Retargeting Is the Best-Performing Line in Your Account. That's the Problem.

Pull up the channel report in almost any account spending real money and the shape is the same. Prospecting sits at two or three times return. Retargeting sits at eight, twelve, sometimes twenty. Somebody points at the retargeting row and asks the obvious question, which is why aren't we putting more money there.

Right question, wrong number.

Retargeting's reported return is the least trustworthy figure in a paid account, and the reason is structural rather than anybody's mistake. Last-click attribution, which every platform defaults to, hands the sale to the final ad a person touched. Retargeting ads are by definition the last thing someone who was already coming back sees. The channel is standing at the finish line collecting credit for the race.

At The RealReal we moved off last-click onto a multi-touch model that gave proportional credit to every touchpoint, and the channel mix reshuffled on the spot. Channels that had looked efficient turned out to be mostly re-acquiring customers we already had. Channels that had looked expensive turned out to be bringing in people who were genuinely new. Nothing about the spend changed that month. What changed was the measurement, and the measurement was what the budget decisions had been resting on.

None of which makes retargeting a bad layer. Built properly it is the most efficient thing in the account, and I want teams spending more there, not less. It is just that spending more on the row with the big number and building the layer so the number means something are two different projects.

The cross-check before the budget conversation

A properly architected retargeting program beats cold prospecting on every line, and by amounts you can hold it to. Cost per click 30 to 60% lower. Conversion rate to lead or demo two to five times higher. Effective return of five to fifteen dollars per dollar in B2B, three to eight in consumer. Average frequency of six to fourteen impressions per converted user.

So the decision rule is a comparison, not a threshold: if retargeting is not clearly beating cold prospecting on those, the audience architecture is wrong, the exclusions are wrong, or the creative is stale, and moving budget in will make all three more expensive. If it is beating cold by a factor that looks too good to be true, check your exclusions before you check anything else. I have written before about why honest conversion data comes before any of this, and retargeting is where dishonest data is most flattering and least visible.

Start with who you are paying to reach twice

The single largest waste in retargeting is not bad creative. It is showing ads to people who have already bought. In accounts I audit, 15 to 20% of the retargeting budget is typically going to existing paying customers, which is a line item nobody ever approved and nobody can see, because in the report it looks like performance.

Four exclusions are mandatory, and none of them take a quarter to implement:

• Current customers, uploaded from the CRM as a suppression list and refreshed weekly, not quarterly
• Anyone with a booked meeting, held out until the meeting outcome is logged
• Employees and their immediate networks, who are often your most frequent site visitors
• Anyone who has hit a thank-you or post-conversion page

Fixing the audience architecture and the exclusions together usually cuts wasted retargeting spend by 30 to 50% in the accounts I get handed, and that recovered budget is already qualified. It goes back into the same layer aimed at people who have not bought. If you want a second read on what your own exclusion list is missing, that is a good use of a 20-minute conversation.

Four buckets, not one audience

Most accounts have one retargeting audience: all site visitors, 180 days, one ad. Every visitor should instead land in one of four buckets defined by what they did and how recently, because intent and message have to match or the frequency just accumulates as resentment.

Casual browsers, days 1 to 7. Visited, did not look at pricing or product. Goal is understanding of the mechanism, so the creative is educational. Cap at three a day.

Active investigators, days 1 to 14. Viewed pricing, product or the demo page. Goal is conversion to a lead, so the creative is a customer story with the return attached. Cap at five a day. This is the only bucket that earns that frequency.

Engaged non-converters, days 14 to 60. Multiple pages, maybe a download, no booking. The angle that did not work is not going to start working, so this bucket gets a different entry point: a new use case, a new proof, an assessment. Cap at four a day.

Stalled opportunities, days 30 to 90. Booked and did not progress, or started a trial and never activated. Cap at three a day with a two-week cooldown between bursts, because these people already talked to you and the wrong cadence here costs you the relationship rather than the click.

Inside each bucket the creative rotates on a fixed sequence: soft reintroduction on days one to three, specific proof on four to seven, the direct ask on eight to fourteen, risk reversal from fifteen. After day thirty with no conversion, drop the cap to one a day for two weeks, then exit the audience entirely. Past 45 days there is no conversion lift left to capture, only accumulating fatigue, and that is a bad trade at any price.

Set the windows from your own data, not the platform default

The 180-day default window exists because it is the widest number the platform can sell you, not because it describes your business.

Working with CamDo, a consumer hardware account, we looked at when purchases actually happened relative to the first visit. Most customers bought on that first visit. Around 20% came back and bought within two to five days. Around 30% bought somewhere between day twelve and day thirty. After that the tail was thin enough to be noise.

That distribution is the retargeting plan. It says the money belongs in a hard push through day five and a second, different push across the twelve-to-thirty window, and that anything past day forty-five on that account was being served to people who had already decided either way. Pull the same report for your business before you set a single window. Time to purchase varies enormously by category, and it is one of the few things in paid media you can know exactly rather than estimate.

What the rebuild is worth

Therma, now GlacierGrid, had been running a single "Request Demo" retargeting ad, unchanged for about six months, against one undifferentiated audience. Click-through was 0.2% and a demo was costing $450.

We rebuilt it as four tiers with a 30-day message sequence, frequency caps by tier, and a creative refresh every three weeks. Click-through went to 0.8% and cost per demo dropped to $120, a 73% reduction. The part I did not expect was the sales team reporting better demo quality, and complaints about being followed around the internet dropping off. Frequency caps are a performance lever and a brand lever at the same time, which is rare enough to be worth saying out loud.

The cheapest improvement to any of this sits outside the ad platform. Every person on your email list who is already in a nurture flow is someone you can stop paying a platform to reach, and every purchase in your CRM sharpens the suppression and the lookalike seeds that make the paid side cheaper. The owned engine lowers what the paid engine costs, and the paid engine feeds the lists and audiences that make owned compound. Neither one is the senior partner.

What to do tomorrow

Export your customer list from the CRM, upload it as a suppression audience in every platform, and set a weekly refresh. That is an afternoon and it is usually worth 15 to 20% of the line.

Then pull your own time-to-purchase distribution and set your longest window to the end of your real tail rather than the platform's. Then split the one audience into the four buckets and put the caps on.

If retargeting is still not beating cold prospecting after that, the problem was never the retargeting.