Since the platform privacy changes, retargeting audiences shrank, attribution windows tightened, and the reported return on retargeting fell. A lot of advertisers concluded retargeting stopped working and moved the budget to prospecting.
Often that was the wrong call, made on a number that no longer means what it used to.
What actually changed
- Audience sizes fell because fewer users can be tracked across apps and sites. Your 30-day site visitor pool is genuinely smaller than it was.
- Attribution windows shortened. A conversion that happens outside the window is not credited, even though it happened.
- Cross-device paths broke. Someone who sees the ad on a phone and buys on a laptop frequently appears as two unrelated people.
Note that two of those three are measurement problems, not performance problems. The sale still happened; the report just does not show it as yours.
A channel that stops being measurable is not the same as a channel that stops working. Most of the industry treated them as the same thing.
How to read it honestly
1. Compare blended, not platform-reported
Take total revenue divided by total ad spend across everything, for the month. Then turn retargeting off for two weeks and look at that blended number again. If blended performance drops, retargeting was contributing regardless of what the platform reported.
This is crude. It is also the most honest test available to a small business, and it beats arguing about attribution models.
2. Ask the customer
A single post-purchase question — what made you decide? — catches influence that no pixel can. When people keep mentioning that they saw you again after visiting, that is your retargeting working, uncredited.
3. Watch new-customer share
If retargeting is doing its job, it should convert people who already know you, while prospecting brings new ones. If your retargeting spend is mostly reaching existing customers, you are paying to harvest sales you already had — a real problem, and a different one from the channel not working.
The structure that still works
- Keep the pool alive with content, not just product pages. Blog readers, video viewers and social engagers are all retargetable and far larger than your cart abandoners. This is the part where content and paid stop being separate budgets.
- Segment by intent, not just recency. Someone who read an article needs a different message from someone who abandoned a cart. Same audience tool, completely different job.
- Cap frequency deliberately. Retargeting is where ad fatigue does brand damage, because these people already know you. Above roughly four or five exposures a week you are buying irritation.
- Change the creative, not just the discount. The default move is to increase the offer. Often the actual blocker is an unanswered objection, and a piece of proof beats ten percent off.
- Send server-side conversion data so the platform can match conversions it would otherwise miss. This recovers a meaningful share of the reporting gap. It is unglamorous tracking work and it pays.
When to actually cut retargeting
Retargeting genuinely does not deserve budget when your traffic volume is too low to build a meaningful audience — if you get 300 visitors a month, there is nobody to retarget, and that money belongs in prospecting or in fixing the funnel.
It also should not be the majority of your spend. If more than a quarter of your budget is chasing people who already know you, you have a growth problem dressed up as an efficiency win.
To see how much the platforms are over-claiming on your account, use the blended ROAS and MER calculator.
Questions people ask
Does retargeting still work?
Yes, but it reports worse than it performs. Smaller audiences and shorter attribution windows mean the platform now sees less of what retargeting contributed, so the number fell further than the effect did.
How should I measure retargeting?
On incrementality rather than on reported return. Turn it off for a defined period and watch total business, not the campaign’s own figure. Retargeting is the format most likely to claim credit for sales that were already coming.
What audience window should I use?
Short for high-intent actions like a cart or a pricing page, longer for broad site visitors. The common mistake is one 180-day pool of everyone, which mixes people who were about to buy with people who bounced in March.
When should I cut retargeting?
When the audience is too small to deliver, when frequency is climbing with no response, or when an incrementality test shows the sales would have arrived anyway. Not because the reported return fell after a privacy update.
Want help telling the difference on your account? Send us the numbers.