ADS Mediapilote, a Google Ads, Meta and TikTok platform

Mediapilote’s paid search team runs campaigns for many clients on Google Ads, Meta and TikTok. Tracking relied on each network’s interface and on an Excel file shared between traffickers, designers and project managers.

The project started in 2025 as a smaller tool: a React and Express app to create Performance Max asset groups with dynamic variables. In 2026 it became a full platform, connected to the intranet for clients and users.

Each ad network has its own API, its own concepts and its own limits. Data from all three had to be synced incrementally into a shared database, and client reports had to stay consistent even when data was missing from the sync window.

Production tracking raised another problem: a workflow with several people involved (brief, creatives, client approval, launch) where each role should only see and do what concerns them, without losing input along the way.

I wrote the Google Ads, Meta and TikTok clients, the scheduled sync and the client reports with PDF and Excel export, then a workflow engine that alone decides production status transitions, with notifications and a reliable auto-save queue. Importing the old Excel file brought over the history, and conversions relayed by the intranet are uploaded to Google Ads through the Data Manager API.

On the AI side, an assistant with function calling queries the Google Ads API live and answers with charts and tables. I also exposed an MCP server protected by OAuth 2.1 so the team’s AI assistants can read and edit campaigns. The first version, picked up again in 2026, had to move from v20 to v25 of the Google Ads API.