Read with care: the State of X study is n=1 account, 983 replies

The stat. Every headline in State of X Engagement 2026 — the 5x warm/cold gap, the twin peaks, the 80% ambient attribution, the bookmark weighting — comes from one account's real operational logs: roughly 983 replies and 224 attributed new followers over the study window. It is a deep, honest single-case study, not a survey of thousands of accounts.

How it was measured. The data is real — actually-posted replies, actually re-scraped engagement, actually-diffed follower lists — not a simulation or a self-reported poll. That's the strength: there's no recall bias, no "what users say they do." The agent logged what it did and what happened next. The findings are internally consistent because they all draw from the same instrumented pipeline.

The honest caveat. It is n=1 account (@thedeepflux, a solo-founder tech account), single niche, single voice, single timezone. None of the numbers should be treated as population constants — your niche, audience timezone, and account size will shift every one of them. Sub-findings split the 983 replies across tiers/archetypes/hours, so some buckets are only ~100 data points. And the account operated in a ToS gray area (automation), which shaped how the data was generated. This is a directional compass, not a benchmark.

The tactic it implies. Use these findings as hypotheses to test on your own account, not as settled facts to copy. The valuable move isn't "copy @thedeepflux's exact peak hours" — it's "instrument your own replies the way this study did: log what you post, re-scrape what it earned, and let your own data tell you your warm/cold gap, your peaks, your best archetypes." The methodology travels even when the specific numbers don't.