Most remarketing campaigns start from the same observation: the majority of people who enter an online store leave without buying, and some of them were one step away from ordering. Remarketing means ads shown only to these people, who have already been on the site, while they browse elsewhere.
Done carelessly, it looks like this: a single audience, “all visitors”, followed for a long time with the same banner, including after they have bought. It costs money, it annoys people and it inflates the reports with sales that would have come anyway.
Three kinds of visitor, by how close they came to ordering
Segmentation is based on the last step taken in the store. Each step says something different about intent and calls for a different message, a different duration and a different amount worth paying to bring the person back.
Durations are chosen according to how long the buying decision takes. With a cheap product, interest fades quickly; with a sofa or a laptop, people compare for much longer. Following them beyond that window is wasted budget.
- Viewed a product: weak interest; dynamic ads, showing exactly the products viewed, on a small budget.
- Added to basket or started checkout: strong interest; a message about what usually holds them back (delivery, returns, payment methods), over a short period.
- Bought: they leave the other segments; later they can be shown complementary products or, for consumables, a reminder to restock.
Remarketing campaigns with capped frequency and buyers excluded
Frequency is the number of impressions the same person sees in a given period. Above a certain threshold, the ad no longer persuades; it irritates. On some campaign types a cap can be set directly; on others, frequency is kept in check through the budget, the audience duration and rotating the ads.
Excluding buyers seems obvious and is missing surprisingly often. It is done in two ways: from the order event measured on the site and, where there is a legal basis, from the store’s customer list, uploaded to the platform in hashed form, if the account has access to this feature.
Some categories are not suitable: Google’s rules do not allow remarketing lists on sensitive topics, such as those related to health, and Meta limits the data used for such categories. Which category you fall into is checked at setup.
The ad and the abandoned basket email, in the same sequence
The abandoned basket email is the message sent automatically to someone who left products in the basket and got as far as giving their address. It is cheaper than advertising, but it reaches only those whose address is known and for whom there is a legal basis for sending, established with your data protection officer. Ads cover the rest.
The two are planned together. Usually the email goes first, shortly after the basket is abandoned, and the ad comes into play only if the person has not come back. Be careful with discounts: if every abandoned basket is rewarded with a voucher, shoppers quickly learn to abandon on purpose.
Cookie consent shrinks the audiences
Remarketing relies on advertising cookies, which are used only with the visitor’s consent. Anyone who clicks “reject” in the consent banner does not enter any audience. At Google, the mechanism is called Consent Mode: the measurement tags change their behaviour according to the choice made in the banner. The Meta pixel, in turn, has to be configured to fire only after consent.
The practical consequence: the lists are smaller than the traffic in the reports. For a store with few visits, the “added to basket” segment can stay below the minimum number of people the platforms require before ads start showing. In that case we widen the segment or the duration, or rely more on email.
A banner designed to extract consent (a hidden reject button, pre-ticked boxes) is a risk, not a solution.
How much of the sales is really down to the ads
Remarketing reports almost always look splendid, because the ads reach people who were getting ready to buy anyway. Some of those orders would have come without them. That is why we do not judge these campaigns by the platform’s figure alone.
Work on a set-up like this usually goes through the steps below.
- Checking the events: product viewed, basket, checkout started and order are recorded correctly and only with consent.
- Ads by segment: dynamic ones, from the product feed, plus a few messages about delivery, returns and payment.
- Pause tests: from time to time, a segment goes without ads for a while, so we can see how much orders actually fall.
- View-through conversions: orders attributed without a click are read separately, because they easily inflate the result.
Frequently asked questions
Why does the store’s ad also appear to people who have already bought?
Either buyers are not excluded from the audience, or the order was placed from another device or without cookie consent, and the platform never found out about it. The first cause is fixed in the settings. The second is reduced by uploading the customer list, where there is a legal basis.
Is remarketing worth it for a store with low traffic?
It depends on whether the audiences reach the minimum threshold the platforms require. With low traffic, a single broad segment and the abandoned basket email are usually enough. The main budget is better spent on attracting new visitors.
Do I have to offer a discount to people who abandoned their basket?
Not necessarily. Many abandon for reasons that have nothing to do with price: a delivery cost discovered late, a missing payment method, an interruption. A message that clears these things up comes before the discount, which is kept as the last step.
This is a typical project description: it shows how we usually approach this kind of work and does not present a project carried out for a particular client. Every real project starts from your company’s situation, and the stages, timescales and price are agreed after the initial discussion.