If you have run ads for more than a few years, you learned a craft that is being retired. Keyword lists, tightly segmented ad groups, manual bids, carefully built audiences. Both major platforms now automate most of that.
Google’s AI Max for Search campaigns uses broad match and keywordless matching to find queries your keyword list would have missed, and generates tailored creative from your existing assets and landing pages. Google reports that advertisers activating it typically see 14% more conversions or conversion value at a similar cost per acquisition, based on its own internal 2025 data for non-retail advertisers.
Meta’s Advantage+ campaigns do the equivalent on the social side, removing manual audience construction in favour of real-time optimisation.
Treat a vendor’s own performance figure as a directional claim, not a forecast for your account. The structural point stands regardless: the targeting decisions you used to make are increasingly made for you.
The inputs did not disappear. They moved.
Automated targeting optimises toward a goal you define, using signals you supply, sending traffic to a page you built, for an offer you wrote. Every one of those is still yours. The system is a very fast optimiser pointed at whatever you tell it to value.
Which means the failure mode has changed. It used to be bad targeting. Now it is efficiently buying the wrong thing.
Control one: what counts as a conversion
This is now the single most consequential setting in your account. If every form fill counts equally, the system will find you the cheapest form fills, and cheap form fills are usually not customers.
Define the conversion as close to revenue as your setup allows. A booked appointment beats a form submission. A qualified call beats any click. If you can pass back which leads became jobs, do that, and the optimiser starts working on your actual business rather than a proxy for it.
Control two: the landing page
Automated systems will send a wider and less predictable range of people to your page. That raises the cost of a page that only makes sense to someone who already knows you.
The page has to answer, quickly and without jargon: what this is, who it is for, what it costs or how pricing works, and what happens if they contact you. If it does not, the platform will keep buying clicks and you will keep concluding that ads do not work.
Control three: the creative and the offer
When targeting is commoditised, the message is the variable. Two competitors can reach the same person with the same budget on the same platform; the difference is what the ad says and whether the offer is worth acting on.
This is the part most small advertisers underfund. It is also the only part a competitor cannot copy by changing a setting.
Control four: the exclusions
Broader matching means more irrelevant traffic reaching you before the system learns. Watch your search terms report in the first weeks, add negatives for the queries that are clearly not your business, and exclude the audiences and locations you cannot serve.
This is unglamorous and it is where a meaningful share of wasted spend lives.
Control five: how long you leave it alone
Automated bidding needs conversion volume to learn. Changing budgets and targets every few days keeps a campaign permanently in its learning phase, which is the most expensive place for it to live.
Set a review cadence and hold it. Weekly for spend and obvious waste, monthly for structural decisions.
What good looks like now
A clear conversion definition tied to real work. A page that answers the obvious questions. An offer worth responding to. A short exclusion list maintained weekly. And enough patience to let the system learn.
That is a shorter list than the old craft. It is not an easier one, because every item on it is about your business rather than the ad platform.
If your ads are producing clicks and not booked work, the problem is usually one of these five and not the campaign settings. Schedule a call with us and we will look at the whole path from ad to booked job.
