Research methodology

How we grade a directory — every point, attributable.

Every directory in the catalog gets a 0–100 quality score. It's deterministic: no AI, no black box — the same directory data always yields the same score, and every point is attributable to a named component. Here's the exact model. The live aggregate numbers are on the benchmark.

Five weighted components

The weights sum to exactly 100. Two things people conflate are kept separate: a directory's quality (is it worth listing on?) and its tier (how much human involvement a submission needs). A guided/manual directory can still be top quality.

Authority

0–45 pts

SEO/citation value of the backlink — domain rating, dofollow status, and traffic. This is the biggest weight because a citation from a weak site is worth little.

Accessibility

0–20 pts

How frictionless it is to actually get listed — free vs. paid, presence of a CAPTCHA, whether an account or email verification is required.

Reach

0–15 pts

How broad an audience the listing reaches — number of languages, market breadth (global vs. single-market), and category discoverability.

Automation

0–12 pts

How reliably our own agent can deliver a listing there — the execution tier (auto/assisted/guided) and the submission method.

Reliability

0–8 pts

Recent real-world success (failure streak) and how fresh our last verification of the directory is.

Grade bands

APrimescore 85–100
BStrongscore 70–84
CSolidscore 55–69
DSituationalscore 40–54
FLow-valuescore 0–39

The default auto-match launch excludes grade-F directories (score below 40) so we never auto-submit you to the bottom of the catalog — but if you deliberately pick one, we respect that choice.

Limitations we won't hide

  • Missing signals score conservatively, never optimistically. When we don't yet have verified domain-authority data, the score is estimated (and labelled as such), leaning low rather than flattering.
  • It's point-in-time. Directories change; a score reflects our last verification, and the reliability component decays as that check ages.
  • Reliability data is still small. Real-world success rates are built from our own dogfood submissions, which is a modest sample today and sharpens as more runs land.
  • A score is worth, not a promise. It estimates how much a listing is worth and how reliably we can deliver it — not whether a given directory will accept you.

Corrections

Because the score is a pure function of a directory's signals, a wrong score is fixed by fixing the signal (re-enriching the data or correcting a field) and re-running the scorer — it then recomputes deterministically. Spotted one that looks off? Our corrections process applies.