What should Google and Meta optimize your ads for?

Google and Meta should optimize your ads for the guests who paid a check, not for the taps, views and clicks that come before one. Left on their defaults, they chase whatever they can see, and in most restaurant accounts that is an action on a phone.

The usual advice for these ads is sound: narrow the audience, bring back people who looked at the menu, keep the offer local, use good pictures. All of it works inside one decision that rarely gets made on purpose, which is what the platform has been told counts as success.

What does the platform think a good result is?

Every automated bid and every widening of the audience runs on conversions. Tell Google that a tap on Directions is a conversion and it will look for more people who tap Directions. Tell Meta that a menu view is a result and it will look for people who view menus. Some of those people come in and spend. Many were coming anyway, or never arrive, and the system can't tell which, because nobody has told it what happened at the table. We walk through what each platform counts in what Google and Meta count as a conversion for a restaurant.

What does that look like in a month's report?

Here is a made-up month. A restaurant spends $1,500 on Google. The report shows 340 conversions: 190 taps on Directions, 80 calls and 70 menu views. That works out to about $4.40 a conversion, which looks efficient.

Over the same month, 41 paid checks can be traced to those ads. That is about $37 for each check. Neither number is wrong. They answer different questions, and the platform is learning from the first one.

The risk is that the cheaper number wins the argument. A $4 conversion is easy to defend in a meeting and a $37 check is not, so the budget drifts toward whatever is cheapest to produce, whether or not it fills tables.

Do targeting, creative and bidding still matter?

They do. A tight local audience, a photograph that makes someone hungry and a budget matched to the week all do real work. The risk is the order. Each of them refines the search for whoever the signal describes, so if the signal is a tap, you get a better-aimed search for tappers. Get the signal right first, then tune.

The same goes for retargeting. Showing the ad again to people who looked at the menu is reasonable. Counting every one of them who later books as proof the ad worked is how a regular who was always coming on Saturday ends up credited to a campaign. By default Meta credits an ad with a result that happens within a day of someone merely seeing it.

What can you send back instead?

Both Google and Meta can take in sales that happened away from the website and treat them as conversions, with their value. It is rarely set up in a restaurant, because it means connecting the ad accounts to the register and matching one to the other.

Before you change anything, list each campaign and write down the one result it is set to get. Then ask how many of those results a person could produce without ever intending to eat.

If you do connect the register, send only the sales you can tie to an ad. An estimate fed back as though it were a check teaches the system the wrong lesson, and it is harder to undo than to avoid.

Where restauWant comes in

The restauWant app sends the checks back to Google and Meta so the campaigns learn from dining revenue rather than clicks. It sends verified matches only, with the amount. Checks start going to both as soon as the register and the ad accounts are connected, and you review each batch before it goes. It can be turned off per campaign and per location. The Optimization page has the detail, and the demo is open if you would like to see it against real numbers.

Why the platforms' own numbers rarely match the register is a separate question, and we take it apart in why Google and Meta ROAS don't match your register.

Next step: Open the conversions list in Google Ads and the result setting on each Meta campaign, and write next to every one whether it is a tap, a view or a sale.