiting for a report.
AI, once the structure was in. One AI Skill builds an interactive report page from a client's data, with charts and the client's logo. Aret reviews and edits it, and a second skill turns it into a client-ready PDF, saved with the date it was sent.
What went wrong on the way. The data was there, but what each report needed to show wasn't. We iterated on real reports with Aret and Dina, updating the skills and adding data checks each round, until every report came out the same way each time.
What changed
- Company reports are generated from live data in Notion, checked by Aret and sent as a PDF, with no downloads or copy-and-paste.
- Every report is logged with its date, so the team can see what each client received and when.
- Data problems that used to surface in front of a client are now flagged inside Notion first.
- Dina answers her own questions about engagement and content without waiting for a report.
- The process is built for growth: about 50 active teams today, with a target of 80.
Who owns it now
Aret runs the reporting: generating each report, editing the page and creating the PDF. Reports are pulled after the morning sync has finished, and Aret can rerun the sync when something changes during the day.
Built the same way for a marketing agency. We used the same approach for Atlas, a healthcare marketing agency, with campaign data from four ad platforms instead of program data. Read the Atlas case study
Still building client reports by hand from exported data? Talk to us about automating them in Notion.
Want results like these in your business?
Talk to us