Programmatic Advertising: Why Aren't All Cookies Worth the Same? ๐ช๐ฒ
An advertiser recently told me: โI don't understand. I'm seeing absolutely enormous drop-off rates between my DMP and my DSP for the audience segments I'm pushing.โ Sound familiar? ๐ช
1. Cookie synchronisation: where audiences disappear ๐
Cookies are specific to the technology that sets or reads them. To target an audience in a DSP, identifiers collected in a DMP often need to be matched to the DSP's identifiers โ and, depending on the buying path, to identifiers recognised by SSPs. Every matching stage can reduce the pool of usable IDs.
So a DMP reporting a million cookies is not promising that a DSP can bid on a million recognisable people. (And cookies, browsers and people are not interchangeable units.)
2. Cookie recency: when was that user last active? โณ
DMPs may count cookies over relatively broad windows, while DSPs can use shorter windows when estimating active cookies. That distinction is sensible: an identifier last seen months ago is less useful for forecasting campaign delivery than one encountered recently.
Recency thresholds vary by platform and configuration, so the right comparison is between equivalent definitions, not simply two dashboard totals.
3. The browser matters: primary versus occasional use ๐
Not all cookies offer the same practical targeting opportunity. One may belong to the browser someone uses throughout the day; another may come from an occasional secondary device or an in-app browser that is rarely seen again.
The first is more likely to be available for effective retargeting. The second can sit in a database without producing much addressable inventory. What matters is whether DSPs and SSPs actually encounter the identifier often enough to buy against it.
| What a DMP may count | What a DSP needs for delivery |
|---|---|
| Cookies collected during a broad lookback window | Identifiers recently active and recognised |
| Cookies from multiple browsers and devices | Identifiers available in relevant bid requests |
| Raw audience-segment membership | Matched, eligible, reachable audience under campaign settings |
4. My practical tip: filter for activity in the last seven days ๐ก
There's a simple way to stress-test the numbers: in your DMP, filter for identifiers that have been active in the last seven days. Compare that figure with the unfiltered segment and the volumes reported by your DSP.
It's not a universal industry standard or a guarantee of impressions, but it is a useful starting point for a more realistic forecast. Delivery will still depend on match rates, browser restrictions, consent, inventory, targeting and bidding conditions.
5. Keep audiences broad โ and think beyond cookies ๐ฏ
Overly narrow segments make these losses worse. Where the technical and legal conditions allow, alternative identifiers can extend addressability on browsers that restrict third-party cookies: ID5, First-ID, Utiq, LiveRamp and The Trade Desk's EUID, for example.
These solutions do not all work in the same way or cover the same environments. And cookieless must not be confused with consentless: privacy obligations still apply, and the permitted activation depends on the context.
For a broader look at this challenge, see how to approach Safari inventory and what programmatic curation can achieve.
6. How GreyFox Ads estimates active reach ๐ฆ
At GreyFox Ads, we base the estimated volumes of our Master Deals on active identifiers, rather than using inflated audience counts as a selling point.
Our curated deals also filter bid requests to help DSP algorithms focus on quality supply:
- SPO: prioritising publishers' own SSP seats and resellers that demonstrably add value
- Inventory quality: excluding MFA sites and problematic subdomains
- Protection: brand-safety and anti-fraud filters
- Creative quality: favouring ad sizes with meaningful share of screen
- Targeting: combining cookie-based and cookieless options where appropriate
The objective isn't to claim the biggest possible audience. It's to help Media Traders reach a credible, active and genuinely addressable one. We also advise programmatic sales houses and trading desks on their activation strategies.
Want to make your programmatic audiences work harder?
Let's talk about active reach, curation and more effective programmatic campaigns across the quality French Open Web.
Original LinkedIn post: Read the original discussion.