Retargeting has the best attributed ROAS of almost any channel. It also has the largest potential gap between what attribution reports and what the advertising is actually causing. For most brands running retargeting at meaningful scale, understanding that gap is worth real money.
This is not a theoretical concern. It changes budget decisions.
A holdout test on retargeting often reveals much of the audience would have converted anyway
Why Retargeting Looks Great in Attribution
Retargeting targets people who have already visited your website. Someone who browsed your product page, added an item to their cart, or spent three minutes on your site has already expressed commercial intent. They know your brand. They're considering your product.
Many of them are going to buy regardless of what happens next.
When these users are shown retargeting ads and then purchase, often within days, attribution models credit the retargeting campaign. In last-click attribution, the retargeting ad gets 100% of the credit. In data-driven attribution, it still gets disproportionate credit because it appeared close to the conversion.
But the ad didn't create the intent. The intent was already there. The ad showed up to someone who was almost certainly going to convert anyway, and it got credit for the conversion.
The Counterfactual Question
The only measurement question that matters for budget decisions is the counterfactual: what would have happened without the advertising?
For prospecting campaigns reaching cold audiences, people who have never heard of your brand, the counterfactual is clear. Without the ad, most of these people would not have purchased. The organic purchase rate among strangers to your brand is very low. So most attributed conversions are genuinely incremental.
Retargeting audiences are different. Cart abandoners might convert at 3–7% organically over the following week, without ever seeing another ad. Site visitors with high engagement might convert at 2–4%. These people are warm. Your retargeting ad is competing against a baseline that is already meaningfully high.
If your retargeting campaign shows an attributed conversion rate of 5% and your organic (no-ad) conversion rate for the same audience is 3.5%, then only 1.5 percentage points of that 5% is actually incremental to the advertising. The rest would have happened anyway.
What the Data Typically Shows
When brands run holdout tests on retargeting campaigns, the results are consistently humbling.
The incremental lift, the additional conversions caused by the advertising, typically falls somewhere between 20% and 50% of the attributed conversions. The rest are organic purchases that the retargeting campaign intercepted on the way to checkout and claimed as its own.
This means a retargeting campaign reporting an 8x attributed ROAS might have an actual incremental ROAS of 1.6x to 4x, still potentially profitable, but a very different picture from what the dashboard shows. In some cases, particularly with broad retargeting audiences and long attribution windows, iROAS falls below 1x. The campaign is incrementally unprofitable while looking great in reporting.
The numbers vary by brand, audience definition, and attribution window. But the directional finding, that retargeting significantly overstates its incremental contribution relative to other channels, is one of the most consistent findings in marketing measurement research.
The Retargeting Tax
When you spend money on retargeting audiences that would have converted organically, you are paying a "retargeting tax", ad spend that recovers no incremental revenue, but does increase your reported ROAS.
Consider what that budget could do deployed elsewhere. Prospecting campaigns reaching cold audiences who would not have discovered your brand otherwise typically show much higher incrementality. Upper-funnel YouTube campaigns driving brand awareness in new audiences can show strong incremental lift even when attribution makes them look mediocre.
The optimization that feels prudent, keep scaling what has the best ROAS, which is usually retargeting, can systematically starve the channels that actually generate incremental growth.
This Doesn't Mean Retargeting Is Useless
It means you need to be more precise about how you use it.
Tighter audience definitions improve incrementality. Cart abandoners with 7-day recency have more genuine incremental potential than all site visitors in the last 90 days. Someone who loaded your checkout page and didn't complete probably needed a reminder. Someone who spent 30 seconds on your homepage three months ago probably didn't.
Shorter attribution windows reduce overcounting. A 7-day click window on a retargeting campaign claims most purchases that were going to happen anyway within that week. A 1-day click window is more honest about what the ad actually caused.
Frequency caps reduce wasted spend on audiences who are going to convert regardless. If someone is going to buy within 24 hours without any nudge, showing them the same retargeting ad six times that day is pure waste.
Sequential creative can add genuine value, moving someone from consideration to decision faster than they would have moved organically. A retargeting creative that addresses a specific objection (shipping time, return policy, comparison with a competitor) can provide new information that actually influences the decision.
How to Find Out
Run a holdout test. It's the only direct way to measure incrementality.
Suppress ads to 15–20% of your retargeting audience for four weeks. Keep showing ads to the other 80–85%. Measure actual orders in both groups from your backend data, not platform-attributed conversions.
Calculate the incremental conversion rate (test group rate minus control group rate). Multiply by the test group size to get incremental conversions. Divide by your ad spend to get iROAS.
What you find will probably be surprising. It's almost universally lower than attribution suggests. But it will also tell you whether you have a real problem (iROAS below your margin threshold) or whether your retargeting, while inflated in attribution, is still incrementally profitable.
If you discover low incrementality: tighten audience criteria, reduce frequency, cut budgets on the broadest audiences, and reallocate toward prospecting. Then re-test in three to six months to see if the changes improved your incremental performance.