Remarketing

Remarketing in 2026 — What Still Works After the Cookie Changes

Jakub Pytlik | May 5, 2026 | 6 min read

In short Third-party cookies are in retreat: Safari has been blocking them for years, ad blockers (uBlock, AdBlock) are intercepting more and more, and GDPR + Consent Mode v2 wipes data from roughly 50% of users without consent. Chrome itself ultimately did not remove cookies (Google walked back the plan in 2024), but the market changed anyway. Remarketing still works — but on different rules. Audience lists fill up more slowly (a "90-day" list now behaves like 30–45), cross-site retargeting loses precision, but Customer Match, Enhanced Conversions and 1st-party-data lists work better than ever. Five things to do: deploy Enhanced Conversions, set up Customer Match from your email database, configure Consent Mode v2, monitor list size, and — for larger accounts — server-side tracking.

What changed — in 30 seconds

For more than a decade, remarketing ran on third-party cookies. Your site placed a file on the visitor's computer; that file was read by Google on other sites — and that is how the customer was followed by ads across the web. Simple, effective, but user privacy was a fiction.

Four things changed the market:

Apple ITP (since 2017, aggressive since 2020) — Safari blocks third-party cookies and limits even first-party cookies to 7 days of inactivity.
Chrome — the deprecation that did not happen — Google announced for years that it would remove third-party cookies from Chrome, but in July 2024 it walked back the plan. Cookies in Chrome still work, but Privacy Sandbox (Topics API) runs in parallel and the user can turn it off — that shifts the balance.
Ad blockers (uBlock Origin, AdBlock, Brave) — intercept the Google Ads tag, so a meaningful share of traffic (estimated 20–40% in Poland, more among more technical audiences) does not enter remarketing lists at all.
GDPR + Consent Mode v2 (since 2024) — without explicit user consent you cannot track for marketing purposes. A cookie banner that most users decline means a real 40–60% data loss — unless you have Consent Mode v2 (more on that below).

Practical outcome: remarketing lists are smaller, shorter and less precise than they were 5 years ago. But that does not mean remarketing is dead — the way you do it has changed.

What still works in 2026

1. Customer Match (from your email database)

You upload a list of your customers' email addresses to Google (hashed, GDPR-OK). Google matches them against its user accounts and creates a remarketing list from that. Pros: independent of cookies, long-term (the list stays alive as long as you keep updating it), very high conversion (these are people who already know you). Minimum: 1,000 matched addresses on the list before Google activates it. Important: Customer Match requires a Google Ads account with history and a clean Policy compliance status — a brand-new account may have to wait several weeks and demonstrate clean history before Google grants access to this feature.

2. Lists from 1st party data (your tags on the site)

The Google Ads tag on your site still works — it builds lists of visitors who have not blocked first-party cookies. The lists are smaller than they used to be (almost nothing builds on Safari), but they are still useful, especially when you add Enhanced Conversions.

3. Enhanced Conversions (hashed conversion data)

When someone converts (purchase, form submission), your site sends a hashed email/phone to Google together with the conversion. Google uses this to fill data gaps — when the cookie expired or was blocked. After deploying Enhanced Conversions I typically see 8–15% more conversions in reports (those that were "disappearing" before). The size of the lift depends on the prior tracking setup — the more data was leaking, the bigger the jump. If you already had server-side tracking working well, the effect is more like 2–4%, not 15%.

4. YouTube remarketing

On YouTube, a logged-in user is identifiable independently of cookies — through their Google account. Lists like "watched my video" or "visited my channel" are stable and resistant to browser changes. This channel got stronger relative to classic Display.

5. Optimized Targeting and Audience Modeling (Google AI fills lists)

The most overlooked piece. Google Ads in 2026 does not only use your "who clicked and came back" list — the algorithm finds similar users based on signals (YouTube behavior, search history, age of Google account). Your list of 500 people behaves as if it had 5,000 because Google models the rest. What this means in practice: a small business with low traffic is not condemned to "remarketing does not make sense". Google fills the gaps with AI. Turning this on in Display/Demand Gen/PMax is usually a checkbox in settings, but the effect is noticeable.

