AI can write, target and optimise faster than any human team. It also confidently takes credit for sales it didn't make. A sober look at where AI actually earns its keep.
What AI is genuinely good at
Drafting variations of ad copy, summarising customer reviews, clustering keywords and spotting anomalies in large datasets. These save real hours and we use them daily.
Platform bidding algorithms are also genuinely strong — when they're fed clean conversion data. That's the catch.
Where it quietly costs you
Every ad platform's AI is optimising for its own reported conversions. If your tracking counts a page view as a 'lead', the AI will happily find you millions of page views.
Attribution is the other trap. Meta, Google and your email tool will each claim the same sale. Add their dashboards together and you've sold more than you made.
How to use AI without being used by it
Fix measurement first: define a real conversion (a qualified lead or a sale) and feed only that back to the platforms.
Reconcile platform numbers against your CRM or accounting every month. The gap between 'reported' and 'real' is the most important number in your marketing.