Retargeting：真实贡献与被归功的成果
Retargeting is a practitioner term for showing ads to people who have already interacted with a brand, usually by visiting its website or appearing on its customer list. The useful question is not whether those people buy afterward, but how many would not have bought without the ad.
That second sentence is where the confusion starts. A retargeted audience is, by design, made of people who already showed interest, and some of them were always going to come back. When one of them clicks and then buys, the report looks excellent. Field experiments that compare exposed and unexposed groups tell a more modest story. Retargeting can add real value. But the lift it causes is often smaller than the credit it receives, and the part that works depends heavily on what the ad offers and when it appears.
Retargeting List Rules: Minimum Size and Membership Duration
How many people a retargeting list must contain and how long each person can stay on it, as stated in platform documentation as of November 2026.
| Platform and list type | Rule | Stated value | Note | Source |
|---|---|---|---|---|
| Google Ads audiences (Display Network, Search Network, YouTube) | Minimum size to serve | At least 100 active visitors or users within the last 30 days | A floor, not a target. A list that barely clears 100 people is eligible but makes later measurement noisy. | Google Ads Help, as of November 2026 |
| Google Ads customer lists | Minimum size to serve | Same 100 threshold; lists uploaded before February 1, 2024 still need 1,000 matched users | The older minimum applies only to lists uploaded before that date. | Google Ads Help, as of November 2026 |
| Google Ads Display and Search lists | Membership duration | Default 30 days; maximum 540 days | A longer window is not automatically better if most of the early effect falls in the first days after a visit. | Google Ads Help, as of November 2026 |
| Google Ads Customer Match | Maximum membership duration | 540 days, rolling out gradually starting April 7, 2025 | Announced in February 2025. | Google Ads Developers Blog (February 12, 2025) |
| Meta Website Custom Audiences | Retention range | 1 to 180 days in one section; maximum of 365 days in another section of the same page | The same Marketing API guide gives two different figures. | Meta Marketing API documentation, as of November 2026 |
How Retargeting Works
Every retargeting campaign starts with a list. Website visitors, or contacts on an uploaded customer list, are placed into an audience, and the platform shows ads to that audience across its network. Two rules shape what the list can do: how many people it must contain, and how long each person stays on it.
Who Gets Added to a List
On Google Ads, the help documentation states that an audience needs at least 100 active visitors or users within the last 30 days to serve on the Display Network, the Search Network, or YouTube (as of November 2026). Customer lists follow the same threshold, except that lists uploaded before February 1, 2024 still need the older minimum of 1,000 matched users.
These are floors, not targets. A list that barely clears 100 people is eligible, but it also makes later measurement noisy.
How Long Someone Stays on a List
Google's documentation lists a default membership of 30 days and a maximum of 540 days for both Display and Search lists. For Customer Match, the Google Ads Developers Blog announced in February 2025 that a 540-day maximum would roll out gradually starting April 7, 2025. Meta's official documentation, as of November 2026, gives a retention range of 1 to 180 days in one section of its Marketing API guide and a maximum of 365 days in another section of the same page.
A longer window is not automatically better. The timing research below found much of the early effect concentrated in the first days after a visit. If that holds in your category, a very long list mostly reaches people who have forgotten the visit.
What Each Retargeting Study Actually Measured
Return visits, conversions, purchases and profitability are different outcomes, so each figure below only supports a claim about the metric its study counted.
