CTV Targeting Is Breaking Trust — Here's the Fix

AdExchanger’s Victoria McNally recently described something that should make every programmatic buyer uncomfortable: she saw a CTV ad for a pregnancy test, inferred correctly that the platform had drawn some probabilistic conclusion about her body, and responded by purchasing from a competitor — out of spite.

That is not a targeting miss. That is a targeting own goal.

When Identity Resolution Becomes Identity Assumption

The mechanics here aren’t mysterious. A probabilistic identity graph — stitching together device signals, browsing behaviour, location pings, and purchase data — flagged McNally as a likely prospect for pregnancy-related products. It was probably right about the demographic cluster. It was catastrophically wrong about the individual, and the penalty wasn’t just a wasted impression: it was active brand damage.

This is the core tension in cookieless identity resolution that most vendors are not being honest about. Probabilistic matching improves reach but degrades precision in exactly the categories where precision matters most: health, finance, relationships. The trade-off is manageable when you’re selling running shoes. It becomes genuinely harmful when you’re in sensitive verticals — and in Southeast Asia, where cross-device household sharing is significantly more common than in Western markets, the false-positive rate on household-level graphs is structurally higher than the benchmarks these platforms were built on.

The Frequency Cap Problem Nobody Wants to Fix

McNally’s experience wasn’t a one-off impression. The ad repeated. That’s a frequency management failure layered on top of an identity failure — and the two compound each other in ways that accelerate negative brand association.

The uncomfortable reality is that most DSPs still manage frequency at the campaign or line-item level, not the consumer sentiment level. There is no signal that says “this user has now seen this ad four times and their engagement pattern suggests irritation rather than consideration.” Clean room environments theoretically allow for richer signal integration — you could in principle model churn risk in your CTV audience exposure — but almost no brand in Southeast Asia is operationalising clean rooms at that level of sophistication yet.

The practical fix available today is blunter but still meaningful: build category exclusion logic into your identity-based audience segments, not just your contextual targeting. If your identity graph can’t tell you someone is not in a sensitive life-stage category with reasonable confidence, suppress the category entirely for that household cluster. The lost reach is recoverable. The lost brand trust often isn’t.

---

What Programmatic DOOH Gets Right (That CTV Should Copy)

Interestingly, the VIOOH–Grupo IMU partnership announced last week in Mexico points toward an infrastructure model that sidesteps some of these identity problems by design. Programmatic DOOH operates on contextual and temporal signals — screen location, time of day, audience composition indices — rather than individual identity resolution. You’re buying the right environment for your message, not trying to follow a specific person across screens.

That’s not a fully satisfying answer for performance advertisers who need attribution. But for brand categories where a mistargeting event carries real reputational cost — pharmaceuticals, financial products, anything touching family or health — the DOOH model’s deliberate distance from individual identity is a feature, not a limitation. As programmatic DOOH infrastructure expands across Southeast Asia (inventory in markets like Thailand, Indonesia, and the Philippines has grown substantially over the past 18 months), media planners should be treating it as a complement to CTV reach, not just an outdoor buy with a digital wrapper.

The strategic question is where in the funnel you can afford to be contextually imprecise versus where individual-level targeting precision is genuinely load-bearing for your business outcome. Most brands have not mapped this clearly enough.

Building Identity Infrastructure That Earns the Impression

The path forward for CTV targeting isn’t abandoning identity resolution — it’s being more honest about what probabilistic graphs actually know versus what they’re guessing. Three implementation priorities worth pressure-testing now:

First, audit your identity graph’s confidence intervals by vertical. Most DSP reporting doesn’t surface this, but your identity resolution vendor should be able to tell you the match rate and confidence distribution for health and family-stage segments specifically. If they can’t, that’s a vendor conversation you need to have.

Second, build suppression logic as a first-class campaign element, not an afterthought. The question “who should we exclude?” deserves the same brief-writing rigour as “who should we target?” In markets like Singapore and Malaysia, where multilingual households are common and device sharing across generations is standard, exclusion logic is often more valuable than inclusion logic.

Third, treat frequency as a brand safety variable. Capping at three or four exposures per week sounds reasonable in a spreadsheet. It doesn’t account for the emotional register of the creative, the sensitivity of the category, or the platform context. CTV in particular — consumed in a lean-back, often shared-screen environment — amplifies the intimacy of a targeting assumption gone wrong.

The infrastructure is shifting fast. The playbooks for doing this well are still being written. But the cost of writing them badly is showing up in real purchase decisions, and it’s measurable.

---

The brands that will win the next phase of programmatic CTV aren’t the ones with the biggest identity graphs — they’re the ones with the most disciplined thinking about what those graphs should and shouldn’t be used for. What’s the one category in your media plan where you’d be most uncomfortable if a consumer correctly inferred what signal you used to target them?

At grzzly, we work with marketing and media teams across Southeast Asia to build identity and targeting frameworks that perform without the liability — from clean room strategy to audience suppression architecture. If your CTV or programmatic setup has grown faster than your governance around it, we should talk. Let’s talk