What retargeting really delivers

Retargeting reaches people who have already sent a signal. They viewed a product, browsed a category or added something to their cart. So the ad doesn't meet indifference. It meets an open decision.

That is the core. Retargeting rarely creates demand. It shortens the path from interest to purchase. It keeps your brand top of mind while your customer compares.

The evidence: Does retargeting work, or would the customer have bought anyway?

That is the legitimate question. Anyone who was going to buy anyway doesn't need an ad. The answer comes from field experiments with real control groups.

Garrett Johnson, Randall Lewis and Elmar Nubbemeyer developed the Ghost Ads method for exactly this purpose. It compares users who see an ad with statistically identical users who don't. In a display retargeting campaign for an online retailer, the ads increased website visits by 17.2 percent and purchases by 10.5 percent.

Navdeep Sahni, Sridhar Narayanan and Kirthi Kalyanam followed up in 2019. In a large randomized experiment with an online home improvement retailer, switching retargeting on led to 14.6 percent more users returning to the website within four weeks. Even more important is timing. Around a third of the first week's effect occurred on the first day, half within the first two days. The paper was recognized as one of the best publications in the Journal of Marketing Research.

As early as 2013, Anja Lambrecht and Catherine Tucker showed that personalized product ads don't automatically perform better. While a customer is still browsing broadly, more general ads often work better. Only once they have narrowed down their choice do dynamic product ads play to their strengths.

The environment has changed since then. Jeffrey Larson and Jeffrey Dotson note in 2026 that privacy regulations such as iOS 14 and the GDPR have significantly reduced the effectiveness of retargeting. In their experiment on Google and Facebook for a mattress brand, prospecting delivered high and stable returns, while retargeting delivered considerably lower ones.

What does that mean for you? Retargeting demonstrably works. But it no longer runs on autopilot. It requires clean consent, fast delivery after the visit and a model in which you don't carry the risk. And it complements new customer acquisition rather than replacing it.

The first visit is rarely the last

A recent survey by a major retargeting provider among 4,500 consumers in nine European countries confirms the pattern. Only 27 percent typically buy on their first visit to a website. Half come back two to three times, almost one in five four to five times. Those who leave without buying usually compare prices, read reviews or look for discounts. (Adzine, September 2026)

The provider concludes that retargeting should run over longer periods. That fits its business model. We are more cautious. The survey shows how people shop. It does not show whether an additional ad triggers the purchase or whether it would have happened anyway.

Our conclusion: The decision can take days. But the ad's impact is greatest in the first days. That is why we start early and focused, reduce frequency afterwards and measure with short windows. A longer runtime does not justify longer attribution windows.

Why its reputation is still poor

We won't sugarcoat it. Retargeting has partly earned its reputation.

First, frequency. Anyone who sees the same ad twenty times a day feels followed. Second, ads after the purchase. Nothing looks less professional than advertising a product the customer already owns. Third, measurement. The return reported in platforms is usually higher than the true incremental return, because ads shown to people with purchase intent collect credit for a decision already made. Retargeting is the textbook example.

That is a real problem. But it is a problem of poor execution and poor measurement. Not a problem of the channel.

What makes good retargeting

The differences between annoying and effective retargeting are concrete.

• Control frequency. Clear caps per user and time period protect your brand and your budget.
• Be fast. The effect is greatest in the first days after the visit. Starting only after a week wastes the most valuable moment.
• Exclude buyers. Anyone who has purchased leaves the audience immediately. That saves money and avoids frustration.
• Segment by recency. A visit from yesterday deserves a different approach than one from three weeks ago.
• Match creative to the stage. If the customer is still browsing, show the breadth of your range. Once they have decided, show exactly their product. Feed based ads pull prices and availability directly from your product feed. So every ad is accurate.
• Choose clean environments. Your ad only works when real people see it in reputable environments. Private deals with premium publishers and an invalid traffic check before every bid rule out bots and questionable sites.
• Respect consent. Retargeting only works with clean consent under IAB TCF. That is not an obstacle. It is the foundation of trust.

Measure what matters

The key question is not: How many conversions does the dashboard show? It is: How many purchases would not have happened without the campaign?

That is what holdout tests and control groups are for. Today they are far cheaper and more accessible than just a few years ago. Anyone who takes retargeting seriously regularly tests its true incremental effect. That includes short, clearly defined post view windows. Long windows embellish numbers. Short windows show impact.

Our approach: We carry the risk

For us, this is the strongest argument for retargeting. We offer it on a CPO basis. That means we only get paid when a purchase happens. We cover the media costs entirely ourselves.

That changes the logic. If we waste budget on people who would have bought anyway, or on bots, we lose money. So we have every interest in clean environments, smart frequencies and real incremental impact. Your goals and our goals are the same.

Conclusion

Retargeting is not a cure all. It doesn't create demand out of nothing. But it wins back customers who were almost there. Research proves a real incremental effect, even if it has become smaller under stricter privacy rules. The channel's weaknesses are known and solvable. What matters is who executes it and how. And who carries the risk.

#morethanperformance