What weakened or stopped working

Cross-site remarketing without login. The customer visits your site, then reads news on a major portal where they would previously have seen your display ad. Today — usually not. Anonymous Safari traffic does not enter lists at all; in Chrome it only enters when the user accepted cookies; in other browsers ad blockers often cut it.
Long remarketing windows. A "90-day" list behaves like "30–45 days" — not because the cookie has a shorter life, but because Safari clears data after 7 days of inactivity and ad blockers intercept the tag on every visit. The cookie is nominally 90 days, but in practice an abandoned cart from a Safari user will not rebuild into the list. If your decision cycle is longer than 30 days, the classic tag-based remarketing will not catch them anymore.
Small activation lists. Google has minimums below which the list does not serve ads. The thresholds differ by type: 1st-party site tag — from 100 users in Search, from 1,000 in Display/YouTube. Customer Match — from 1,000 matched addresses. A small business with low traffic may struggle to activate Display, but Search remarketing usually works.
Tracking without Consent Mode v2. Without Consent Mode, even consenting users are skipped in some categories.

Five things to do in 2026

Deploy Enhanced Conversions in Google Ads. 30 minutes of work with a developer. Effect: 8–15% more visible conversions + better Smart Bidding learning (size depends on how much your previous tracking was leaking).
Set up Customer Match from your email database. Export customers from your CRM, upload to Google Ads. Refresh once a month. You get the cheapest and most durable remarketing list that exists. If you have fewer than 2,000 contacts — skip this for now and revisit when the database grows.
Deploy Consent Mode v2. Without it, even consenting users are skipped in some categories. Implementation usually goes through Google Tag Manager + a banner CMP (e.g. Cookiebot, CookieYes).
Consider server-side tracking. More work (a separate server GTM), but it bypasses ad blockers and part of ITP. Worth it for larger accounts with a budget of 10,000+ PLN/month. For smaller ones — skip for now, the ROI is not there.
Monitor list size every 2 weeks. In Google Ads → Audiences — check "Membership status" and "List size". If the list suddenly drops by 40%, something broke with tracking (consent, tag, blockers).

What I see after 11 years

Cosmetics e-commerce — before deploying Enhanced Conversions: 240 conversions/month in reports. After deployment (1 day of dev work): 270 conversions in the first month and 285 in the second. Those "extra" conversions were already happening — Google simply did not see them before. Smart Bidding started optimizing better, customer acquisition cost dropped 12%.
B2B services — Customer Match from a database of 4,800 customers. Remarketing campaign on that list: CPC 1.80 PLN, conversion rate 22% (click → quote request — not a purchase; on a narrow list of known customers these levels are realistic, vs. 3% on cold traffic). The most profitable campaign on the account — "warm-list" campaigns usually are.
Local service — the client was running classic Display remarketing; ROAS dropped from 3.1 to 1.8 over a year. Diagnosis: the list shrank from 12,000 people to 4,100 (Safari ITP cuts traffic, ad blockers intercept the tag). We switched to YouTube remarketing + Customer Match + Optimized Targeting. Two months later ROAS was 2.9 — not what it used to be, but stable.

Who this affects most Small businesses with up to 5,000 PLN/month and low site traffic suffer most — lists need scale, and Customer Match needs 1,000 matched addresses. Mid-sized businesses with 10,000+ email addresses gain (from 5,000 addresses Google realistically matches 1,500–2,500, which works but is not yet powerful — at 10,000+ the database becomes a real engine). Big players run server-side tracking and behave as if nothing happened.

Summary Third-party cookies are gone, but remarketing still works — on different rules. Customer Match from an email database, Enhanced Conversions, 1st-party-data lists and YouTube are the new toolkit. Lists are smaller and shorter, but more accurate (less spam, more real customers). Five things to do: Enhanced Conversions, Customer Match, Consent Mode v2, monitor list size every 2 weeks, and server-side tracking for larger accounts.

I will check your remarketing setup for 2026 — for free Drop me a short note: do you have Enhanced Conversions, Customer Match and Consent Mode v2 in place? I will reply within 24h with what is probably missing on your side and what to fix first (usually 1–2 things deliver the biggest return). If you want a full configuration audit, I will need access to the account and Google Tag Manager. No contract, no pressure. Email me at jakub.pytlik@jp-ads.pl or use the contact form.

About the sources: Article based on observations from 11 years of running Google Ads campaigns at JP Ads + Google documentation on Enhanced Conversions, Customer Match and Consent Mode v2. Numbers from my own practice (8–15% more conversions after Enhanced Conversions, ROAS 2.9 vs. 3.1 after switching to new lists) are flagged "from my experience" — distinguish them from hard market research.

Related articles

← Blog Check your remarketing