Remarketing: Definition, How It Works and Types

Remarketing needs to know who visited what, and it relies on three technical mechanisms that are usually combined.

The most common is the remarketing pixel, a snippet of code from Google Ads, Meta Ads or another platform, usually installed on the site through a tag manager. When someone visits a page, the pixel logs that visit and, if the user already has a cookie from that platform, links it to their identifier. From then on, the platform can show them ads while they browse other sites in its advertising network.

The second mechanism is cookies, either first-party (set by the site itself) or third-party. They store which pages a user visited, which products they viewed, or whether they added something to the cart without completing the purchase. As third-party cookies are gradually phased out in Chrome, the first scenario carries more weight: data a brand collects directly on its own site.

The third mechanism is uploading customer lists, also called customer match or personalised audiences: a file of email addresses or phone numbers, usually hashed, uploaded to the ad platform. This route doesn't depend on cookies and is mainly used to reach existing customers, for example past buyers who should be offered a complementary product.

All three methods feed the same thing: an audience, meaning a group of identified users the platform can serve ads to. The more specific that audience, the better the campaign tends to perform.

A special case is dynamic remarketing: instead of always showing the same ad, the platform combines the pixel with a product feed, the same catalogue that powers Google Shopping or Meta's catalogue, to automatically generate an ad featuring the exact product each user looked at. That's the mechanism behind ads that seem to "remember" the item someone was browsing in an online store. It takes more setup work upfront, but it usually performs noticeably better than one generic ad shown to everyone.

An additional technical detail is cross-device matching: if a user logs into their Google, Meta or store account, the platform can link the same browsing history across their phone, laptop and tablet instead of treating them as separate visitors. Without that login, each device generates its own cookie, and remarketing loses continuity if the user switches screens midway through the buying decision.

Not every remarketing audience is worth the same. Lumping them all into one campaign usually performs worse than splitting them by purchase intent.

General site visitors. The broadest group: anyone who was on the site in the past days or weeks, with no behavioural filter. Useful to keep the brand top of mind, but with a lower conversion rate than tighter segments.

Cart abandoners. Users who added a product to the cart but didn't complete the purchase. One of the best-performing segments, because purchase intent was already clear: an ad featuring the exact product, sometimes paired with an incentive like free shipping, is often enough to close the sale.

Product page visitors without conversion. Someone who browsed several product pages or categories without adding anything to the cart. They can be shown dynamic ads featuring the exact products they looked at.

Existing customers. People who already bought, segmented for cross-selling or up-selling: offering a complementary product or a higher-tier version. Here the source is usually the directly uploaded customer list, not a browsing cookie.

Segmentation by country or language. Relevant for businesses selling in several markets: the same remarketing campaign needs to adapt the ad's language, and sometimes the offer or currency, depending on which country the user visited the site from.

Inactive subscribers. Users who signed up for a newsletter or created an account but never engaged further. Remarketing helps re-activate them before they're written off.

High-value customers. Customers with a purchase history or an average order value above the norm, who deserve different treatment: early access to new products, a loyalty programme, or simply less ad pressure, because the goal is no longer to win a first purchase but to keep them coming back.

Each segment needs its own message: a cart abandoner doesn't respond to a generic brand ad the way they respond to one showing the exact product they left behind. That segmentation work is what separates a profitable remarketing campaign from one that just burns budget.

Remarketing depends on identifying a specific user, which puts it squarely under the GDPR and cookie rules (ePrivacy) in the European Union. This isn't an optional detail: setting a remarketing pixel or tracking cookie without the user's prior consent is a violation, and data protection authorities have actively enforced it in recent years. This affects not just the classic cookie but any form of pixel-based or server-side tracking, as soon as the data can be linked to an identifiable person.

In practice, this means the remarketing pixel or cookie should only activate after the user accepts that specific category in the consent banner, never before and never by default. Most consent management platforms let you configure this distinction by cookie category, separate from strictly necessary cookies.

The privacy policy also needs to state clearly what data is collected, for what purpose, and for how long it's kept, and give users a simple way to withdraw consent at any time. This summary doesn't replace legal review: every remarketing setup should be checked with whoever handles compliance at the company, especially when combining several data sources such as cookies and customer lists.

Since 2024, Google has also required Consent Mode v2 to keep building remarketing audiences on its platform within the European Economic Area: without a correctly implemented consent signal, Google Ads stops building those audiences even if the pixel is still installed on the site. Meta Ads applies an equivalent requirement for its EU campaigns: without a valid consent signal through its own consent management setup, it stops receiving the events needed to build those audiences.

Well-executed remarketing shows up in results, not just in impression counts. A few habits make the difference:

• Segment by intent, not just by visit: a cart abandoner and a casual visitor shouldn't see the same ad.
• Set a frequency cap so the same ad doesn't show dozens of times to the same person.
• Exclude anyone who already converted, to avoid spending budget on someone who already bought what's being advertised.
• Refresh the creative every few weeks; an ad the user has already seen ten times stops generating clicks.
• Adjust the audience's time window to the purchase cycle: a few days for an impulse product, weeks or months for a longer decision.
• Measure performance by segment, not just overall, to see which audience actually drives sales and which one just adds cost.

The most frequent mistakes aren't technical, they come down to setup and judgement:

• Skipping the frequency cap: the same person sees the ad fifteen times in two days and associates it with intrusive advertising instead of the brand.
• Firing the pixel before the user gives consent, which is both a legal problem and a source of unreliable audience data.
• Chasing buyers with the same ad for the product they just purchased.
• Using a single generic audience for all traffic instead of splitting by behaviour.
• Leaving remarketing campaigns running indefinitely without reviewing performance or refreshing the ads.
• Ignoring the phase-out of third-party cookies and not preparing a first-party data strategy.
• Skipping a small test run before scaling the campaign; it's worth checking with a limited budget that the audience was built correctly and the ad is showing where it should.