AI in ads: where it helps, and why we never set and forget.

Never set and forget

AI tools promise ad accounts that run themselves. We don’t recommend it, for Google or for Meta. Every ad account should have a person checking search terms, leads and spend every week, and making the calls.

Ad platforms optimise for whatever you tell them is a result. If tracking counts the wrong things, or nobody checks lead quality, the automation will happily spend more to get more of the wrong thing. It will not tell you.

Google Ads: don’t hand Google’s AI the wheel

Google’s built-in automation is, frankly, not good enough for most local businesses. Auto-applied recommendations raise budgets and loosen keywords. Broad match drifts into searches you would never pay for. Performance Max spends across placements you can’t see properly. All of it is designed to grow spend on Google first.

If you want AI in your Google Ads, we recommend not using Google’s native options to run the account. Use AI for the checks and the drafts, and keep a human layer with long-standing Google Ads expertise on the decisions: which searches, which areas, what a lead is worth and when to say no.

In practice we switch off auto-applied recommendations, avoid Performance Max at launch, start with tight control: exact and phrase match keywords, strong negative keyword lists and capped bids, and only consider automated bidding once conversion data is clean, with a person checking it every week.

Facebook and Instagram: the AI finds the audience, people make it work

Meta’s AI has changed things. It is now good at finding the right people, so the old approach of stacking interests and lookalikes matters far less. Broad targeting in the right location usually wins.

That makes the creative and the analytics the whole game. Which hook stops the scroll, which offer gets the form filled, which leads turn into jobs. Those are human decisions, and they need a strong human layer: fresh creative every month, lead form questions that filter, and someone reading the results against real sales.

What still needs a person

• Deciding what counts as a lead, and checking that tracking only counts that.
• Reading search terms and blocking waste every week.
• Choosing services, areas and budgets, and saying no to spend that won’t pay back.
• Writing and refreshing creative, and judging which ads actually bring jobs.
• Checking lead quality with the business owner.
• Catching broken tracking, disapproved ads and platform changes before they cost money.

Where AI does help

Use AI to draft the weekly search-term review, flag wasted searches, spot unusual spend and write the first draft of reports. That gives whoever runs the account more time for the decisions above. Never let it change budgets or pause campaigns on its own.

Already running your ads with AI tools? Co-pilot adds the human layer: a senior strategist reviews your accounts monthly or fortnightly and tells you what to change.

Real results come from people plus tools

K2 Car Removal went from 126 to 266 Google Ads leads a quarter, July to September 2025 against 2026, with 40% of clicks turning into leads, up from 26%. That came from fixing tracking, cleaning up searches and managing the budget by hand, not from switching Google’s automation on.