Short answer
Remarketing means showing ads again to people who have already visited your site. The order is: cookie consent first, then lists built with a tag, then excluding buyers, a frequency limit and a message that differs from the first ad. Whether the sales a panel shows are extra sales is tested separately.
A visitor looks at a pair of shoes, does not add them to the basket and closes the tab. The next day the same shoes follow them onto a news site, then into the start of a video. If someone who ordered those shoes three weeks ago still sees the same ad, the list has a fault. If someone who clicked “reject” in the cookie banner sees it too, the fault is bigger. Remarketing is showing ads again to people who have already visited your site, but the hard part is not the ad: it is who goes onto the list, who comes off it and who never gets on it.
This guide gives the order: consent, lists, exclusions, frequency, creative and measurement. If you have not opened your first campaign yet, start with how to run your first Google Ads campaign and how to run Instagram and Facebook ads; none of that setup is repeated here. For a short definition of the term, the glossary entry on remarketing is enough.
What remarketing does and does not do
Google’s help page describes it as reaching existing customers and people who have previously visited your website or app. The same page says that in Google Ads the “remarketing” heading is now called “your data” segments, and older write-ups and screenshots still use the old name. In Meta the same job is done with custom audiences.
Remarketing does not find new customers; it reminds people who have already shown interest. That means it can do two things at once: bring back someone who was undecided, and also take credit for a sale that was going to happen anyway. The last step of this guide deals with that difference.
Step 1: set up consent first, then the tag
Lists are filled by cookies and tags. According to Google’s help page, when a visitor reaches a page their cookie ID is added to the segment. So the cookie choice is the door to the list. A visitor who refuses must not get onto it, and in technical terms that means tags that do not run before the choice is made.
- Google side: Google’s consent mode guide asks for the
gtag('consent', 'default', ...)command to run on every page, before any command that sends measurement data. Besidesad_storageandanalytics_storage, two more parameters are sent for advertising:ad_user_data(consent to send advertising-related user data to Google) andad_personalization(consent for personalised advertising). In the guide’s example the default is set todenied; when the visitor chooses in the banner, anupdatecommand changes the value. With Google Tag Manager the same job is done with the tag manager’s consent APIs. - Meta side: Meta’s developer documentation shows
fbq('consent', 'revoke')to pause Pixel events until consent is given, andfbq('consent', 'grant')once it is. The same document says revoke must be called on every page. - Customer lists: Lists of emails or phone numbers that you upload do not depend on cookies, but the people must have allowed this use. We explain how email and SMS consent is built in email and SMS consent in Türkiye; which legal basis is enough for advertising is a legal question for your own lawyer.
How we set up consent mode alongside GA4 is in the GA4 setup checklist, step by step. If you skip this step and build the lists first, every later step rests on the wrong foundation.
Step 2: build the lists in the platforms
The logic is the same on both platforms: a source (tag or Pixel) records the visit and rules define the list. The differences are in this table.
| Google Ads | Meta | |
|---|---|---|
| Source | The Google tag (optional event snippet) | Meta Pixel; it must be installed and active |
| How a list is defined | A rule, for example “URL contains”; ready-made “all visitors” and “all converters” segments | Audiences, Create Audience, Custom Audience, Website Traffic; you pick the Pixel, the visitor type and a number of days |
| By page | With a URL rule | “People who visit specific web pages”, or “specific pages but not others” |
| Customer list | Customer Match: contact details customers gave you, matched when the person is signed in to Google | Customer list custom audience |
| Watch for | Google says only segments with more than 100 active users can run ads | How many days people stay on the list is your choice |
Google’s help page recommends starting with the automatically created segments as best practice. The “all converters” segment stays empty if you have no conversion measurement, so finish that first if you are still setting it up. Very narrow rules can also keep a list below 100 people, so for a first setup a few broad lists, split later, is safer.
Step 3: which audiences to build first
It is tempting to open a separate list for every page. Three or four lists are enough at the start; the table below shows the order. We explain how to build audiences in GA4 with conditions, sequences and membership duration in our GA4 audiences article.
| Order | Audience | Who | Why in this order |
|---|---|---|---|
| 1 | Basket abandoners | Added to the basket, did not finish paying | The clearest intent to buy |
| 2 | Product or category page viewers | Looked at a product, did not add it | There is intent but it is less clear; the message has to follow the product |
| 3 | All visitors | Everyone who entered the site | Large and rough; a fallback until the first two fill up |
| 4 | Customer list | People who bought before | For repeat purchase and cross-sell; needs a consent basis |
We single out basket abandoners because the reason a basket was abandoned may sit on the site itself. Before you call them back with an ad, look at the cause in our guide on reducing cart abandonment; otherwise you lose the same person twice. For a service business, “opened the form but did not send it” or “saw the pricing page” plays the same role as the basket.
