← All experiments

Flasky reach engine — 60-post sprint to 1M

Refuted
Jul 3, 2026 — Aug 27, 2026

Hypothesis

The glassworks can produce 60 high-quality animated reels in 30 days (2/day, IG primary + TT cross-post). That volume lifts Flasky cumulative reach from 23,686 to 1,000,000. Animated series accounts convert reach to followers at a stable rate, so we optimize for reach and treat followers as downstream.

Success criteria

≥1,000,000 cumulative reach at day 30 (60 posts). Follower conversion holds ≥0.1% (1M reach → ~1,000 followers).

Key metrics

Cumulative reach, median views/post, follower conversion % (followers/reach)

Baselines

As of 2026-06-23 (locked): - Reach: 23,686 (24 posts) → ~987 median/post - Followers: 18 → 0.076% conversion - Target conversion assumed: 0.1% (→ 1M reach = 1,000 followers) - Glassworks throughput: 10 outputs / 3h → 18h for all 60 (~0.6h/day, ~5× the cadence) — volume is not the bottleneck

Results

Reach: ~100,163 cumulative views vs 1,000,000 target (~10%) — missed Conversion: 0.051% vs ≥0.1% criterion, flat vs pre-automation baseline — missed Throughput: 58/60 reels, 1.9 → 10.0 posts/week at ~0.6h/day machine time — validated Followers: ~30 → ~81 Per-post: median views 1,802 → 1,524 (−15%); like rate 2.37% → 1.85% Sprint closed at 58/60 — pipeline outage ~Aug 11; the restart pivoted to quality changes, so no later posts were eligible. 43 of 58 posts landed inside the 30-day window. Caveats: views used as reach proxy; TikTok cross-posting dropped mid-experiment; IG native insights unavailable at audit time.

Conclusion

Verdict: Refuted — volume alone does not buy reach. Learning: the constraint is content and funnel, not distribution. Posts clearing 1,000 views convert; posts below convert ~nothing. Per-post reach fell ~45% while conversion never moved. Decision: persevere. Next iteration targets production quality (prop variety, second character, editing rhythm); new headline metric = posts >1k views/week.

Log

  1. Baselines locked, sprint defined (60 posts / 30 days, IG primary + TT cross-post).

  2. Registered in the Experiments register; structured content moved here from the draft lab notes (both drafts since removed).

  3. Metrics policy: views-first for ongoing tracking (scraped twice daily). The 1M target stays reach-denominated — criteria locked at baseline — verdict via the next export; views are the live proxy in between.

  4. First sprint post confirmed (Viggo): 'to talk, or not to talk…' 2026-07-03. Started moved 06-23 → 07-03 (06-23 was baseline-lock/sprint definition). 7 IG posts to date tagged (Jul 3–9); 2 recent posts possibly not yet scraped — tag on arrival. Jun 22–29 posts (7 IG, first Flasky Studio format incl. 'hbu? are you also a drinker?') deliberately EXCLUDED — manual-assist pilots, several outperformed; marked Tool Used='Flasky Studio' + pilot note in Content instead. IG-only tagging for now (TT cross-posts untagged; TT tracker shows nothing after 06-29).

  5. Tagged the remaining 51 IG sprint posts in Content (now 58/58 tagged, Jul 3 – Aug 11). Post count: 43 landed inside the 30-day window (by Aug 2), 15 after; 58 of 60 total, cadence still active (~9 in past week). TT cross-posts remain untagged (TT tracker quiet since 06-29). Reach verdict still pending next export.

  6. Closed out: Refuted against pre-registered criteria (reach ~10% of target, conversion 0.051% < 0.1%). Final count 58/60 — posting stopped Aug 11 (pipeline outage) and the restart pivoted to Glassworks quality changes, so no later posts are sprint-eligible. Write-up distilled from Viggo's 2026-08-13 audit + 2026-08-14 pre/post table; full background notes in cortex strategy/experiments/exp-03-flasky-reach-engine.md.