| Outcome type | What was counted | Reported result | How to read it | Source |
|---|---|---|---|---|
| Return visits | Users returning to the website within four weeks, across more than 230,000 randomized BuildDirect site visitors | 14.6% more users returned; 33% of the first week's advertising effect occurred on the first day | Shows retargeting brings people back, not that it sells to them. The data is several years old, so it says more about direction and timing than about current benchmarks. | Sahni, Narayanan & Kalyanam (2019), Journal of Marketing Research |
| Visitors and visits | Site visitors, total site visits, transactions and sales for one online retailer | Site visitors +26.6%, total site visits +17.2%, transactions +12.0%, sales +10.8% | Visitors count people and visits count visits, so the first two figures are different metrics. Conference draft, single retailer. | Johnson, Lewis & Nubbemeyer (2016), Ghost Ads, NBER conference draft |
| Conversions | Baseline conversions across more than 2,000 advertisers using Chrome traffic | Retargeting raised baseline conversions by 4.6% | A positive effect, but far smaller than the multiples in vendor roundups. Preprint, not peer-reviewed; countries not specified. | Kobayashi, Johnson & Gu (2024), SSRN preprint |
| Conversions | Conversions when retargeting was combined with the right placement and message, in a field experiment with more than one million users and a European automaker | Lifted conversions by as much as 17% for some product categories | An upper bound for certain categories, not an average. The summary does not say which action counted as a conversion. | Valenti, Miller & Tucker (2025), Management Science, via IESE Business School summary |
| Purchases | Purchase rate among shoppers who left without buying | Free shipping +18.6%; free returns +31.6%; simple product reminders showed no significant increase | Relative increases, not percentage-point changes. The published abstract does not name the country. | Luo, Wang, Wu & Moriguchi (2025), Manufacturing & Service Operations Management (MSOM) |
| Profitability | Targeting profitability, as defined by the authors | Free shipping +28.2%; free returns +37.8% | Keep the authors' metric as written. Targeting profitability is not return on ad spend or company profit. | Luo, Wang, Wu & Moriguchi (2025), Manufacturing & Service Operations Management (MSOM) |
| Purchases | Purchases after cart abandonment, by how soon the cart retargeting ad arrived | Ads within 30 minutes to one hour: less likely to purchase than no ad. Ads one to three days later: positive incremental effect. | Measures cart reminders and purchases, not banner ads and return visits. | Li, Luo, Lu & Moriguchi, Journal of Marketing (online November 2020), via AMA release |
| Returns vs. prospecting | Returns from retargeting compared with prospecting on Google and Facebook, in a geography-based test with more than US$5.7 million in ad spend | Prospecting produced large and consistent returns on both networks; retargeting returned significantly less | One large controlled test in a long buying cycle category (mattresses). It does not prove retargeting fails everywhere. | Larson & Dotson (2026), Journal of Interactive Marketing, via American Marketing Association summary |
| Enrollment | SNAP enrollment after retargeted Facebook and Instagram outreach in Los Angeles County, among people not previously enrolled | Estimated effect of -0.03 percentage points (standard error 0.20); no statistically significant increase | A public program, not a store, but being reached again did not produce action. | Rogers (2024), working paper manuscript, Kelley School of Business |
What Field Experiments Say Retargeting Actually Adds
Platform reports measure what happened after an ad. Experiments measure what happened because of it, using a group that qualified for the ads but was held back.
Return Visits Rise, and the Effect Is Front-Loaded
One widely cited experiment ran with BuildDirect, a Canada-based online seller of home-improvement products whose customers are mostly in the United States. Navdeep Sahni, Sridhar Narayanan, and Kirthi Kalyanam randomized retargeting across more than 230,000 site visitors and published the results in the Journal of Marketing Research in 2019. Switching on retargeting caused 14.6% more users to return to the website within four weeks, and 33% of the first week's advertising effect occurred on the first day.
The outcome was a return visit, not a purchase, so the figure says retargeting brings people back, not that it sells to them. The data is also several years old, which makes it better evidence of direction and timing than a current benchmark.
When Prospecting Beat Retargeting
Jeffrey Larson and Jeffrey Dotson ran a geography-based field experiment on Google and Facebook for a major bed-in-a-box mattress brand, with more than US$5.7 million in ad spend, published in the Journal of Interactive Marketing in 2026. According to the American Marketing Association's summary, prospecting ads aimed at new audiences produced large and consistent returns on both networks, while retargeting returned significantly less.
Mattresses have a long buying cycle, so a final retargeting click can collect credit for a decision already made. The result does not prove retargeting fails everywhere; it shows that in one large, controlled test, the money worked harder elsewhere.
A Cross-Advertiser Estimate
A 2024 paper by Shunto Kobayashi, Garrett Johnson, and Zhengrong Gu of Boston University, posted on SSRN as a preprint and not peer-reviewed, ran an industry-wide experiment across more than 2,000 advertisers using Chrome traffic, with countries not specified. Retargeting raised advertisers' baseline conversions by 4.6%. That is a positive effect, and far from the dramatic multiples that circulate in vendor roundups.
A 2016 conference draft of the Ghost Ads paper by Garrett Johnson, Randall Lewis, and Elmar Nubbemeyer, which aimed to make ad experiments cheaper to run, reported results for one online retailer: site visitors rose 26.6%, total site visits rose 17.2%, transactions rose 12.0%, and sales rose 10.8%. The first two are different metrics, one counting people and the other counting visits.
Why Platform Reports Overstate Retargeting
Retargeting audiences are pre-selected for intent, so they are easy to convert and hard to credit fairly. In the mattress study, the authors argue that platform attribution overstates retargeting because it credits the final retargeting click for purchases people were already planning, a problem that grows with long purchase cycles.
The metric can flatter too. More return visits are easy to retell as more sales. Before repeating any retargeting figure, ask what was counted: visits, conversions, purchases, or profit.
A working paper manuscript by Parker Rogers of Indiana University's Kelley School of Business, dated September 2024, tested retargeted Facebook and Instagram outreach encouraging SNAP food assistance enrollment in Los Angeles County. Among people not previously enrolled, the estimated effect was negative 0.03 percentage points, with a standard error of 0.20 points, and no statistically significant increase. It is a public program, not a store, but being reached again did not produce action.
