What AI Media Tools Miss: the Opportunity That Creates for Local Advertisers

In 2025, Futuri Media ran an experiment. They fed media planning prompts to eight AI systems (GPT-4o, Gemini, Grok, Claude, Perplexity, and others) across 20,000 generated plans covering 60 business sectors, budgets ranging from $25,000 to $10 million, at both national and local market levels. They wanted to know: what does AI tell advertisers to do with their money?

The answer should concern every business owner in the Treasure Valley.

Radio appeared in just 3% of plans. Google's Gemini and Anthropic's Claude excluded radio in 100% of the plans they generated. Digital video dominated, appearing in 80–99% of plans. A $150,000 campaign for a home improvement brand in Phoenix, a driving market not unlike Boise, came back with zero allocation to radio or television.

This isn't a quirk or an edge case. It's a structural feature of how AI planning tools work. And it's steering advertisers toward a strategy that decades of marketing effectiveness research says is quietly destroying brands.

Why AI Can't See Radio

AI media planning tools learn from data. The problem is the data they have access to. Google's ecosystem has thousands of documented, indexed, measurable case studies: clicks, conversions, cost-per-acquisition, ROAS. The feedback loop is clean and abundant.

Radio's effectiveness data exists, but it's sparse, inconsistently formatted, and rarely structured in a way AI systems can parse. There's no pixel on a radio ad. There's no click-through rate. The attribution requires separate tools and deliberate methodology, none of which feeds back into the AI's training data automatically.

The result, as Futuri's research describes it, is an "algorithmic snowball": AI directs budgets to digital, more digital data accumulates, that data reinforces the AI's digital bias, repeat. Futuri projects that if AI adoption in media planning reaches 60%, which it's on track to do, radio could lose $6 billion in annual ad revenue by 2028. Not because radio stopped working. Because it became invisible to the tools making the decisions.

What AI Actually Optimizes For

Google's Performance Max and Meta's Advantage+, the two dominant AI-driven ad platforms, are engineered around a single objective: conversion signals. Purchases, form fills, calls, leads. The algorithm is extraordinarily good at finding people who are likely to convert right now.

That is not the same thing as building a brand.

One paid media agency founder analyzed roughly 40 Performance Max campaigns across dozens of companies and found only one showed genuine profitability. The pattern in the others: PMax absorbs branded search clicks, conversions that were already going to happen because the consumer already knew the brand, then claims credit for them. Meta Advantage+ does the same thing, over-indexing on retargeting audiences rather than discovering new prospects.

These platforms are not growing your customer base. They are efficiently harvesting the demand that already exists, demand that someone, somewhere, at some point created through brand-building. When there's no brand-building in the plan, these tools are consuming seed corn.

In a single year, the industry crossed a threshold. Brand investment dropped below one-third of total ad spend. The recommended ratio, established by Les Binet and Peter Field through the world's largest database of marketing effectiveness case studies, is 60% brand building, 40% activation. The industry is now running the inverse.

The Death Spiral

At the IPA Effectiveness Conference in October 2025, Les Binet, the researcher behind most of what the industry knows about long-term marketing effectiveness, named what's happening directly.

"Budget is eight times more important than ROI," he said. "Yet 65% of marketers say ROI is most important and only 35% say budget."

The data behind that statement is sobering. Since COVID, marketing ROI has risen 4%. Net profit generated by marketing has fallen 11%. The industry is getting more efficient at doing less. It is optimizing itself toward irrelevance.

Binet called it a death spiral: optimize for ROI, cut reach and scale, weaken mental availability, make brand-building harder, force more reliance on performance tactics to make up the gap, repeat. The deeper you go, the harder it is to get out.

The mechanism is well understood. Byron Sharp and the Ehrenberg-Bass Institute have documented it for decades: brands grow by building mental availability, being easy to think of at the moment someone needs what you sell. Mental availability is built through broad reach, consistent exposure, and memorable creative. It requires reaching people who are not currently in-market, the light buyers and non-buyers who represent the majority of future growth.

