First-party data and cookieless audiences: the new targeting stack
Published August 20, 2026
For a decade, targeting meant renting someone else's data: third-party segments that followed users around the web. That era is over — browsers deprecated third-party cookies, privacy regulation tightened, and identifiers fragmented. What replaces it isn't a single solution but a stack, and the strongest layer is the one you build yourself.
Want to put this into practice?
SeatNMD gives you DV360 access, supply and flexible operation.
Let's talk →The four layers that survive
| Layer | What it is | Who owns it |
|---|---|---|
| First-party audiences | Your site visitors, app users, CRM lists, converters | You |
| Platform audiences | Intent and in-market signals from the platform's own graph (e.g. Google audiences) | The platform |
| Lookalikes / similar | Modeled audiences seeded from your first-party data | You + the platform |
| Contextual | Page content, category and keyword relevance — no user identity involved | Nobody needs to |
Notice what's missing: rented third-party segments. They still exist, they're still sold, and they're the layer degrading fastest.
Building the asset: site tag first
Everything starts with tagging your site. In the Google stack, a Floodlight tag on every page turns ordinary traffic into a growing audience: all visitors, visitors to specific sections, people who started a form but didn't finish, converters to exclude from prospecting. Each of those becomes a targetable segment inside DV360 — and, critically, one that keeps compounding while you do nothing.
The timing matters more than the sophistication: an audience takes months to reach useful size and cannot be bought retroactively. Tagging the site the week before a campaign launches means launching with nothing. Tagging it today means having something real by the time budget shows up.
Turning first-party data into reach
Your own audience is precise but finite. Two ways to scale it without going back to rented data:
- Lookalikes: seed a model with your best converters and let the platform find similar users at open-web scale. Quality of output tracks quality of seed — a lookalike from 50 random leads is noise; from 500 qualified customers, it's your best prospecting source.
- Contextual: place ads where the content signals intent, no identity required. It's the layer that made a comeback for exactly this reason, and it's the native logic of channels like DOOH.
Sequencing beats blasting
Once audiences are structured, the campaign changes shape: prospecting reaches new users with contextual and lookalike targeting, retargeting speaks to site visitors with a different message, converters get excluded so you stop paying to reach customers you already have. Layering these deliberately is the craft covered in inventory and targeting in DV360, and where a lot of wasted budget hides.
Do it right, legally
First-party data is a privacy responsibility, not a loophole: disclose what you collect and why, honor deletion and opt-out requests, and keep your privacy notice current with the technologies you actually run. Advertisers that treat this seriously end up with more durable data than the ones that don't.
The takeaway
The advertisers who'll target well in three years are the ones tagging their sites and structuring their audiences now — while it looks like it isn't urgent. If you're setting up a DV360 seat, the tagging conversation should happen on day one, not before the first big campaign.
Newsletter
Stay ahead of the market.
Programmatic moves fast — DV360 updates, CTV and DOOH trends, supply opportunities. Get the essentials in your inbox, no noise.