Benchmarks
Original studies from real DTC ad accounts, not survey data and not scraped filings. Every figure is computed from the same reporting warehouse Interconnections runs client accounts on, aggregated so no advertiser is identifiable.
The Meta Creative Fatigue Benchmark
A study of 368 Meta creatives across 15 DTC brands, 48.5M impressions and $685,791 in spend. Most creatives do not fatigue. They die of not working, and the top 10% carry 68% of spend.
Read the studyThe DTC Repeat Purchase Benchmark
A study of 468,386 orders and $69,656,336 of revenue across 17 DTC brands. Repeat revenue is inherited from earlier acquisition rather than produced this year, and repeat rate varies 5.6 times between comparable brands.
Read the studyThe Entry Product Benchmark
A study of 139,121 first orders across 17 DTC brands and 8,403 products. The product a customer buys first predicts whether they come back, and it holds out of sample on months that played no part in the ranking.
Read the studyHow these are built
Measured, not surveyed
Every figure is computed from the same reporting warehouse Interconnections runs client accounts on. Nothing here is a survey response, a vendor estimate, or a number aggregated from other publishers. That is the whole reason these studies sometimes disagree with the consensus.
No advertiser is identifiable
Every published figure pools at least five brands. Results are reported as per-brand medians alongside the pooled value, and each unit is indexed to its own baseline, so no figure reveals any advertiser’s costs, spend, or performance.
The window is stated, always
Each study names its measurement window, its sample size, and its unit of analysis. A benchmark without those three things is an anecdote with a percentage sign, and Interconnections will not publish one.
What the data cannot show is stated too
The creative fatigue study says plainly that the widely quoted frequency thresholds of 2.5 and 3.5 have no primary source we could trace, and that this dataset cannot test them because it measures impressions rather than reach. The repeat purchase study declines to rank categories or make any claim beyond six months, and says why.
Questions about the data
- Where does the data in these benchmarks come from?
- From client ad accounts and store data, computed inside the same reporting warehouse Interconnections uses to run those accounts day to day. It is measured rather than surveyed, modelled, or aggregated from other publishers, which is why some findings contradict numbers the industry repeats without a source.
- How do you publish this without exposing client data?
- Every figure Interconnections publishes pools at least five brands, is reported as a per-brand median alongside the pooled value, and is indexed to its own baseline. No absolute spend, cost, or revenue figure for any individual advertiser appears in any study.
- Why do these numbers disagree with other published benchmarks?
- Usually because the other number has no primary source. When Interconnections went looking for the origin of the widely quoted Meta frequency thresholds of 2.5 and 3.5, there was none to find. Where our data genuinely cannot settle a question, the study says so rather than filling the gap.
- Can I cite these studies?
- Yes. Cite the study by name with a link to the page, and please quote the sample size and measurement window alongside any figure, because a benchmark without them is misleading. If you need a clarification before publishing, email Interconnections and you will get a straight answer.
- How often are these updated?
- Each study states its own measurement window, and Interconnections re-runs them rather than quietly editing figures in place. When a study is refreshed the window moves and the numbers move with it, so a citation always points at a specific measured period.
- Do you publish findings that are inconvenient for your own services?
- Yes, and two of them are on this page. The creative fatigue study found that early click-through does not predict which creatives scale, which undercuts a common way agencies justify fast creative decisions. The repeat purchase study found retention is a multi-year lever rather than a quarterly one, which is not what a retention pitch wants to hear. Interconnections publishes them because a benchmark you only trust when it agrees with you is not a benchmark.
Related
Interconnections also publishes teardowns built entirely from public sources rather than client data. See all research and teardowns, including a breakdown of how Polymarket and Kalshi acquire users.