Target CTV from your own customer segments and measure the reach you actually added above linear, rather than buying impressions on faith
Connected TV is sold like digital but measured like television, which is to say barely. Brands buy CTV impressions against platform-defined audiences, get back impression counts and a frequency figure, and have almost no way to answer the two questions that matter: did this reach people the brand actually wanted, and did it add reach above what linear TV was already delivering. The default is to trust the platform's audience and the platform's reach claim, both of which the brand cannot inspect.
The information to do better exists in the CRM. The brand knows its high-value segments and can match them to CTV platforms for targeted reach, the same first-party activation move used for display and social. What CTV adds, and what makes it a distinct recipe rather than just another destination, is the measurement problem: CTV exposure happens on the television, the response happens on a phone or laptop, and proving incremental reach above linear requires both a cross-device link and a counterfactual.
This recipe targets CTV from owned segments and measures the reach genuinely added, rather than accepting impression counts as evidence.
CTV campaigns targeted from first-party segments, with an honest read on incremental reach above linear. The headline metric is incremental reach: the audience CTV reached that linear did not, measured against a baseline rather than asserted from impression logs. Realistic framing: incremental reach is meaningful for brands whose linear buy already saturates a demographic, where CTV's value is reaching the cord-cutters and light TV viewers linear misses; for a brand with little or no linear presence, the "incremental above linear" framing matters less and the recipe collapses toward standard first-party activation. The size of the incremental reach depends on the linear-CTV audience overlap, which is exactly what the measurement is there to reveal.
The second outcome is cross-device credit: knowing that a CTV exposure preceded a response on another device, so the channel gets the attribution it earned rather than none.
A path from the warehouse to the CTV platforms (the major CTV and streaming ad platforms, or a CTV-capable DSP) for the hashed segment match, plus a measurement capability that can join CTV exposure to cross-device response. Composable stacks suit this because the exposure-to-response join lives in the warehouse alongside the identity graph. The capability that decides whether the recipe delivers is measurement design: targeting CTV from CRM is the easy half, and a brand can do it and still learn nothing if it cannot connect the household-level exposure to the cross-device response and compare against a counterfactual.
Compare the tools on Martech Stack Builder
Marketing ops owns the segment definitions, the CTV platform mapping, and framing the linear-versus-CTV reach question the campaign is meant to answer. Data engineering builds the hashed match export and the exposure-to-response join across devices. Analytics is critical, because incremental reach and cross-device attribution are the entire point and both are genuinely hard to measure honestly. Legal covers the basis for sharing hashed segments with CTV platforms, which is the same shape as other activation but worth confirming per platform. The recipe takes months because the measurement is the hard part and needs a test design that runs long enough to read. The targeting can be live quickly; the credible incremental-reach answer takes a measurement cycle.
Work in this order. Measurement is designed before the first impression, or impressions end up counted as reach.
Consent-scoped signal collection covers steps 1 and 2, anonymous-to-known stitching steps 3 and 4, cross-device identity resolution step 5, and incrementality testing as default measurement step 6.
The first failure is counting impressions as reach. CTV platforms report impressions and frequency readily, and it is tempting to treat a large impression count as a large reach addition. Without a baseline, impressions say nothing about incremental reach; a campaign can deliver millions of impressions entirely to households linear already saturated and add almost no reach. The holdout is what separates delivery from incrementality.
The second is the household-individual mismatch. CTV resolves to households, so a segment built at the individual level is matched at the household level, and both the targeting precision and the measurement inherit that coarseness. A high-value individual in a household of five is reached alongside four others, which is fine if understood and misleading if the brand believes it bought individual-level precision.
The third is the broken cross-device link. If the CTV exposure cannot be connected to the cross-device response, the channel shows no attributed conversions and gets cut for underperforming, when the real problem is that the measurement could not see the response it drove. The cross-device resolution is a prerequisite for the channel getting fair credit, which is why it is in the recipe rather than assumed.
The Workshop works out with your team which of these matter for your stack right now, and what to do first: a 90-minute session with the people who own the decision.
The household-level match and what it means for precision, the cross-device link that lets CTV earn credit, and the incremental-reach test against linear: those are the decisions that turn a CTV buy on faith into one you can defend.
Did this recipe match your situation?Anonymous response. Sign up to leave a longer note tied to your account.
Activate the target account list with one identity resolution and one exclusion list across paid display, paid social, direct mail, and BDR outreach, so the same account stops getting contradictory treatment
Aggregate the signals from every contact in the account into one account score, because the buying committee is buying together and scoring leads in isolation misses the picture that matters
Lower acquisition cost by feeding the bidding algorithms better conversions, deduplicated, consent-valid, and weighted by value, rather than just more of them