The problem this solves
Audiences decay in both directions. New companies start matching the criteria and never get added; contacts who became customers keep seeing acquisition ads; people who changed jobs stay matched to an account they left. An audience uploaded once is accurate for the month it was made, and every month after that the campaign pays a little more to reach a little less of the right audience.
A refresh should touch only what changed. Rebuilding from scratch resets the platform's learning and produces a new audience name every month; adding the new matches and removing the departures keeps the same audience live, keeps its history, and keeps the campaign's optimization intact. The size delta, posted where the media team looks, is the evidence that the audience is being maintained.
How the mission runs
- Re-run the criteria behind audience <name>. On the first of the month the Audience Agent re-executes the saved criteria - the list, the ICP filter, the signal conditions - against current data to produce this month's full set of matching contacts and companies.
- Diff against the current audience members. The new set is compared with the members recorded at the last refresh. Contacts and companies that newly match are marked as additions; those that no longer match, and anyone who became a customer in the CRM since the last run, are marked as removals.
- Apply additions and removals on both platforms. Additions are added and removals are removed on the LinkedIn matched audience and the Meta custom audience, staged as one batch and applied on approval, so the audiences stay the same objects with updated membership.
- Post the size delta to #paid. Once both platforms report the updated matched sizes, a post to #paid states the additions, the removals and the new size for each audience, with the reasons for removal summarized.
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
The LinkedIn matched audience and Meta custom audience behind <name> kept current every month: newly matching members added, members who no longer fit or who became customers removed, both applied to the existing audiences, and a #paid post with the additions, removals and updated matched sizes. A membership history is kept so the audience's growth can be charted.
Make it yours
- Run weekly for audiences built on fast signals such as intent or hiring.
- Add Reddit custom audiences to the refresh alongside LinkedIn and Meta.
- Post the delta to Telegram or email instead of Slack.
- Keep removed customers in a separate expansion audience instead of dropping them entirely.
- Send the delta to the account owner in the CRM when the audience is an ABM list with named owners.
Frequently asked questions
Why keep the same audience rather than recreate it?
Campaigns attached to the audience continue without interruption and the platforms retain what they learned about it; recreating resets both.
Where do the customer removals come from?
From the CRM connection - any contact or company that reached a customer stage since the last refresh - and from an attached customer list if you prefer.
What if the criteria need to change?
Edit them in the thread; the next refresh applies the new criteria and notes the change in its post.
What happens if a platform rejects part of the update?
The accepted and rejected rows are reported separately with the platform's reason, and the rejected rows are retried on the next refresh.
Can the refresh also update an exported file?
Yes. For platforms served by export files, the refresh regenerates the files in Drive so the uploaded audiences can be replaced with current ones.
Does the refresh touch campaigns?
No. It changes audience membership only; campaigns, budgets and bids stay as you set them.