Most ad problems don’t announce themselves. A creative fatigues and CTR bleeds out over two weeks. A catalog sync breaks and half your products stop serving. Spend spikes on a Saturday and nobody checks until Monday. None of these are dramatic on any single day, which is why they cost so much by the time someone notices.

We built Gavin for that gap between when something goes wrong and when a human looks.

Meet Gavin

Gavin is the performance marketing agent inside ShopOS. It connects to your ad accounts on Meta and Google, plus your analytics, and works with your real data. You talk to it like an analyst, and it has read your numbers before you ask.

Ask why ROAS dipped last week and it pulls the data, breaks it down by campaign and creative, and tells you what changed and what to do about it. Ask which products deserve more ad spend and it answers with your brand and catalog in mind, not generic advice.

It keeps watching after you close the tab

A conversation only helps when you start one. The problems above start on their own schedule. So Gavin runs recurring checks on the accounts you tell it to monitor:

  • Daily account audits that flag unusual spend and sudden metric drops, the things that deserve a look today rather than at month’s end.
  • ROAS digests so you get the number that matters without logging into two ad managers.
  • Creative fatigue checks that catch the early signs: climbing frequency, sliding CTR, rising costs on the same audience.
  • Catalog health checks that catch feed errors and disapproved products before they quietly eat your reach.
  • SKU quadrant reports that cross your ad spend with your store revenue and sort every product into four buckets: the ones to scale, the organic winners you’re not advertising yet, and the ones quietly draining budget.

Every run is stored. Ask Gavin “what did my last audit find” and you get the findings, the headline metric, and the full report. When Gavin pings you, it should be worth reading.

It watches everything. It touches nothing until you say so.

Out of the box, Gavin is read-only. It can pause a campaign, edit a budget, or modify an ad, but those write tools stay disabled until you enable them yourself. Until then, Gavin diagnoses and recommends, and when you agree with a change, it gives you the exact steps to make it in Ads Manager. Then it re-checks the data after you do.

A lot of AI ad tools pitch full autopilot from day one. We think trust should work the other way. Your ad account is your revenue engine, so Gavin starts with zero write access and you decide when, and whether, to hand over the keys. Plenty of teams never do, and that’s a perfectly good way to use it.

Early days

Gavin’s monitoring is young and we’re improving it every week: better signal quality, fewer false alarms, smarter thresholds. What’s already working is the daily attention. Your account gets looked at every day by something that never gets bored, never skips a day, and remembers what it found last time. Even a good marketer can’t give every account that; there aren’t enough hours in the day. Computers have the hours.

If you’re running ads and want an analyst that has read your data before your first coffee, Gavin is waiting inside ShopOS.