Step 4: exclude people who already bought
Every list needs an exit rule: once someone completes the conversion, they leave that ad. There are two routes.
- Google Ads: According to the help page you can exclude segments from Search, Display, Demand Gen, Standard Shopping, Video and Performance Max campaigns. An exclusion cannot be added while the campaign is being created; you add it to an existing campaign: in the Campaigns menu go to Audiences, then the Exclusions module and Edit Exclusions. The same page notes that an exclusion may not apply to users who have opted out of ads personalisation.
- Meta: The example on the website custom audience page selects people who visited a sales page but not the “about” page. In the same way you can select people who saw the product page but not the thank-you page. The thank-you page or the purchase event must be measured on your site as a separate address or event.
Exclusion has one limit: if measurement is incomplete, the buyer does not leave the list. So check that the purchase event works by comparing it with your order count; the method is in measure before you raise the budget.
Step 5: cap the frequency and say something new
Showing the same ad to the same person every day is nagging, not reminding. Two settings need to be thought about together.
Frequency. Frequency is the average number of times a person is shown an ad: impressions divided by reach. The frequency setting differs by campaign type on each platform, so check in your campaign settings which type has one. Where there is none, your controls are a short time on the list and frequency in the weekly report. To calculate it, use the ad frequency calculator.
Creative. The first ad introduced; the second has a different job. To people who looked at a product, show the product itself and an answer to an objection (delivery, returns, sizing); to basket abandoners, give a message that removes the obstacle at the payment step. Repeating the same image and the same sentence tells someone who has already seen the first ad nothing new.
This is the same flow as in Detmore’s remarketing work: flows built around visitor behaviour, with frequency under control. See the advertising service page for details.
Step 6: did it bring sales or only claim them
Everyone on a remarketing list already knows the site. Some of them would have come back without an ad. A panel can credit the whole sale to the campaign, so the panel figure alone should not decide anything. Three checks, in order.
- Read what the panel counts. Find out in your account whether conversions are counted after a click or after an impression, and which window is used; read the number with that in mind.
- Match orders to the till. Compare the sales the panel shows for a week with your shop or CRM records; the reasons for gaps are in why GA4 and ad platforms disagree.
- Pause on purpose. If the platform’s own lift test is available in your account, use it. If not, stop one list for a period you wrote down in advance and put total orders next to a comparable earlier period. If total orders stayed almost the same, the list may have claimed more than it sold. One period gives no firm answer, so note the season and the campaign calendar.
Before the budget goes up, these three questions should have written answers; the order is the same as in measure before you raise the budget.
Common mistakes
- A visitor who refused is on the list. If the tag runs without waiting for consent, the whole list cannot be trusted. Test the banner by opening the site and clicking “reject”.
- Not removing buyers. A customer sees the ad for something they own for weeks; it harms both budget and reputation.
- Never limiting the time. Showing today the same basket reminder to someone who looked six months ago makes no sense. Choose the time on the list to match your buying cycle.
- One ad for every list. A basket abandoner and someone who saw the home page should not get the same message.
- Counting the panel figure as profit. Remarketing looking best in the panel is normal; the problem is when that view steers decisions.
- Starting with narrow lists. Because of Google’s 100-user threshold, lists split too finely may never show an ad.
Common questions
Are remarketing and retargeting the same thing?
In everyday use, largely yes: both mean showing ads again to people who visited a site or app. Platforms use their own names: “your data” segments in Google Ads and custom audiences in Meta.
Can a visitor who refused cookies never be shown any ad?
A person who refuses must not join a list through their behaviour on your site. Seeing an ad from another source, for example through targeting of an audience you do not know, is a separate matter and is not remarketing. Which method may be used with which permission is a legal question.
How much traffic does remarketing need?
According to Google’s help page, a segment needs more than 100 active users to run ads. Meta’s help pages give no such threshold, though frequency climbs quickly on a very small list. With little traffic, keep the number of lists low and start with “all visitors” and “basket abandoners”.
One thing you can do today
Open your site in a private window, click “reject” in the cookie banner, browse a few pages and check, in your tag manager’s preview or the browser’s network tab, whether the advertising tags run. If they do, fix consent before you touch any list. If they do not, the next job is to write the exit rule for one list: who enters, who leaves, how many days they stay.
Sources
This article is general information, not legal, financial or technical advice. Platforms, their rules and the law change over time, so some details may have changed since 11 November 2026 or contain small errors. Before you make a decision for your business, check the current official sources or write to us.