The first 90 days with a performance marketing agency
What should actually happen in months one, two and three, and the point at which a bad start stops being a learning phase and starts being a bad agency.
Most agency relationships are judged far too early or far too late. Three weeks in, nothing can be concluded. Six months in, a lot of money has gone. Here is what a competent first 90 days actually looks like, so you can tell which one you are in.
Days 1-14: plumbing, not campaigns
- Account ownership transferred or created in your name, with the agency on partner access.
- Conversion tracking installed and verified, not just a pixel on the page, but a confirmed event firing on a real form submission.
- Server-side tracking for Meta where it matters, because browser-only tracking has been under-reporting for years.
- A defined offer. What exactly is being advertised, at what price framing, to whom.
- Agreement on what counts as a qualified lead. Without this, month two becomes an argument.
If week two arrives and nobody has asked you what a customer is worth, that is a bad sign independent of anything else.
Days 15-45: the learning phase is real
Ad platforms need conversion volume before their optimisation does anything useful. Early cost per lead is close to meaningless, and the single worst thing you can do in this window is panic and change the budget, which resets the learning you have paid for.
Days 46-90: compounding
- Budget shifting toward the winning combinations rather than being spread evenly.
- New creative tested against the current best performer, not against nothing.
- Cost per lead stabilising, then improving.
- Lead quality discussed, not just lead count, the point at which a good agency starts asking how many actually converted.
When a bad start is actually a bad agency
Patience is warranted for the numbers. It is not warranted for the process. Treat these as separate: a high cost per lead in month one is normal; not knowing why it is high in month two is not.
- No conversion tracking live by week three.
- Reports that show impressions and reach but not cost per lead.
- No stated hypothesis for what they are testing next.
- You are the one chasing them for the monthly call.
Month-to-month terms exist precisely so this is your decision rather than a contractual obligation. If you are inside a lock-in and seeing the list above, that is worth a conversation.