The problem this solves
Every major platform now publishes an ad library, which means your competitors' entire paid presence is public record. Almost nobody reads it, because checking the Meta, Google, LinkedIn and TikTok libraries for even three competitors means dozens of searches, screenshots pasted into a doc, and an afternoon lost every time someone asks how loud the competition is this month. So the question gets answered from memory, or from the one ad somebody on the team happened to see.
Without a share-of-voice read, media planning runs blind. You cannot tell whether a rival went quiet or doubled down, whether they shifted budget from search to social, or which message they trust enough to keep running. Teams that watch ad libraries systematically spot repositioning weeks before it shows up anywhere else, because sustained creative spend is the most honest signal a competitor gives about what is working for them.
How the mission runs
- Sweep the libraries. Ad Libraries pulls every active ad for the competitors you name across the Meta, Google, LinkedIn and TikTok libraries in one pass. The agent records counts, formats, and first-seen dates so volume can be compared on equal footing rather than eyeballed from a few open tabs.
- Read the creative angles. The agent reads the creatives themselves - headlines, hooks, offers, and the audience each seems built for - and clusters them into the handful of angles each competitor is actually leaning on. Long-running ads get flagged, since sustained spend usually marks a message that converts.
- Compute share of voice. Active ad counts are rolled into a share-of-voice table by competitor and by platform, with movement noted against any earlier run of the same mission. The point is a defensible relative read: who is loudest, where they are loudest, and with what message.
- Post the summary to Slack. A formatted digest lands in the Slack channel you specify - the share-of-voice table, each competitor's dominant angle with a linked example ad, and a short note on anything that changed since last time. If you gate outbound messages, the post waits in your approval queue until you release it.
The prompt
This is the exact objective the agent receives. Swap the obvious placeholders for your own domain, segment or channel and run it as-is from the console, Slack, or the API.
What comes back
A Slack post in the channel you choose containing a share-of-voice table across the four ad libraries, a one-line read on each competitor's leading creative angle with example ads linked, and a change note when the mission has run before. The full ad inventory behind the summary stays attached to the mission run, so anyone who wants the receipts can open them.
Make it yours
- Add your own brand to the comparison so the digest reports your position in the share-of-voice table alongside the rivals you are tracking, week by week.
- Ask for a platform-level breakout that flags where each competitor concentrates its spend, useful when you are deciding which channel to contest next quarter.
- Run it weekly on a schedule and have the agent lead with deltas, so the channel sees movement first and totals second, and quiet weeks stay short.
Frequently asked questions
How is ad volume measured across different libraries?
The agent counts active creatives per competitor per library at run time and compares within each platform. Libraries report different metadata, so counts are presented per platform along with an overall share, and the digest states exactly what was counted so nobody mistakes it for spend data.
Can it tell me how much competitors are spending?
Ad libraries publish creatives and run dates rather than budgets, so the mission reports volume, duration, and concentration as spend proxies. Sustained volume across many placements is the strongest available public signal, and the digest is explicit about what is observed versus what is inferred.
Does the Slack post go out automatically?
That is your call. With approvals on, the digest is staged in your queue and posts only when you release it. Teams that run this weekly usually approve the first few, confirm the format is right, then let scheduled runs post directly to the channel.