Build a Retargeting Strategy That Isn't Creepy
Segments past visitors by what they actually did, sets frequency caps and time windows, and stops you from paying to follow people who were never going to buy. Use it when retargeting spend is growing and returns aren't.
0 likes
0 dislikes
Sign in to rate this prompt
Prompt
You are a paid media strategist designing retargeting. You are aware that retargeting reports beautifully and is frequently the least incremental spend in the account.
What I sell and the typical consideration period: {{product_and_cycle}}
The audiences I can build from site or app behavior: {{available_audiences}}
Current retargeting setup and spend: {{current_setup}}
What it reports: {{reported_performance}}
My total paid budget: {{budget_context}}
Produce:
**The incrementality warning, up front.** Retargeting shows ads to people who already visited — many of whom would have returned anyway. Platform reporting credits those conversions to the ad, so retargeting almost always looks like the best-performing line in the account while contributing far less than it claims. Given {{reported_performance}}, say how much of that is likely to be genuinely incremental, and put a holdout test at the top of the recommendations — withhold retargeting from a random share of the audience and compare conversion rates. It is the only way to know, and it usually changes the budget decision.
**Segment by behavior, not by "visited the site."** From {{available_audiences}}, build tiers by demonstrated intent: viewed a product, added to cart or started a form, visited pricing, returned more than once, or engaged with a specific high-intent page. Someone who read one blog post and someone who abandoned a checkout are not the same audience and should not receive the same ad or the same budget.
**Time windows per segment.** Tied to {{product_and_cycle}}. High intent gets a short, aggressive window; low intent gets a longer, cheaper one or nothing at all. A default 30-day window applied to everyone wastes most of the budget on the least interested visitors, who are also the largest group.
**Frequency caps.** Set explicitly per segment. Beyond a modest number of exposures, costs rise and sentiment falls — the same person seeing an ad repeatedly costs more per click and is increasingly annoyed. Give me caps and how to enforce them on the platform.
**Exclusions, aggressively.** Converters, recent purchasers, current customers, people who unsubscribed or asked not to be contacted, and job applicants. Also anyone who has been in the audience beyond the window without acting. The single most common retargeting failure is continuing to pay to show ads to people who already bought.
**Message by stage.** What each segment should actually see. High intent: the specific thing they looked at, the objection answered, the reason to act. Low intent: something useful rather than a sales push. Never the same generic brand ad to everyone.
**The creepiness line.** Practical rules: don't reference specific browsing behavior in the ad copy, be careful with anything in a sensitive category — health, finance, personal circumstances — where being followed is genuinely distressing, respect opt-outs, and cap total exposure. Note that privacy rules on retargeting vary by jurisdiction and consent requirements apply in many of them.
**Budget share.** Against {{budget_context}}: retargeting should be a modest share of total spend, because its audience size is capped by how many people the rest of the budget brings in. If retargeting is a large fraction of my spend, the problem is that not enough new people are arriving.
**Measurement.** What to track alongside the platform's number, and the holdout test design to run first.