Remarketing after third-party cookies: what still works
Remarketing has the best numbers in the report and the worst incrementality. What survived the cookie restrictions, how to build audiences from your own data, and how to tell when you are chasing people who would have bought anyway.
- Remarketing has the best numbers in the account and the lowest incrementality. Both for the same reason.
- Audiences now rest on your own data: customer lists, site events and offline conversions.
- Exclusion is now more valuable than targeting — especially excluding existing customers from acquisition.
Remarketing always has excellent numbers in the report: a low cost per conversion, a high return. That makes sense, because you are reaching people who have already been with you. For exactly the same reason it is the channel with the largest gap between what the platform credits itself with and what actually got added.
What survived third-party cookies
Classic remarketing built on cross-site tracking is in decline. What works today:
- Your own site events. Product views, add-to-cart, checkout start. Built on your data and your tracking code.
- Customer lists. Hashed emails and phone numbers uploaded to the platform. The most reliable input today, because it does not depend on the browser.
- Server-side event sending. More resistant to blocking and to browser restrictions. See first-party data.
- Offline and CRM data. Who bought, who enquired and did not buy, whose contract is coming up for renewal.
The common denominator: all of it is data you hold. Anyone who has not started building it over the past two years has lost more than just remarketing.
Audiences worth building
| Audience | What for | Watch out for |
|---|---|---|
| Abandoned cart, 1–3 days | Highest value, short window | A large share of these people return without an ad |
| Product view without cart, 7–14 days | Reminder of a specific item | Do not remind indefinitely; after two weeks it is just irritating |
| Existing customers | Repeat purchase, complementary range | Never in acquisition campaigns, only in their own |
| Customers nearing the end of a consumption cycle | The best ratio for consumable goods | Needs knowledge of the cycle, not just a purchase date |
| Enquiries that did not close | For services, the most underrated audience | Requires the CRM connected to advertising |
| Blog visitors | Cheap reach, low value | Do not mix into performance campaigns |
Exclusion is now more valuable than targeting
This is the most underrated part. If you do not exclude existing customers from acquisition campaigns, you are paying to be found by people who already buy from you regularly. For subscriptions and repeat purchase that is an immediate, measurable saving.
The minimum that should be excluded in every acquisition account:
- Existing customers (from a customer list, not from site visits).
- People who have just ordered — at least for a few days, longer for services.
- Job applicants and current employees, if you have a list.
- Your own IP addresses and staff, on smaller campaigns.
How to tell remarketing is just taking credit
Two checks that cost nothing:
- Look at the time window. If most credited conversions arrive within hours of an impression, from people who were already in the cart, it is very likely they would have bought without you.
- Narrow it and watch the total. Shorten the audience window from thirty days to three and watch what happens to total revenue. If nothing happens and only the remarketing campaign’s numbers fall, you have your answer.
The only reliable answer, though, is one: a holdout test where part of the audience sees no ads at all. Platforms now offer tools for this directly — see the article on incrementality.
For shops with a short purchase cycle, assume a substantial share of remarketing conversions would have arrived anyway. That does not mean switching remarketing off — it means not judging it by the same metric as acquisition, and not shifting budget into it simply because it has the best numbers in the report.
Frequency and fatigue
The most common customer complaint about advertising is not that it is badly targeted — it is that there is too much of it. Practical measures:
- Cap frequency, especially in display and social placements.
- Shorten audience windows. Thirty days is unnecessarily long for most shops.
- Change the message by stage. The same announcement for the tenth time does not work like the first.
- Remove people who have bought immediately, not when the window expires.
What to take away
Remarketing survived the end of third-party cookies, but it moved onto first-party data — and anyone without it is now in a materially worse position than two years ago. The main advice is unchanged though: do not judge it by platform numbers. It is the channel that looks best in the report and adds least in reality, and the right size for its budget can only be found by testing.
Common questions
Does remarketing still work without third-party cookies?
Yes, but on different foundations. Instead of cross-site tracking, audiences are now built from your own data: events on your site, hashed customer lists uploaded to the platform, server-side conversion sending and your CRM. Customer lists are the most reliable, because they do not depend on the browser or on cookies. Companies that have not started building their own data are now in a markedly worse position for remarketing.
Why does remarketing have such good numbers and still get doubted?
Because both follow from the same thing: you are reaching people who have already been with you and were often about to buy. The platform credits itself with their order, but a large share would have arrived without the advertising. Remarketing therefore has the best cost-to-return ratio in the whole account and the lowest incrementality. That is not a reason to switch it off, but it is a reason not to judge it by the same metric as acquisition and not to shift budget into it on the strength of the report alone.
Who should I exclude from campaigns?
From acquisition campaigns, above all existing customers — using a customer list, not merely site visits. For subscriptions and repeat purchase that is an immediate saving. Also exclude people who have just ordered, at least for a few days and longer for services, plus job applicants and your own staff. Exclusion is now often more valuable than finer targeting, because it directly stops you paying for people you already have.