The Circular Logic of AI-Optimized Audiences

Is your AI finding new customers, or shining a light on purchase-ready consumers?

Following up on this week's theme of ways to avoid marketing analytics misfires and be successful in 2026. The first part touched on issues with AI-driven ad optimization, and today will discuss potential concerns with CPA optimization.

Everybody strives for the lowest CPA (cost per acquisition) they can get.  That makes sense. But during my holiday catch-up I came across a post claiming that “AI-driven targeting” had cut CPA in half for a business.  The CMO was thrilled.  But should they have been?  Not if you break down the buying audience.

The Misfire: AI-driven targeting shifted 90% of the budget toward branded search and heavy retargeting.  Nothing revolutionary there – those techniques have been common for years. But they are hardly focused on finding an undiscovered audience, which is really what you are hoping when you are measuring CPA improvements. That approach is simply finding those who have already provided intent signals.  Where is the optimization?

The Reality: The AI tool discovered that the easiest people to convert are the ones that already know, like and trust your brand.  Shocking. That’s part of the challenge with today’s iteration of AI models – they only know what has already been discovered.  So this solution wasn’t growing the business, it is another example of a tool reaching already high-intent consumers and claiming credit for their purchases.

The Lesson: Efficiency is easy when you target consumers who were going to buy anyway.  We don’t need AI for that – branded search and re-targeting have been around for a minute.  Real growth requires incremental lift, and reduced CPA isn’t a proxy for that.  Instead, focus efforts on finding consumers that need your product or service, but don’t know it themselves or just haven’t shown intent, essentially pulling them forward in time.  And develop more than one metric that measures success of your efforts - like CPA for low-intent consumers vs. CPA for high-intent consumers. Could be eye-opening. Also think about ways to expand consumption or encourage cross-sell/up-sell, and reach your existing customer base with relevant messaging and offers.  If AI can help in that effort, great, but those areas offer fertile ground whatever the tools you use.

What do you think - have you had similar or different experience with tracking CPA? Do you agree or disagree? Want to chat? Find a convenient time here.