Home/The Benchmark
The Launch Readiness Benchmark
Everybody approaching a launch wonders whether they are unusually unprepared or perfectly normal. This is the answer, drawn from every launch scored with the Launch Readiness Score, aggregated and stripped of anything identifying.
Collecting
0 of 100 qualifying assessments in
The first report publishes at 100 qualifying completions, not 100 raw ones. Publishing at thirty and calling it research is the sort of thing that damages a new publication permanently, so it waits.
What will be published
- Median and distribution of overall readiness scores
- Median score per dimension, and which dimensions move together
- The share of pre-launch products blocking at least one AI retrieval crawler in their own robots.txt, which is a verified fact here rather than a self-report
- How scores differ by company stage and by how close the launch is, which tests whether people improve as the date approaches or simply run out of time
- The individual checks failed most often, which is the most directly useful list of the lot
Methodology in short
Every figure comes from the same forty checks, each answered No, Partly or Yes and scored zero, one or two, with six of them verified against the respondent's live site where a URL was given. Dimension scores are the percentage of available points. The overall score is the weighted average. The full method, the weights and the reasoning are on the methodology page.
Exclusion rules, published in advance
A run is excluded from the figures if it was completed in under three minutes, if thirty six or more of the forty answers are identical, if it is a second run from the same source that day, or if it is flagged as a test. Runs reporting "already launched" are reported separately rather than mixed in. Every published figure will state its n.
What is stored
Scores, the answer pattern, the four profile answers, and the verified check evidence. IP addresses are hashed with a daily rotating salt for rate limiting and same-day deduplication only. Email addresses, where somebody unlocked a full report, live in a separate file and never enter this dataset. The privacy page has the detail.
The limitations, stated in advance rather than in a footnote later
This is largely self-reported data from a self-selected group, and neither problem is fixed by a larger sample. Founders who seek out a launch readiness assessment are probably more diligent than average, which biases scores upward. People assessing their own work are generally generous, which also biases upward. Both push the same direction, so the fair reading is that these numbers represent a ceiling rather than an average. Moving six checks to live verification was the main thing done about it.