Library status
We publish the thresholds the tactic library must clear and where it stands against them, so you see the library as it is rather than a marketing claim. The values in the "Now" column are read from the database when this page loads, and nobody edits them. A count of zero is shown as 0. A ratio or percentile that does not yet have enough data behind it gives the reason instead of a number.
| Criterion | Launch bar | Six-month bar | Now |
|---|---|---|---|
| Published tactics | ≥ 30, each with ≥ 2 evidence sources | ≥ 60 | 6 |
| Approved examples per published tactic (fewest) | ≥ 5 with crops | ≥ 15 | 3 |
| Corpus sites captured and annotated | ≥ 100 | ≥ 300 | 15 |
| Golden-set precision per published tactic (lowest) | ≥ 0.85 | ≥ 0.9 | Not measured: hand-labelling of the golden set has not started, so no tactic has the 20 labelled presences a figure needs. |
| Two-pass agreement, κ per published tactic over its last 200 pairs (lowest) | κ ≥ 0.8 | κ ≥ 0.85 | κ 0.107; 2 of 6 tactics have too few pairs for a figure |
| Page audit wall time, 95th percentile | ≤ 180 s | ≤ 120 s | Fewer than 20 audits have finished, so a 95th percentile would not mean anything yet. |
| Model cost per page audit, 95th percentile | ≤ £1.50 | ≤ £1.50 | Not measured: customer audits make no model calls today (detectors only), so there is no per-audit model cost to report. |
| Published tactics graded tested or better | ≥ 10 | ≥ 25 | 6 |
| Published tactics graded measured | None required; the count is shown | ≥ 5 | 0 |
Read from the database at 2026-10-15 13:17 UTC, when this page loaded.
The launch bar is what the library must meet before it is sold; the six-month bar is where it must be six months later.
By page type
Tactics drafted and published for each page type, and corpus pages of that type captured and annotated. Tactics that are not tied to a page type are counted under "All page types", against every annotated corpus page.
| Page type | Tactics in draft | Tactics published | Corpus pages captured and annotated |
|---|---|---|---|
| Homepage | 384 | 5 | 7 |
| Pricing | 224 | 4 | 12 |
| Signup | 254 | 2 | 0 |
| Demo | 83 | 0 | 4 |
| Contact | 105 | 0 | 4 |
| Product | 503 | 4 | 5 |
| Customers | 38 | 0 | 4 |
| About | 49 | 1 | 5 |
| Comparison | 48 | 1 | 3 |
| Landing | 327 | 2 | 0 |
| Other | 366 | 2 | 0 |
| All page types | 0 | 0 | 44 |
What the criteria mean
- Published tactics counts tactics that passed the publish rule. Drafts do not count.
- Approved examples are cropped screenshots of the tactic on real sites that passed annotation review. The figure is the lowest count across published tactics.
- Golden-set precision is how often the system is right when it says a tactic is present, checked against hand-labelled crops.
- Two-pass agreement is Cohen's kappa (κ) between two independent annotation passes (different model and prompt) over the last 200 observations, with positive agreement (agreement on the cases where either pass saw the tactic) reported alongside. It is agreement between two models, not accuracy. It is never raw agreement: most crops show neither pass finding a given tactic, and those both-absent pairs push raw agreement up whatever the passes actually do (in our pilot, raw 0.974 against kappa 0.838).
- Page audit wall time is the time within which 95 in 100 page audits finish, queue included.
- Model cost per page audit is what the model calls behind one page audit cost, priced at published rates.
- Tested or better and measured count tactics by evidence grade.
What happens when a tactic slips
Any tactic whose golden-set precision falls below 0.85, or whose two-pass kappa falls below 0.75 over its last 200 observations, is demoted to draft automatically and stops being returned to agents. A figure counts only once there is enough data behind it: at least 20 labelled presences for precision, and at least 50 two-pass pairs including 5 where a pass saw the tactic for kappa. Below that, the table above says why there is no figure.
Every tool response carries tactic_count and corpus_version, so your agent always sees the current size of the library.