Should you send leads, qualified leads or sales as conversions to Google Ads and Meta? A framework

There's no free lunch here, so let me lay out the trade-off as a framework instead of a slogan.

Every stage you could send has two properties, and they pull in opposite directions. Signal quality: how well does this stage separate a real buyer from a form filler? And volume: how many of these do you get per month, and how fast? The ad platform needs both. Quality tells it who to find. Volume gives it enough examples to learn from.

The lead: high volume, low quality. Bidding learns fast, but it learns to find people who fill in forms. Junk and robots included.

The qualified lead: medium volume, much better quality. The catch is that somebody on your team has to mark leads honestly, and "qualified" has to mean the same thing every time, or the signal gets noisy again.

The booked call or appointment: a good middle. Bots rarely book, so it filters out most fakes, and there's still decent volume.

The showed stage: better quality again. A person who walked in or joined the call is a real prospect.

The sale or paid invoice: the truest signal, and the rarest. It arrives days or weeks later, and there may not be enough of them to learn from.

Notice there's no winner. There's a ladder, and the right rung depends on your volume. If you get a few hundred leads a month, the sale might be enough to optimise on. If you get a few dozen, booked or showed is usually the sweet spot. If you get a handful, you probably need more volume before you can optimise on anything deep.

And the trick that breaks the either/or: don't pick just one. Send several stages as separate events, so booked, showed, won and paid each have their own line. You optimise on one and watch the others. On Google Ads, that means booked and showed go to their own conversion action, not into "Purchase", so a booking never gets counted as a sale.

A made-up funnel makes it concrete. 400 leads a month. 120 book. 70 show up. 25 buy.

Optimise on leads and the campaign sees 400 examples, but 330 of them never became a customer, so it's learning mostly from the wrong people. Optimise on sales and it sees 25 examples, which is precise but thin. Optimise on showed and it sees 70 examples, each one a real prospect who walked in.

In funnels shaped like this, "showed" often gives the best trade: the signal is close to the sale, and there are almost three times as many examples as sales. That's why I'd start there and revisit as volume grows. The numbers are invented, but the shape is common in appointment-based businesses.

The objection I hear most is "but we don't have the volume to optimise on a later stage". It's fair, and the answer isn't to skip the later stages. It's to send them anyway, as secondary, and keep optimising on the earlier stage until the later one has enough history. The data starts flowing now, so the day you do have volume, you also have months of clean history. Starting late means starting from zero.

What did you pick? And did the ads get better, or just fewer?