How the Score works

Every ranked influencer gets one Score from 0 to 100. It blends four parts, so a well-documented, distinctive character can outrank a slightly bigger but thinly documented one.

Score = 0.35 × Fame + 0.30 × Momentum + 0.20 × Originality + 0.15 × Data × 100

The number comes with a tier, so a score means something at a glance. Colour follows the value along the same spectrum everywhere on the site.

How big a character already is. Mostly audience size and how much posts usually get liked, plus the single biggest moment, whether they keep posting, how long they have been around and how recognisable they are. Inputs are scaled 0 to 1, weighted, and added up to a score out of 100.

How much attention is happening right now. Only the last 14 days count: recent likes (the lower of two averages, so a fading account cannot coast), the biggest recent post, posting rhythm, follower growth for new accounts, and how recently they posted.

Four analyst ratings from 0 to 5, added up and scaled to 100. The data files call this distinctiveness. It is the most subjective part of the Score, and it is labelled as inferred everywhere.

Five checks, each yes (1), partial (0.5) or no (0): is the identity clear, are there enough posts for Fame, are there recent dated posts, did we load the full history, and did we observe links to other characters. The average is the sufficiency score. Thin data pulls the Score down instead of being guessed.

Audience numbers span five orders of magnitude, so they are scored on a log scale: each step of 10× adds the same amount. Below the low end scores 0, above the high end scores 1.

How sure we are

Observed

Seen directly on Instagram: profile, captions, tags, counts.

Inferred

An analyst interpretation of observed material, or a rule applied to it (relationship types, phase, status).

Unknown

Not verifiable from Instagram, such as tickers reported only by press.

Momentum labels

Live status

From the latest captured post, never faked as realtime.

Every value from the research workbook, exactly as used by the scoring code.

• Avg/median likes & comments = the 12 most recent NON-PINNED posts. Posts whose like count is hidden by the owner (Instagram returns a placeholder of 3) are excluded from like averages and counted in 'Likes hidden'.
• Posts last 7d counts every post; posts last 14d / 30d exclude pinned posts and only count posts inside the analysed sample (lower bound when the sample is shorter than the window).
• 14-day window = posts dated on/after 2026-09-25 12:00 UTC.
• Engagement rate = (avg likes + avg comments) / followers. Engagement level: Very high >=5%, High >=2%, Medium >=0.5%, Low <0.5%.
• Reel view/play counts are NOT exposed to non-owners on Instagram web -> not captured (UNKNOWN). Likes are used as the reach proxy.
• Distinctiveness Score = (visual + personality + lore + cross-character) manual ratings (each 0-5, blue inputs) / 20 x 100. Ratings are analyst judgements based on the observed posts/bios.
• Recognizability rating (0-5, blue input) = analyst judgement: verified badge, press coverage found (non-IG, see SOURCES), and how often other characters reference the account.
• Data sufficiency = (YES=1, PARTIAL=0.5, NO=0) averaged over the five flags in DATA SUFFICIENCY.
• Rankings (TOP sheets) only include rows with Inclusion status = INCLUDED; ties are broken by row order.
• OBSERVED = seen directly on Instagram (profile, captions, tags, counts). INFERRED = analyst interpretation of observed material. UNKNOWN = not verifiable from Instagram.
• Followers/likes are point-in-time snapshots (2026-10-09, UTC+8) and change hourly for fast-growing accounts.

Where the numbers come from is listed on the sources page.