Preliminary result
In the Lion's Forge aggregate snapshot generated on 6 August 2026 at 23:30 UTC, the latest valid complete-profile score for each of 23 users had:
This is an early product dataset, not a representative sample of everyone using Tinder, Hinge or Bumble. It should not be presented as the average dating profile score for the general population.
Profile-score distribution
| Score band | Latest scores | Share of 23 users |
| --- | --- | --- |
| 0-39 | 5 | 21.7% |
| 40-59 | 7 | 30.4% |
| 60-79 | 4 | 17.4% |
| 80-100 | 7 | 30.4% |
The median is the more stable summary for this small dataset because the minimum score of 1 pulls the mean downward. Neither number is a percentile or an outcome prediction.
What a Lion's Forge score is
Lion's Forge uses consistent 0–100 model scores to help you compare your own photos, profile drafts and conversations. A score is decision support: it summarizes the model's assessment and points you toward the accompanying feedback. It is not a percentile, attractiveness fact, or prediction of matches, replies or dates.
Why we removed the early benchmark figures
An earlier version of this page published preliminary averages without the exact sample count, deduplication rules and reproducible query behind each number. Even with caveats, that was not enough context for readers to judge the evidence. We removed those figures rather than present an incomplete internal snapshot as a public benchmark.
That correction matters. Photo analyses, profile scoring events, Face Card runs and conversation analyses use different units. One person may run a tool more than once, failed inputs may need exclusion, and a model's "potential" estimate is not an observed before-and-after result. Combining those outputs without a clear protocol can create a precise-looking but misleading conclusion.
Reproducible inclusion and deduplication rules
The snapshot was produced by a separate aggregate-only reporting function. It does not change the scoring model, stored scores, percentile display or score-boost mechanism.
Why other score categories are not headline benchmarks yet
The same aggregate check found 45 Face Card analyses from only 2 unique users and 64 conversation analyses from 13 unique users. Those samples are too small and concentrated for responsible headline averages, so their means are not published as claims about typical users.
How to interpret your own result
Use the score to compare like with like:
A higher Lion's Forge score does not guarantee a better dating outcome. Location, preferences, app activity, timing and other people's choices all matter.
Our publication standard for future benchmarks
We will only publish a new aggregate benchmark when it includes the snapshot date, exact analysis count, unique-user count, inclusion and exclusion rules, treatment of repeated runs, score distribution, median and mean, and a plain-language limitation statement. Any outcome claim would additionally require opt-in match or reply data and a suitable comparison design.
This August snapshot meets that disclosure standard for complete-profile scores. It does not meet the higher standard required for an outcome claim. Among 23 unique users in this early dataset, the defensible conclusion is limited to a median latest model score of 55. You can review and compare your own complete profile without treating the result as a promise.