Does duplicating an ad campaign lower its click-through rate?
Does running two identical campaigns aimed at the same audience on Facebook and Instagram lower the click-through rate? The researchers estimated a relative reduction of about 28% compared with a single campaign if the campaigns reached entirely the same people. Showing the same ad more often may produce fewer clicks per impression.
ACADEMIC WORK REVIEWED
Digital Media Mergers: Theory and Application to Facebook-Instagram
ResearchersJustin Katz, Hunt Allcott
NBER · July 2025 · May 2026 revision, not peer-reviewed · Working paper · Source language: English
English review: 13 November 2026 · The source date appears in the citation above.
Editorial review draft
The source has been checked; the review still carries its original editorial status. DijitalPi did not conduct a customer experiment for this paper.
LET'S READ THE RESEARCH TOGETHER
We first explain the researchers' question, method and findings. We then discuss how to interpret the results, clearly separating DijitalPi's commentary from the source.
01 / WHAT DID THE RESEARCHERS WANT TO UNDERSTAND?
How does targeting the same audience with two identical campaigns instead of one change the click-through rate?
Katz and Allcott wanted to measure whether duplicating a campaign reduces the clicks obtained per impression. The experiment informed their broader study of what happens when Facebook and Instagram cannot coordinate advertising delivery.
Explaining the context · DijitalPi commentary
DijitalPi commentary: Running more campaigns does not establish that every impression produces clicks at the same rate. The experiment suggests that repeat exposure to the same people deserves attention when assessing identical campaigns.
02 / HOW WAS THE RESEARCH CONDUCTED?
How did they test the question?
The researchers created 120 separate audiences for 15 products across five categories. They randomly assigned audiences to receive either a single campaign or two identical campaigns. Ads came from product recommendation pages created by the authors, rather than the brands themselves. Follow-up campaigns ran on Facebook and Instagram from 24 February to 3 March 2025. The primary analysis includes 174 campaigns.
Understanding the method · DijitalPi commentary
Each product had eight separate audiences. The researchers varied budget and objective alongside campaign count. They recorded click-through rate, impressions and reach. Reach here means the share of the initial audience shown a follow-up ad. Frequency means impressions per user.
They calculated overlap using Meta’s combined unique reach estimates. They scaled the difference between the groups’ average logarithmic click-through rates by the average duplicated share to estimate full duplication. A logarithm is the mathematical transformation used here to calculate relative change. The primary estimate accounts for product, budget, objective, frequency and reach.
03 / RESEARCH FINDINGS
What does the estimate scaled to full duplication show?
The primary sample contains 174 follow-up campaigns. The average duplicated share was 0.855: 85.5% of impressions in the average duplicated campaign went to people also exposed to the other campaign. The primary coefficient was -0.335. The authors’ calculation, 1-exp(-0.335), implies an average click-through rate about 28% lower in a fully duplicated campaign than in a single campaign, a relative difference. This is neither a 28-percentage-point decline nor the directly observed average gap between experimental groups. Average frequency was 4.07 impressions per user per week, and the average click-through rate was 1.4%. That rate is not specific to the control group. Alternative data treatments produced similar duplication estimates.
Click-through rate under full duplication
About 28% lowerRelative to a single campaign, scaled from partial overlap to full duplication.
Average duplicated share
0.855 (85.5%)The share of impressions in the average duplicated campaign going to users also exposed to the other campaign.
DijitalPi commentary: Examine reach, frequency and click-through rate together when campaigns target the same audience. The estimate for full duplication is not an expected loss rate for every advertising account.
DijitalPi commentary: Click-through rate alone does not establish sales performance. This experiment cannot calculate the sales effect of merging or stopping campaigns.
05 / CONCLUSION AND OPEN QUESTIONS
What did we learn, and what do we still not know?
This is an NBER working paper that has not been peer-reviewed. The advertising outcome was click-through rate; sales, conversions and brand effects were not measured. Ads came from the authors’ product recommendation pages. Products and users were not randomly sampled from the entire market. Working outside Meta, the authors could not implement their ideal experiment by directly randomising each user’s exposure count. Both authors disclose previous employment at Microsoft Research. Allcott previously consulted on social media matters, unrelated to FTC v. Meta. He was also an unpaid researcher on Meta’s Facebook and Instagram Election Study. The authors report no consulting work or conflict of interest related to this paper’s topic. Katz received support from an ICMM pre-doctoral fellowship awarded through NBER.
Our open question at DijitalPi: how much do your identical campaigns reach the same people, and do you have the data to assess click and sales outcomes separately?
We created these scenarios to make the topic concrete. They are not cases from the paper or measured client results.
EXAMPLE 01
E-commerce: identical product ads
An illustrative coffee retailer is considering identical campaigns using the same product image and audience.
Does copying a campaign make advertising more efficient?
Open the recommendation for example 01+
Assess extra impressions separately from click-through rate
DijitalPi commentary: Record the product, objective, budget and target audience together. Examine how click-through rate changes as impressions increase. Comparing campaign reports alone does not establish the effect of duplication. This is an illustrative example, not a measured retailer outcome from the study.
EXAMPLE 02
Tourism: separate repetition from bookings
An illustrative hotel is assessing campaigns that show the same room ad to the same audience again.
If the click-through rate falls, do bookings fall by the same amount?
Open the recommendation for example 02+
Measure booking outcomes separately
DijitalPi commentary: Report completed bookings separately from click-through rate. The experiment did not measure bookings, sales or brand effects. Its click finding cannot be converted into an expected revenue loss for the hotel. This proposal has not been tested at the hotel.
EXAMPLE 03
Agency: examine campaigns aimed at the same audience
An illustrative agency wants to know whether its client’s identical campaigns repeatedly reach the same people.
Does selecting the same target audience guarantee full overlap?
Open the recommendation for example 03+
Distinguish the target audience from the people actually reached
DijitalPi commentary: Examine reach and frequency alongside campaign settings. If available reports do not identify shared users, do not assume the overlap is known. The researchers used Meta’s combined unique reach estimates for that calculation. This illustrative review plan is not a replication of their experiment.
TRY IT WITH YOUR TEAM
How can you assess identical campaigns?
DijitalPi commentary: We suggest recording product, budget, objective, reach, frequency and click-through rate together for identical campaigns aimed at the same audience. State when overlap data are unavailable. Measure sales separately. DijitalPi has not run a client experiment for this study. This proposal has not yet been implemented.
The academic work belongs to the researchers named above. This page contains DijitalPi's explanatory review and original business examples; it is not a full translation of the paper.
Cookie preferences