The short answer: set a frequency cap of about 3 to 5 impressions per person per week, let the window follow the buying cycle (7 days for quick purchases, 30 for most, 90 for long B2B decisions), exclude people who've already bought, and change the message at least every ten days. And keep retargeting to a small share of the budget, because it doesn't bring in new customers.
We've all seen the opposite. You looked at a pair of shoes for three minutes, and the shoes followed you for three weeks: in your feed, on news sites, in the weather app. That isn't marketing. It's a setup mistake with a budget.
Why is retargeting so profitable, and so easy to get wrong?
People who've already visited your site are a warm audience. They know who you are and what you sell, so in my accounts the conversion rate for retargeting is often two to five times higher than for cold audiences. It looks fantastic in the report, so it's tempting to pour more money in.
Two things go wrong. First, a small audience can't absorb more budget by reaching more people, only by showing the same people the ad more often. Double the budget on 800 visitors and you double the frequency, not the reach. Second, the report credits retargeting with purchases that would have happened anyway. Someone who's already made up their mind will happily click an ad on the way to checkout, and the credit goes to "retargeting".
How often should someone see your retargeting ad?
Frequency is the number of times the same person sees your ad in a given period. My rules of thumb:
- 3 to 5 impressions per week in the short window, when the person has just visited and is actively thinking about it.
- 1 to 2 impressions per week in longer windows, where you just want to stay top of mind.
- More than 10 per week is where irritation sets in. If you see 15 or 20 in the report, nobody's been checking.
How you set the cap varies by platform. In Google Ads, Display campaigns have a campaign-level frequency cap where you set impressions per person per day, week or month. In programmatic platforms such as DV360, you can cap at several levels, including across every campaign in the account, so someone in three audiences still only sees you five times a week in total.
Meta is trickier. Regular conversion campaigns have no direct frequency cap; you only get one in campaigns with reach as the objective. So on Meta you control frequency indirectly: check the Frequency column with a seven-day filter, and if it goes above 5, lower the budget, widen the window or add more creative. Meta shows frequency for whatever period you've selected, so a campaign can look calm overall until you zoom in.
How long should someone stay in your retargeting audience?
The window (how many days a visitor stays in the audience) should match how long people actually take to make up their minds.
- 7 days: low-priced e-commerce, abandoned baskets, impulse buys.
- 30 days: most services and more expensive products where people compare a few options.
- 90 days: B2B, quote-based sales and anything where several people are involved in the decision.
Better still, don't pick one window: use three, in layers. One audience for days 0 to 7 with high intensity, one for days 8 to 30 at a calmer pace, and one for days 31 to 90 that mostly keeps the door open. Each layer excludes the shorter one, so nobody is in two at once. Technically, Meta allows website audiences of up to 180 days and Google up to 540, but just because you can doesn't mean you should.
Most important of all: exclude people who've already bought. Build an audience from the thank-you page or the purchase event and add it as an exclusion in every retargeting campaign. Chasing customers with an offer they've already accepted is the quickest way to make them feel like a number. Upload your customer list as an exclusion too, to catch people who bought over the phone or in a shop.
Why your audiences are smaller than they used to be
Retargeting relies on a pixel or tag being allowed to set a cookie in the visitor's browser, and under GDPR and the EU cookie rules, that means the visitor has to accept marketing cookies in your banner. Anyone who clicks "reject" never ends up in the audience. The share who say yes varies hugely depending on the banner, but expect a sizeable chunk of your visitors to disappear. Add Safari and iOS limiting how long cookies last, and an audience that was 10,000 a few years ago might be half that today.
That makes the frequency cap even more important, because the budget is concentrated on fewer people. It also makes the alternatives more attractive: customer lists with email addresses (where you have consent), Meta engagement audiences based on who's interacted with your posts and ads (they live inside Meta and need no cookie on your site), and YouTube viewers in Google Ads. Together they fill part of the gap. I explain why your ad accounts show fewer conversions than you actually get in the article on Consent Mode and server-side tracking.
What should the ad say at each step?
This is usually where it falls apart. The same ad runs for 60 days to the same people, who just get more and more tired of it. Follow up in steps instead, the way you would if you were calling someone:
- Remind (days 0 to 7). Show what they looked at, with clear branding and no fuss. They already know what it is; they just need to remember you.
- Inform (days 8 to 30). They haven't bought, so something is holding them back. Answer their objections: how it works, what it costs, what other customers say, what happens after they buy.
- Give a reason to act (days 31 to 90). A limited-time offer, a free walkthrough, limited places. But not a discount on everything for everyone; that just teaches your customers to wait until you chase them.
In practice, that means three ad sets (or ad groups), each with its own audience, creative and budget. Programmatic platforms often have built-in sequencing that does this automatically. Either way, an ad that looks the same for two months isn't a campaign. It's a screensaver.
How much of your budget should go to retargeting?
My rule of thumb is 10 to 20% of the total ad budget: the higher end if you have lots of traffic, the lower end if you don't. If retargeting takes 50%, you've stopped bringing in new visitors and are just milking the same 2,000 people. The warning signs: rising frequency, flat reach and a cost per thousand impressions that keeps creeping up.
If you want to know what retargeting actually adds, run the simplest test there is: pause it for two weeks and see whether total conversions fall. In my experience they do, but by a lot less than the report claimed retargeting was delivering. The difference is the purchases that would have happened anyway.
When should you skip retargeting altogether?
If your site has fewer than roughly 1,000 visitors a month, the audience is too small for the platforms to deliver evenly, and it costs more time than it's worth. Spend the money on getting more visitors first. And if you're already running retargeting with no cap, no exclusions and a single ad, switch it off today and rebuild it. If everyone sees the same ad 40 times, you haven't built awareness, you've built irritation, and people will remember your brand for that just as well.
What to do next
- Check the frequency column in every platform with a seven-day filter. If it's above 5, you've found your first problem.
- Build the exclusion: an audience based on purchases or the thank-you page, with a 30- or 90-day window, added to every retargeting campaign.
- Split the window into three layers (0 to 7, 8 to 30, 31 to 90 days) and give each layer its own creative: remind, inform, act.
- Set frequency caps where you can (Google Display, programmatic, Meta reach campaigns) and use budget and window to control frequency where you can't.
- Work out the share of your total budget that goes to retargeting. If it's over 20%, move money into finding new customers.
- Not sure where to start? Take the quick analysis: five questions, two minutes, and you'll get a picture of where your advertising stands today.
