AI Max on a small budget: let it expand early or keep it off until you have data?

AI Max worries me on small budgets. Say a dental clinic turns it on for a Search campaign. Google widens matching beyond your keywords and pulls in queries it considers close, which sounds fine until "dental implant cost" or "free dentist near me" starts eating budget with no bookings. The usual advice is to let the system learn for a few weeks, but I'm not sure that holds when you only get a handful of conversions a month. The bidding just doesn't have enough signal to learn what a good query looks like. I'd keep it off until you have a month of clean search term data and a solid negative list, or if you do turn it on, review the terms every few days. How are others handling the first two weeks?

Before deciding anything about AI Max, check what your conversion actually counts. With only a handful of conversions a month, an action that also fires on calls, chat clicks or newsletter signups can make the bidding look like it's learning when it's really chasing cheap junk. Make sure the primary conversion fires once per real booking, and then turn it on.

A good rule of thumb is that any expanded query that spends about twice your target CPA with zero conversions gets a negative that week. Judge the expansion by that list of queries, not by the campaign average, because the average hides the bad terms when volume is low.

Agree on the conversion check, but on a small budget I'd tighten the trigger to 1x target CPA. At 2x, one bad term can burn most of a month's spend before anyone acts. Cutting earlier costs very little, and you spot the bad patterns faster.

Tightening it makes sense, but 1x target CPA is too aggressive as a cutoff. A term with one click and no booking has already hit it, and on a clinic with a low click-to-booking rate that's just normal noise. I'd treat 1x as a review flag and only add the negative once a term has around ten clicks with nothing booked.

Also, AI Max has final URL expansion as its own toggle, and it's easy to miss. If "free dentist near me" gets sent to your homepage or a generic service page, a clean negative list won't fix that mismatch. For a small clinic I'd keep final URL expansion off, so the expansion only widens which queries trigger your ads and you still control where people land.