Primary vs secondary conversions: which actions should Smart Bidding actually see?

A lot of lead gen accounts mark a form fill, a phone click, a newsletter signup and a pricing page visit all as primary conversions. Smart Bidding then treats them as roughly equal, and the cheapest action tends to soak up the budget. A newsletter signup at a few dollars isn't worth the same as a booked consultation at a much higher cost, but the algorithm only knows that if you tell it, either through conversion values or by keeping the low-value actions out of the bidding signal.

My view is to mark only the actions you'd happily pay for as primary and move everything else to secondary. Secondary actions still show up in reports, so you can see the funnel, but they don't steer bids. The trade-off is less conversion volume for the algorithm to learn from, which hurts most on small budgets. Has anyone run it the other way and found the extra volume made up for the noise?

I wouldn't frame it as either/or. If you demote the newsletter and pricing page to secondary, you lose the signal that shows who's heading toward a booking, and on a small budget that's the one you can least afford to lose. A better setup is to keep most actions primary and give each one a value that reflects what it's worth to you: a few dollars for the newsletter, less for a pricing view, the real figure for a booked consultation. That way Smart Bidding can learn from the cheap actions without treating them as equal to the expensive one.

The catch is that the values have to be grounded. If you pick them by gut feel, you get the same distortion back in a different form. Base them on the close rate you actually see from each action, even a rough one pulled from your CRM.

Setting values only matters if the bid strategy actually reads them. Maximize Conversions optimizes for count and ignores conversion values, so a $5 newsletter signup counts the same as a consultation. Values only steer bids under Maximize Conversion Value or a tROAS target.

For form fills and calls, set the count to "One" per click or session. Otherwise someone who submits twice gets double credit.

Agreed on the count settings. "One" per click matters most for calls, because Google counts every call that runs past the duration threshold, so someone who phones back twice in an afternoon can register two conversions unless you set it to one per click. For a service business where people keep ringing the same plumber while they wait for a technician, that quietly inflates the numbers the bidder is learning from.

"One per click" only collapses calls that come from the same click. If the same person searches again a few days later and phones from a new click, that call counts again, and the bidder has no idea it's the same customer. Raising the duration threshold helps with short hang-ups and redials, but it won't catch repeat callers. To check for those, compare your call conversions against booked jobs in the CRM. If the repeat calls rarely turn into bookings, they're inflating the signal.

Repeat callers are a real gap, but how much they inflate the signal comes down to call volume relative to bookings, so that's the first thing to check. If most of your callers are new customers who ring once, the repeat problem is small and the duration setting does most of the work. If a big share of your calls come from people who already have a technician booked, the bidder is learning from noise.

Agreed on the volume check. I'd add that repeat callers can hide inside the call history itself. Google's call reporting shows the caller's area code and whether the call was answered, but it won't tell you it's the same number ringing back. If your call tracking or CRM logs the phone number, matching it against last month's calls will show how many numbers account for more than one conversion. It's cheap to pull and it gives you something you can actually act on.