What Makes a Retargeting Ad Earn Its Lift
If retargeting often takes credit it did not earn, the next question is when it does earn it. The evidence points to two levers: what the ad gives people a reason to do, and when it reaches them.
An Offer Beats a Reminder
A field experiment by Xueming Luo, Le Wang, Qingsheng Wu, and Takeshi Moriguchi, published in Manufacturing & Service Operations Management in 2025, tested retargeting for shoppers who left without buying. Simple product reminders did not significantly increase purchases. Offers made late in the shopping process did. Free shipping raised the purchase rate by 18.6% and targeting profitability by 28.2%. Free returns raised them by 31.6% and 37.8%. These are relative increases, not percentage-point changes, and the published abstract does not name the country where the experiment ran.
A shopper who left already knows the product exists. Free shipping or free returns removes a cost or a risk that may have stopped them, which changes the decision instead of repeating it. Keep the authors' metric as written: targeting profitability is not return on ad spend or company profit.
Timing: Early Helps, Too Early Backfires
A study of cart retargeting complicates the first-day finding. In research by Jing Li, Xueming Luo, Xianghua Lu, and Takeshi Moriguchi in the Journal of Marketing, published online in November 2020, customers who received a cart retargeting ad within 30 minutes to one hour of abandoning their cart were less likely to purchase than customers who received no ad, according to the AMA's release. Ads sent one to three days later had a positive incremental effect.
The findings need not conflict. One measured banner ads and return visits; the other measured cart reminders and purchases. Together they suggest a window: soon enough that interest has not faded, not so soon that the ad lands on someone still deciding. That window will differ by product and channel, so test it.
Dynamic vs. Generic, and Context
Dynamic retargeting shows the exact product a person viewed, Anja Lambrecht and Catherine Tucker found otherwise in a 2013 Journal of Marketing Research study: on average, dynamic retargeted ads were less effective than their generic equivalents.
Context can help. In a 2025 Management Science paper by Albert Valenti, Chadwick Miller, and Catherine Tucker, based on a field experiment with more than one million online users and a European automaker, combining retargeting with the right placement and message lifted conversions by as much as 17% for some product categories, according to IESE Business School's summary. That is an upper bound for certain categories, not an average, and the summary does not specify which action counted as a conversion.
Retargeting vs. Remarketing vs. Prospecting
Prospecting reaches people who have not yet interacted with a brand, while retargeting reaches people who have. How remarketing relates to retargeting, along with campaign types and worked examples, is covered in the ad retargeting entry, so this page stays on a narrower question: how much retargeting changes the outcome.
Privacy, Cookies, and Consent
Google announced on April 22, 2025 that Chrome would keep its existing approach of giving users a choice about third-party cookies and would not roll out a new standalone prompt. Users can set their preference in Chrome's privacy and security settings, and Incognito mode blocks third-party cookies by default. Treat that as the stated position as of November 2026.
The cross-advertiser preprint also tested Google's Privacy Sandbox as a replacement. Adjusted for ad spend, Sandbox reached 86.4% of traditional retargeting's clicks per dollar and 81.8% of its click-through conversions per dollar. It worked, but less efficiently.
In Indonesia, Law No. 27 of 2022 on Personal Data Protection was enacted and took effect on October 17, 2022. It governs the processing of personal data, including the rights of data subjects and the obligations of data controllers. This entry does not interpret specific provisions.
A survey of 154 Generation Z e-commerce users in Greater Jakarta (Jabodetabek), published by Nada Jauza Adinda Dara and Willy Arafah in Bisnis-Net Jurnal Ekonomi dan Bisnis in December 2025, used structural equation modeling. Brand ethics, ad effectiveness, and ad relevance were positively associated with attitudes toward retargeting ads, while privacy concern showed no significant effect. It shows associations in one sample, not causes.
How to Check Whether Your Retargeting Is Incremental
Every study above compared people who could see retargeting ads with similar people who could not. The same logic works in a single account.
Name the outcome first, and keep it fixed. Return visits, purchase rate, and profitability are different measures, and mixing them produces numbers that cannot be compared.
Then hold some eligible people back, by individual user as in the BuildDirect study or by region as in the mattress study. The held-back group shows what would have happened anyway. The gap between groups is the lift; the platform's attributed conversions are not.
Finally, test the levers the evidence points to: a plain reminder against an offer that removes a cost or risk, an ad within the hour against one a day or more later, a dynamic product ad against a generic message. Judge each by the conversion rate in the test and control groups, and weigh it against cost. A small, real lift can be worth paying for. A large attributed number with no measurable lift is not.
