Identity Graph Activation Across CTV and Email

How Does an Email Audience Reach CTV?

The buyer starts with a CRM audience collected and approved for the intended campaign. Email addresses are normalized and transformed for the designated onboarding path. An identity graph resolves eligible keys to device, CTV, or household identifiers accepted by the destination. The MAID feed and CTV ACR data represent distinct identity and exposure layers; the contract and join recipe should say which one participates at each step.

The result is an activation audience with lineage back to the approved source cohort. It is not proof that every linked identifier belongs to the same individual. Some paths operate at device level, others at household level, and some rely on modeled relationships. Preserve relationship type and confidence so activation and measurement teams do not silently treat household reach as person-level certainty.

Consent, Suppression, and Destination Governance

The CRM audience's permission does not automatically authorize every downstream destination or measurement use. Record the source, notice, intended use, destination, retention, and suppression behavior for each path. If a record becomes ineligible, the workflow should identify which onboarding files, platform audiences, clean-room tables, and reports require action.

Raw and hashed data are both licensed in the broader GSDSI catalog under written terms. For this email-to-CTV workflow, hashing is a join and transport control, not a claim that every GSDSI product is hashed-only. Buyers should review the privacy compliance hub, destination contracts, and the NAI code and guidance when designing advertising activation.

Control Household Expansion and Match Uncertainty

CTV commonly operates at a household or device level while email begins closer to an account or person. The graph may therefore expand one seed record to several eligible endpoints. Buyers should cap or segment expansion, retain relationship labels, exclude ambiguous joins where the use requires precision, and compare outcomes with and without broader household links.

A useful match report separates input validity, eligible resolution, destination acceptance, household expansion, suppression, and unmatched records. One blended match rate hides where loss or overreach occurred. The identity-graph diligence guide provides questions for reviewing those stages.

The Match Rate That Matters Is the One at the Destination

Activation projects are scoped on a graph's match rate and judged on what actually reaches a platform, and those two numbers are rarely close. Every hop loses records: the graph resolves some share of your file, the destination platform matches some share of what it receives, and platform minimums and policy filters remove more before anything becomes addressable. A strong first number can still end in an audience too small to run.

Model the whole chain before committing, and ask each platform for its own expected match rate on the identifier type you will send rather than assuming the graph's figure carries through. Then test with a real audience at small scale and measure what arrives at the far end. The number that determines whether the program works is addressable records at the destination, and it is the one least often quoted in a proposal.

Suppression Has to Travel With the Audience

Audiences are built once and pushed repeatedly, which makes suppression an ongoing obligation rather than a step in the build. A record suppressed after an audience is created remains in every copy already sitting in a destination platform unless something removes it there, and platforms differ in how they handle updates: some replace an audience wholesale, others merge additions without honoring deletions.

Establish per destination how removals propagate, how long they take, and whether a refresh replaces or merges. Then schedule refreshes at a cadence you can defend rather than when somebody remembers, and keep a record of which audience version went where and when. That record is what makes it possible to answer whether a suppression was honored, a question that arrives without warning and cannot be answered retrospectively without it.

Keep one more thing in the record: the identifier type sent to each destination and the date it was sent. When a platform changes what it accepts, that log is what tells you which audiences need rebuilding and which are unaffected, without re-deriving it from memory.

Where an audience underperforms, check the delivered size against the expected size before questioning the creative, since a shortfall at the destination explains far more disappointing activations than messaging does.

Pilot Activation and Measurement Together

A pilot should use a permissioned seed, a named CTV destination, agreed suppression handling, and a report that reconciles the audience from CRM intake through platform acceptance. Re-run the same seed during the evaluation to observe stability and explain meaningful changes. Do not substitute a vendor-selected demo audience for the buyer's actual data conditions.

Plan measurement before launch. Decide whether the outcome is delivery, reach, site activity, store visits, or another approved event, and identify the resolution level used in reporting. Cross-channel measurement and audience targeting cover the solution paths. The MRC standards library provides useful measurement context for buyer review.