Narrow targeting, which is the core feature of every AI performance tool, systematically fails to do this. It finds the people most likely to convert today and ignores everyone else. It is fishing in a pond, not a river.

What Happens to Brands That Go Dark

The Ehrenberg-Bass Institute tracked 41 brands across the alcohol category over nearly two decades. Brands that stopped advertising saw:

A Harvard Business Review analysis of three separate recessions found that 80% of companies that cut marketing costs had not regained pre-recession sales and profits within three years of the recession ending. Econometric consultancy Data2Decisions found that negative consequences of going dark are four times more severe in the long run, with a significant delay before the full damage shows up. By the time the numbers confirm the problem, the brand equity is already gone.

This is why a tough economy is exactly the wrong time to abandon brand investment. Analytic Partners analyzed hundreds of billions in marketing spend across 750+ brands in 45 countries and found that over 60% of brands that increased marketing investment during a recession saw improved ROI and incremental sales. Brands that cut back saw an 18% contraction in incremental sales, two-thirds of which came directly from reduced investment, not from the economy itself.

The Specific Case for Radio in This Equation

Radio isn't valuable here simply because it's broad-reach. It's valuable because it's the highest-reach, lowest-cost-per-point medium available for building mental availability in a local market, which is exactly what the effectiveness research says brand-building requires.

Peter Field's 2024 analysis of the IPA Databank found that brands using AM/FM radio experience +13% greater mental availability, +42% greater profit, and +58% lift in brand trust compared to non-radio brands. These aren't theoretical benefits. They're measured outcomes from the same methodology that underpins all of Binet and Field's effectiveness research.

In the Treasure Valley, a market built around commuting, radio holds 86% of in-car ad-supported audio time. It does something digital targeting cannot replicate: it reaches adults during their daily routines, before they're in-market, while their purchase intent is still forming. That's precisely where mental availability is built.

Switching from a performance-only approach to a balanced brand-plus-performance strategy improves total revenue ROI by an average of 90%, according to WARC's 2025 Multiplier Effect research. Upper-funnel brand tactics are 60% more effective over the long term than lower-funnel tactics and only 25% less effective in the short term, meaning the trade-off is far more favorable than most AI-optimized plans assume.

What to Do About It

The AI tools aren't going away. Performance Max and Advantage+ are legitimate tools for capturing in-market demand. Used correctly, within a plan that also builds demand, they work. The problem is using them as the plan rather than part of one.

Binet's prescription at the IPA conference was direct: "Go big or go home." Statistically significant sales uplift requires 30–60 million exposures. Major, durable brand results require 200 million or more. Those are numbers you cannot reach through precision targeting. They require scale, the kind of scale that broad-reach audio in a local market actually delivers.

If your current media plan was generated by an AI tool, or by an agency optimizing primarily for trackable conversions, it is almost certainly underinvesting in brand. It is optimizing for what is easy to measure rather than what actually grows a business. And in a market where your competitors are doing the same thing, the business that holds its brand investment, or increases it, is the one that will be standing when conditions improve.

The data on this isn't new. What's new is that AI is accelerating the mistake at a scale that makes the consequences harder to escape.

Sources: Futuri Media, "The Algorithmic Snowball," July 2025 (20,000+ AI-generated media plans across 8 LLMs and 60 business sectors). Les Binet and Will Davis, IPA Effectiveness Conference, October 2025. WARC, "Performance Budgets Rise at Expense of Brand," November 2024. WARC / Analytic Partners / System1, "The Multiplier Effect," 2025. Analytic Partners ROI Genome Intelligence Report, "The Rules of Recession-Proofing." Ehrenberg-Bass Institute for Marketing Science, brand investment withdrawal study (41 brands, 18-year longitudinal). Peter Field / IPA Databank analysis for Westwood One, January 2024. Edison Research Share of Ear®, Q3–Q4 2024. Friday Agency Ireland, "Performance Marketers Are Sleepwalking Into an AI Black Box," 2025.