researchdata as of 2026-10-20
our own verdicts. misses included.
Cogny markets itself with Cogny. Our changes become tickets, and each ticket closes with a verdict. Here is the count for 2026, including the 19 that didn't work.
01the count
- 167measured changes to Cogny's own marketing, 2026-02-14 to 2026-10-14
- 7042%worked
- 7847%partial
- 1911%didn't work
- worked 70
- partial 78
- didn't work 19
- inconclusive 0
None closed as inconclusive. Tickets in this period were closed as worked, partial or didn't work; Cogny now closes a change as inconclusive when repeated checks can't tell.
02by kind of change
- AdsSearch, social and display campaigns, budgets, targeting, creative3922 worked · 11 partial · 6 didn't work · 0 inconclusive
- Site and SEOPages, content, structured data, indexing9233 worked · 49 partial · 10 didn't work · 0 inconclusive
- Tracking and measurementConversion tracking, UTM tags, pixels, audiences2410 worked · 12 partial · 2 didn't work · 0 inconclusive
- Email and communityNewsletters, list hygiene, community posts125 worked · 6 partial · 1 didn't work · 0 inconclusive
03what the misses look like
Most misses were new pages and articles that did not reach the rankings or clicks we expected within the window, ad changes that did not move cost or click-through the way the ticket said they would, and two tracking changes that did not hold. Partial usually means the change shipped and part of the expected outcome showed up, for example pages indexed but not yet ranking.
04method
- Source: the tickets in Cogny's own workspace, the one we use to market Cogny. Each ticket states its expected outcome before the change ships. Cogny measures it afterwards in the connected data (3 days later for ads, 7 days for content and SEO) and records a verdict when it closes.
- Counted: closed tickets with a recorded verdict that changed our own marketing: ads, the cogny.com site and its SEO, tracking and measurement, email and community posts.
- Not counted: changes to the Cogny product itself, bug reports, feature and support requests, research or analysis that changed nothing, tickets still being measured, and closed tickets that never got a verdict.
- Mapping: internal success, partial and failed are shown as worked, partial and didn't work.
- Read on 2026-10-20 from every closed ticket in the workspace, not a sample. Each ticket's kind of change was assigned by hand from what it changed.
- Anonymised: this is Cogny's own marketing, so no customer data is involved. Ticket titles are not published.
- Verdicts are Cogny's own assessment against each ticket's expected outcome, not a controlled experiment and not audited by a third party.