This is aimed at the part of media buying that is not decision-making: filters, date ranges, custom report builders, CSV exports. Handing the connection to a reporting platform rather than to the Meta API is what removes the app review and the token rotation, and it also sets the scope — the questions it answers are reporting questions, over whatever ad accounts that platform account can already see.
A hosted endpoint from Two Minute Reports that puts live Meta Ads data in front of Claude or ChatGPT. You connect a Two Minute Reports account that already has the ad account linked, and the assistant queries through it — there is no Meta API app to create and no access token to generate or refresh, and several ad accounts can be queried inside one conversation. Because the connection runs on the Two Minute Reports platform, the same chat can also reach the other marketing sources that account has connected, Google Ads, GA4 and Shopify among them. The server keeps nothing: data moves from the source through the platform to the assistant.
- Campaign performance over a date range — spend, results and cost per result, down to a daily breakdown
- Spend and budget comparisons period over period, and which campaigns spend most for the least return
- ROAS and purchase value by campaign and ad set, including which ad sets are losing money
- Creative-level analysis: click-through rate by ad, video against image, top ads by conversion rate
- Ad set and audience breakdowns, including lookalike audiences against interest-based ones
- Placement and platform splits — Facebook Feed against Instagram Stories, and which of the two converts
- Funnel metrics: link click-through rates, ThruPlay rates, landing page view to purchase
- The same figures plotted as a chart when you ask for them that way rather than as a list
