How Atlas brought four ad platforms into one Notion database and automated client reporting
Atlas was exporting campaign data from four ad platforms, cleaning it in Excel and rebuilding client reports by hand. We synced Meta and Google Ads into Notion with Notion Workers, built a CSV import route for LinkedIn and Reddit, and added AI Skills that generate bi-weekly and final impact reports from that data. A bi-weekly report now takes about four minutes to generate.
4 minutes to generate a bi-weekly client report
Meta and Google Ads data flowing into Notion automatically, less than three weeks after kickoff
Atlas is a Canadian healthcare marketing agency that runs paid campaigns for its clients across Meta, Google Ads, LinkedIn and Reddit. In Phase 1 we rebuilt their Notion workspace around a central Program Tracker. Phase 2 brought the campaign data and the client reporting into the same place.
What was going wrong
Every report started with exports. Michelle, who manages Atlas's ad campaigns, downloaded data from each platform and cleaned it into a master Excel file. Bi-weekly updates went out with the insights written by hand in an email. Final reports came from an old web template that couldn't be changed per programme, and when a client asked for slides, the whole report was rebuilt in PowerPoint. Garrett, Atlas's co-owner, called the link delivery process "a huge pain".
How we built it
Matt led the build, starting with the data and putting the reports on top of it.
Centralisation. Campaign data from all four platforms lands in one Campaign Results database. Meta and Google Ads sync automatically through Notion Workers that Matt built against each platform's API. LinkedIn and Reddit only open their APIs to vetted partners, so their exports go into an Imports database and flow into the same place.
Collaboration. Campaigns match to their programme through Atlas's naming convention, and each programme holds its KPI targets, so anyone can check delivery against guarantees. Every metric has a plain-language description, and a data flag shows which rows have been checked.
Personalisation. Ian and Garrett each work from their own dashboard, with their programmes and tasks in one view.
AI, once the structure was in. AI Skills build the bi-weekly report from a programme's data, with an impact summary, headline metrics and guarantees marked green or yellow. They turn it into an Atlas-branded PDF, and build the final impact report as a PDF or an editable PowerPoint.
What went wrong on the way. The first reports overstated clicks and reach, because each platform counts them differently from how Atlas reports them. We matched every metric to Atlas's definitions and checked the output against the platforms until the numbers agreed.
What changed
- A bi-weekly report takes about four minutes to generate from data already in Notion.
- Meta and Google Ads data refreshes on its own, with no exports.
- On a completed campaign, reach, link clicks, unique link clicks and saves matched Ads Manager exactly. Impressions came within 0.07%.
- Atlas has decided that final impact reports will go out as Notion links, with PowerPoint kept for the few clients who want slides.
- The setup costs less than the per-client reporting tool it replaces.
Who owns it now
Michelle runs the imports and generates the reports, but the process doesn't depend on her. If someone is on leave, a teammate can follow the onboarding guide, switch on the skills from Atlas's AI Skills database, and ask Notion AI to pull the right data.
Built the same way for a team culture business. We used the same approach for The Well, with programme data from Airtable instead of ad platforms, to report across more than 50 corporate teams. Read The Well's reporting case study
Still exporting ad data into spreadsheets to build client reports? Talk to us about bringing your campaign data and reporting into Notion.
Want results like these in your business?
Talk to us