Quick answer
Retargeting shows ads to people who already visited your site or app but did not buy, using a list built from what they did. Results depend mostly on list quality: who is on it, who is excluded and how often they see ads. Only a comparison with a held-out group shows whether it added sales.
Formula: Incremental conversions = (conversion rate of exposed group − conversion rate of holdout group) × size of exposed group; incremental cost per conversion = spend ÷ incremental conversions
What is retargeting?
Retargeting is paid advertising aimed at people who have already interacted with your business, such as visiting your site or using your app, but have not done what you wanted, usually buying. A list of those people is built from what they did, and ads are shown to them as they browse elsewhere. Google Ads describes the same idea under the name remarketing: re-engaging people who have previously interacted with your brand while they browse Google and partner sites (Google Ads Help).
The reason it is popular is that these people are warmer than strangers, which makes the campaign cheap per conversion. The reason it is easy to misjudge is the same: warm people often buy anyway. Retargeting performance is decided mainly by the quality of the list, not by the bid: who is on it (what they did, how recently), who is excluded (people who already bought), and how often they see ads. Even a perfect list leaves one question that only a comparison can answer: did the ads cause the purchase, or would it have happened anyway? That is the idea of incrementality.
Retargeting, remarketing, re-engagement and repositioning: what is the difference?
The four words sound alike and mean different things.
| Term | What it is | Where it runs | Who it reaches | Often confused with |
|---|---|---|---|---|
| Retargeting | Paid ads to people who visited or used your product | Ad platforms, display networks, social feeds | People on a list built from their behavior | Remarketing |
| Remarketing | Google's name for showing ads to people who interacted with you; some writers use it for emails to existing contacts | Google Ads and partner sites; sometimes email | Past visitors or existing contacts | Retargeting (often used interchangeably, so check the definition in use) |
| Re-engagement | Messages to lapsed customers from channels you own | Email, push, in-app | Lapsed users you can already contact | Retargeting, since the audience is similar; see re-engagement campaigns |
| Repositioning | Changing what the brand stands for in the market | Brand, product and pricing | The whole target market | Retargeting, because of the similar spelling; no list is involved |
If someone says "we need retargeting", ask which one they mean. A paid ad to past visitors and an email to lapsed buyers have different costs, different consent rules and different ways to measure.
What lists do retargeting campaigns use?
A list is a rule: an action, did or did not, a time window and sometimes a count. The most useful lists are the ones where the action shows intent and the window is short.
| List | Rule | Exclude | Typical window |
|---|---|---|---|
| Viewed pricing, did not buy | Viewed the pricing page, no purchase | Anyone who bought in the last 30 days | 7–14 days |
| Added to cart, did not check out | Cart add, no checkout complete | Anyone who completed an order since | 3–7 days; see cart abandonment rate |
| Lapsed customers | Purchased 90–180 days ago, nothing since | Anyone who purchased recently | 90–180 days |
| Product viewers | Viewed a product page, no add to cart | Past purchasers of that product | 7–30 days |
Three rules make a list work:
- Exclude buyers. Someone who bought yesterday should leave the list, or you pay for ads to people who have just finished converting. A common failure is a list that is never updated for purchases.
- Keep the window short for high intent. A pricing-page view from 90 days ago is weak evidence; one from 3 days ago is not.
- Cap how often people see ads. Google Ads lets you set a frequency cap per day, week or month on Display campaigns, at the campaign, ad group or ad level (Google Ads Help). Without a cap the same few people see the same ad dozens of times.
Lists also have to be large enough to run. In Google Ads, only segments with more than 100 active users are eligible to run ads (Google Ads Help). That is a platform limit, not a quality bar: a 150-person list can run, but it will not give you a stable measurement.
How do you measure whether retargeting adds sales?
Platform reports normally credit a retargeting ad with every conversion from a person who saw it. But the people on the list were already likely to convert. The fair comparison is with a group that was eligible and did not see the ads: a holdout.
Incremental conversions = (conversion rate of the exposed group − conversion rate of the holdout group) × size of the exposed group
Incremental cost per conversion = spend ÷ incremental conversions
The method is a standard two-proportion comparison (NIST/SEMATECH e-Handbook, 7.3.3). Economists have also shown that measuring the returns to advertising is statistically hard, because sales are noisy compared with the small effect any single campaign has, so experiments often need to be large (Lewis and Rao, 2015).
Worked example: 1,000 people, a holdout and what the ads added
Sample data: a list of 1,000 people who viewed pricing and did not buy. The list is split at random into 500 who are shown retargeting ads and 500 held out and shown none. Ad spend is $600; an order is worth $80.
| Group | People | Orders | Conversion rate |
|---|---|---|---|
| Shown ads | 500 | 30 | 6.0% |
| Holdout (no ads) | 500 | 24 | 4.8% |
| Difference | 6 | +1.2 points |
How to recompute: 30 ÷ 500 = 6.0%, 24 ÷ 500 = 4.8%, 6.0% − 4.8% = 1.2 points. Incremental orders = 0.012 × 500 = 6.
