Demographic fit tells you who could buy; behavior tells you who is ready to act right now. When we were inside a digital health company, we spent $20,000 on a partner channel whose audience almost perfectly matched our core customer’s age profile: people 55 and older. It made zero sales, even after a relaunch.
That lesson matters more now that Google is folding its Display campaigns into AI-driven Demand Gen. There, the intent and context signals you give the system count for more than the placements you hand-pick.
Why can a perfect demographic match produce zero sales?
Because demographic fit doesn’t tell you whether someone is in a buying moment, and the audience we paid to reach wasn’t. The company’s product served people managing type 2 diabetes, and its core user was 55 or older. When a large partner channel came up whose audience matched that age profile almost exactly, it felt like a safe bet. The overlap was close to perfect, the targeting options were precise, and the logic looked airtight.
The channel cost us $20,000 and produced exactly zero sales. Not a weak ROAS. Zero. It failed twice.
The partner’s own team was puzzled enough by the results to offer a creative workshop. Their experts reviewed the ads, made specific recommendations, and helped relaunch the campaign. Still nothing.
The tracking was clean. The placements were live. The audience was real. But the people clicking through weren’t in a buying mindset; they were in a browsing mindset. They’d come to read articles and catch up on news. A health product ad, however well matched to their age, was an interruption, not a solution. The demographic fit was real. The behavioral fit wasn’t.
Many media plans skip that distinction. Demographic targeting asks, “Who is this person?” Behavioral targeting asks, “What is this person doing right now, and does that behavior suggest they’re open to acting?” The second question is harder to answer. It’s also the one we should have asked before spending $20,000.
What does Google’s move from Display to Demand Gen change?
Hand-picking placements matters less, and the signals you give Google’s AI matter more. On May 26, 2026, Google announced that its Display Ads product “has a new home in Demand Gen” and said it expects the transition to be complete by 2027. According to Google Ads Help, a migration tool began a phased rollout in June 2026. The same page lists two later steps: new Display campaigns will be created only in Demand Gen, and remaining eligible campaigns will migrate automatically. The Display Network itself stays. It can now be bought through Demand Gen, a campaign type that also reaches YouTube, Discover, Gmail, and Google Maps (in beta).
Demand Gen still has levers, such as channel controls and contextual targeting. But Google describes its AI as choosing the combination of visuals, messages, and placements for your ads. So your effort shifts to the signals you feed the system: who your buyer is, and what they’re doing just before they act.
For years the dominant logic was simple: find the right person, put the ad in front of them, and the conversion follows. Demographic targeting was the tool, and placement selection was the craft. Our $20,000 test showed the flaw in that logic. You can reach the right person in the wrong moment and get nothing. We read Google’s shift as a push away from placement precision and toward behavioral precision.
How do you build campaigns around behavior instead of demographics?
The instinct to hunt for demographic overlap isn’t wrong; it’s incomplete. Age, income bracket, and job title are useful filters. They narrow the universe. But they tell you nothing about intent or context.
A 58-year-old with type 2 diabetes reading a news article about retirement isn’t the same buyer as the same 58-year-old who just searched “how to lower A1C naturally.” Same person. Completely different moment.
So start from the signals that reveal a moment, and use demographics as a filter on top:
- Search intent. What someone just searched for is a strong sign that they’re trying to solve a problem now. Keyword-targeted Search ads reach that moment directly.
- Context. What someone is reading or watching when the ad appears. Demand Gen supports contextual targeting outside sensitive categories such as health.
- Your own behavioral data. Site visits and past purchases can seed remarketing and lookalike segments.
- Channel mindset. Whether people come to a channel to browse or to act. Our partner channel’s audience came to read, not to buy.
For health products, condition-based audiences are restricted. Google’s personalized advertising policy treats health as a sensitive interest category, so health advertisers can’t use advertiser-curated audiences such as remarketing lists or lookalike segments. Meta began removing health-related detailed targeting options in January 2022, and GDPR Article 9 restricts the processing of health data in the EU. In Demand Gen, Google limits sensitive-category ads to predefined Google audiences and generally to YouTube inventory (About Demand Gen campaigns). So for a health product, find the moment through search keywords, not through audiences built on an inferred condition.
Before you build your next campaign, ask two questions. First: does your audience’s behavior in this channel match the action you’re asking them to take? Second: are you targeting a person, or a moment?
If you can’t answer the second one, you might be about to spend $20,000 to find out the hard way.
Our approach to paid acquisition and conversion is on our Acquisition & Retention page.
Sources: Google Ads & Commerce Blog, “Google Display Ads has a new home in Demand Gen” (May 26, 2026); Google Ads Help, “Google Display Ads campaigns have a new home in Demand Gen” and “About Demand Gen campaigns” (checked October 23, 2026); Google Advertising Policies Help, “Health in personalized advertising” (checked October 23, 2026); Meta for Developers, “Updating Meta’s Detailed Targeting options starting January 19, 2022” (December 8, 2021); GDPR Article 9, “Processing of special categories of personal data” (Regulation (EU) 2016/679; unofficial text via gdpr-info.eu, checked October 23, 2026).
Updated October 23, 2026: we rewrote this post, re-checked its facts, and added sources.