Frequently Asked Questions
Does retargeting actually increase sales?
Sometimes, and usually by less than platform reports suggest. A preprint covering more than 2,000 advertisers estimated a 4.6% rise in baseline conversions, while a 2026 mattress-brand experiment found retargeting returned significantly less than prospecting.
How soon after a visit should a retargeting ad appear?
Early, but not instantly. In a 2019 experiment, 33% of the first week's effect came on the first day. A cart retargeting study found ads within 30 minutes to one hour reduced purchases, while ads one to three days later helped.
Are dynamic product ads better than generic retargeting ads?
Not by default. A 2013 study found dynamic retargeted ads were, on average, less effective than generic versions.
How many people does a Google Ads retargeting list need?
As of November 2026, at least 100 active visitors or users in the last 30 days for Display, Search, and YouTube. Customer lists uploaded before February 1, 2024 still need 1,000 matched users.
How long can someone stay on a retargeting list?
On Google Ads, the default is 30 days and the maximum is 540 days. Meta's documentation, as of November 2026, gives 180 days in one section and a maximum of 365 days in another.
Does Chrome still allow third-party cookies for retargeting?
Per Google's April 22, 2025 announcement, Chrome kept letting users choose, without a new standalone prompt. Check for newer announcements before relying on this.
Related Terms
- Behavioral Targeting
- First-Party Data
- Google Ads
- Display Advertising
- Conversion Rate Optimization (CRO)
Sources
- VERIFIED. Sahni, Narayanan, and Kalyanam (2019), Journal of Marketing Research 56(3). Stanford GSB publication record. https://www.gsb.stanford.edu/faculty-research/publications/experimental-investigation-effects-retargeted-advertising-role.
- REPORTED. Larson and Dotson (2026), Prospecting vs. Retargeting, Journal of Interactive Marketing. American Marketing Association summary. https://www.ama.org/research-insights/research-insight-ad-retargeting-is-overrated/.
- VERIFIED, preprint, not peer-reviewed. Kobayashi, Johnson, and Gu (2024), Privacy-Enhanced versus Traditional Retargeting. SSRN. https://papers.ssrn.com/abstract=4972368.
- VERIFIED, conference draft. Johnson, Lewis, and Nubbemeyer (2016), Ghost Ads. NBER conference draft. https://conference.nber.org/conf_papers/f84173.pdf.
- VERIFIED, working paper manuscript. Rogers (2024), Retargeting Social Media Outreach for SNAP Take-up. https://parkerrogers.github.io/Papers/Rogers_SNAP_MS_v3.pdf.
- VERIFIED. Luo, Wang, Wu, and Moriguchi (2025), From Stopping to Shopping, Manufacturing & Service Operations Management 27(4). https://ideas.repec.org/a/inm/ormsom/v27y2025i4p1224-1241.html.
- REPORTED. Li, Luo, Lu, and Moriguchi, cart retargeting study, Journal of Marketing (online November 2020, issue 2021). AMA release and EurekAlert. https://www.ama.org/2020/11/12/boosting-returns-on-e-commerce-retargeting-campaigns/ and https://www.eurekalert.org/news-releases/754706.
- VERIFIED. Lambrecht and Tucker (2013), When Does Retargeting Work?, Journal of Marketing Research. MIT DSpace. https://dspace.mit.edu/handle/1721.1/88160.
- REPORTED. Valenti, Miller, and Tucker (2025), Management Science. IESE Business School summary, February 19, 2025. https://www.iese.edu/insight/articles/audience-placement-message-personalizing-marketing-journey/.
- VERIFIED, platform documentation, as of November 2026. Google Ads Help, audience size and membership duration. https://support.google.com/google-ads/answer/7558048?hl=en and https://support.google.com/google-ads/answer/2472738.
- VERIFIED. Google Ads Developers Blog (February 12, 2025), Customer Match membership duration. https://ads-developers.googleblog.com/2025/02/update-to-customer-match-membership.html.
- VERIFIED, platform documentation, as of November 2026. Meta Marketing API, Website Custom Audiences. https://developers.facebook.com/docs/marketing-api/audiences/guides/website-custom-audiences.
- VERIFIED. Google Privacy Sandbox blog (April 22, 2025), next steps for Chrome. https://privacysandbox.google.com/blog/privacy-sandbox-next-steps.
- VERIFIED. Law No. 27 of 2022 on Personal Data Protection. JDIH BPK. https://peraturan.bpk.go.id/Details/229798/uu-no-27-tahun-2022.
- VERIFIED, correlational survey. Dara and Arafah (2025), Bisnis-Net Jurnal Ekonomi dan Bisnis 8(2). https://jurnal.dharmawangsa.ac.id/index.php/bisnet/article/view/7110.
