Ancaster Labs · experiment 01 · not a product yet
Lead with the finding. The offer comes later, if at all.
We pay to put our research where people already scroll: Threads, Instagram, Reddit, TikTok, X, sometimes ChatGPT. The post is the finding, complete on its own. Nobody is sold anything on the first touch. We measure whether a reader is still with us three months later, not what the click earned.
Experiment 01No spend yet
The lab results so far: every paid-traffic test, published whether it won or lost →
The path
- Touch 1A finding in the feedOne sourced number, complete on its own. Paid placement, labelled.No offer
- Touch 2The finding pageThe full research, a chart and a two-minute check. Ends with “get the next one”.No offer
- Touch 3They followOn the platform they came from, or The Receipt by email.No offer
- Touch 4+More findingsOne a week or so. A relevant offer may appear, labelled, only where it fits the finding.One offer, if it fits
- Day 90Judge the cohortDid they stay? Did they come back? What did they earn — as a side effect, not the goal?Measure, don’t push
Touch 1 · in the feed
ancaster.research Sponsored
>40%
of affiliate links on publisher pages had rotted. The pages still loaded. The money didn’t.
Source: impact.com / Trackonomics study of 7,000+ pages
The usual cause: a program changes networks and nobody tells the people linking to it. We traced how.
Read the finding · ancaster.link/f/link-rot
A mock-up of the first post; it has not run. Complete without the click. Labelled by the platform as an ad, and by us as research.
Touch 2 · the finding page
Your links load. That doesn’t mean they pay.
What the study measured, the three ways a link breaks without breaking, and a two-minute check anyone can run.
- What the study measured, and what it didn’t
- Three ways a link breaks without looking broken
- A two-minute check, with a free scan for the lazy version
- No offer, no sponsor on this page
Ends with: “Get the next finding” — follow on the platform you came from, or The Receipt by email.
The ledger · by cohort
Judged at day 90, not day 1.
| Finding × channel | Spend | Readers | Followed | Still here d30 | Per engaged | Offer touches |
|---|---|---|---|---|---|---|
| No spend yet — the ledger fills when the first cohort runs. | ||||||
“Per engaged” is spend ÷ readers who opened or returned at least twice in 30 days. Earnings are tracked per cohort as they arrive, but they don’t decide whether a finding runs again. Whether readers stay does.
House rules
- The post must be worth it alone
- If someone reads only the ad and never clicks, they should still have learned something true.
- No offer before touch 4
- Nothing is sold on the ad or the finding page. Offers come only after someone chose to follow.
- Only offers that allow it
- An offer appears only if its terms permit paid traffic to content, checked in the index. Ancaster’s own advertisers set their own rules.
- Labelled, always
- The platform labels the ad; we label the research as ours and any later offer as paid.
- Curiosity, not compulsion
- Target interests and communities. No late-night dayparting, no retargeting people who left.
- Stop what doesn’t keep readers
- Any finding × channel above $5 per engaged reader for two weeks is paused, and the reason is logged.