The two ways of reading the same campaign:
| Measure | Platform view (all 30 orders credited) | Incremental view (6 orders) |
|---|---|---|
| Cost per order | $600 ÷ 30 = $20.00 | $600 ÷ 6 = $100.00 |
| Return on ad spend | 30 × $80 ÷ $600 = 4.0 | 6 × $80 ÷ $600 = 0.8 |
The platform view says the campaign is excellent; the incremental view says it returned 80 cents per dollar. That gap is the entire argument for holdouts. Of the 30 orders from the exposed group, an estimated 24 (4.8% of 500) would have happened without ads.
The honest caveat: with 500 people per group, a 1.2-point difference is not statistically convincing. The pooled rate is 54 ÷ 1,000 = 5.4%, the standard error is √(0.054 × 0.946 × (1/500 + 1/500)) ≈ 0.0143, and z = 0.012 ÷ 0.0143 ≈ 0.84 (p ≈ 0.40). The same rates in groups of 5,000 would give z ≈ 2.65 (p ≈ 0.008). So a small list can run ads, but it cannot prove they worked; treat the 6 orders as a rough guide until the test is larger.
A second leak is waste on past buyers. If 150 of the 1,000 people (15%) had already bought and were not excluded, and impressions are spread evenly, 15% of the spend, 0.15 × $600 = $90, went to people who had just converted.
Mobile retargeting
Mobile retargeting applies the same idea inside apps and mobile browsers: show ads to people who installed your app, or browsed your mobile site, but then stopped. A typical list is "installed, has not opened the app for 14 days" or "added to cart in the app, no purchase".
The difference is how people are recognised. App campaigns rely on a device advertising identifier, and on Apple platforms an app must request the user's permission through the App Tracking Transparency framework before tracking users across other companies' apps and websites (Apple Developer). Lists on iOS are therefore smaller, and measurement by holdout matters more because platform-reported numbers are harder to verify. The list rule, exclusion and cap principles above are unchanged.
Common misunderstandings about retargeting
- "High return on ad spend means it works." Return on ad spend (see ROI and ROAS) credits every order from a warm audience; in the example, 4.0 became 0.8 once the holdout was applied.
- Counting natural repeat customers as retargeting wins. People who would have returned anyway fill the list and flatter the results.
- No exclusion for purchasers. In the example this wasted about $90 of $600.
- No frequency cap. The same people see the same ad again and again, which wastes spend and irritates them.
- Judging a tiny list. At 500 per group, a 1.2-point gap gave z ≈ 0.84. Wait for more data before changing the budget.
- Treating retargeting as a strategy. It collects people the rest of your marketing has already brought; it does not create demand. For how to tie spend to results across channels, see marketing ROI.
Retargeting in TapCub
TapCub does not run ads or place bids; it defines and counts the audience and measures the results. Here is how the example is checked in TapCub.
- Define the list as a rule. In All-platform analytics › Users › Segments & tags, create a rule segment: did the pricing page view in the last 14 days and did not do the purchase event. Preview the count; it refreshes daily and keeps a 30-day size trend. Anonymous visitors have no profile and never join a segment, so this list covers identified people (see segments); for anonymous visitors the ad platform's own audience defines who sees ads.
- Check exclusions. Add a "did not purchase" condition with a short window so recent buyers fall off the list at the next refresh.
- Import the spend. In All-platform analytics › Business › Channel ROI, open the Costs tab and import the campaign's daily spend by channel and campaign. The ROI tab then shows cost per acquisition (spend ÷ purchases on the order ledger) and return on ad spend (revenue ÷ spend) grouped by campaign. Those are the platform-style numbers; the incremental view needs your holdout comparison.
- Compare the groups. Record who was held out as a property or a separate segment, then compare conversion between the groups with Analyze on each segment. TapCub does not randomize the audience or split ad delivery.
- Hand the list to the ad platform. TapCub does not run ads or build audiences inside an ad platform. If a webhook is configured, it emits a membership change when a segment refreshes, which is the hand-off point for whatever tool you use to build the audience.
See also: Holdout group: definition, how big it should be and a worked example.
Sources
- Google Ads Help. About remarketing. Re-engaging people who have previously interacted with your brand.
- Google Ads Help. About your data segments. Segments need more than 100 active users to be eligible to run ads.
- Google Ads Help. About frequency capping. Limiting how often the same person sees Display ads.
- Randall A. Lewis and Justin M. Rao. The Unfavorable Economics of Measuring the Returns to Advertising. Quarterly Journal of Economics 130(4), 2015.
- Apple Developer. App Tracking Transparency. Permission for tracking and access to the advertising identifier.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Product Experience Specialist
Explores usability, user journeys, and feature adoption through practical examples of onboarding and everyday product use.