Local first, published deliberately.

I collect and analyse data locally with MaxX, my own local app. This website contains selected chart data, cached portraits from the original graphics, articles, and exported figures. Exploring it doesn’t call the X API.

The sample is my mutual network, not a representative sample of X. The charts are dated research editions, not live account metrics.

Followers per combined post.

The network snapshot was observed on 17 October 2026 at 20:47:34 UTC (22:47:34 in Prague). Of 804 mutuals, 628 had valid counts and at least 500 combined profile posts.

Ratio = followers ÷ combined total posts. The denominator includes original posts, replies, and reposts. Band means are unweighted means of individual ratios across every eligible account in the band. Medians use the same group. Rankings use the unrounded ratio, then follower count, then handle.

Bands use lower-inclusive, upper-exclusive boundaries: 0–999; 1,000–4,999; 5,000–9,999; 10,000–24,999; 25,000+. The histogram uses nine 0.25-wide intervals and a final 2.25+ overflow bucket. None of this measures follower conversion or proves posting causes growth.

What counts as viral here?

At least 1,000 observed likes. Replies and reposts are excluded; quote posts are included. The search covered all 804 saved mutuals with no 500-post minimum.

Post creation window: 17 September 2026, 23:42:45 UTC inclusive, to 17 October 2026, 23:42:45 UTC exclusive. In Prague: 18 September, 01:42:45 → 18 October, 01:42:45.

Public engagement metrics were observed on 18 October at approximately 00:00:39–00:02:06 UTC. They are cumulative at capture. Summed impressions are not unique viewers or the account’s total impressions during the window.

The count endpoint reported 356 matches; search returned 329 resources. That gap remains unresolved. Three returned posts were below the observed like threshold, leaving 326 qualifying saved posts from 58 authors. This is why the site describes saved hits rather than complete coverage.

Rank is qualifying-post count descending, then summed impressions descending. Changing chart height does not change the rank. Account size is from the frozen profile snapshot. No-hit accounts in the wider cohort aren’t proof that no qualifying post existed.

The 107-post content reading.

The content study removes whole accounts with unusually many saved qualifying posts. Across the 58 authors with hits, Q1 = 1 and Q3 = 4; Q3 + 1.5 × IQR = 8.5. Accounts with nine or more hits are excluded in this view.

Eight accounts and 219 posts are removed; 50 authors and 107 posts remain. The global cutoff is an analytical choice, not a claim that prolific accounts are invalid. Original charts remain unfiltered.

Themes are a qualitative reading of text and captions. Media contents were not reviewed. The retained under-1k band has only one post, and 25k+ has four. Those groups cannot support broad content claims. Non-viral posts and a full posting denominator were not collected, so success rates and causal writing advice cannot be estimated.

Inspect the published data.

The website publishes the aggregates and original chart leaders needed for these views. It does not distribute the full network membership or third-party post bodies.

Published chart data (JSON) ↓Definitions and source fingerprints (JSON) ↓

Maximum X is an independent personal project. It is not affiliated with, endorsed by, or sponsored by X.