Reader demand
Books delivering it
Reverse-harem 170 : 1
Slow-burn 76 : 1
Found-family 74 : 1
Dark-romance 52 : 1
Forced-proximity 42 : 1
As measured 22 August 2026, 04:00 UTC
What this ratio is, and what it is not. “Demand”
is the number of reader signals in our corpus — Goodreads reviews, Reddit reader
requests — that name the trope. “Supply” is the number of titles in our
registry tagged with it. Both are counts of what we hold, not censuses of the market, so
the multiple is a ranking device, not a headcount: it says reverse-harem has less
published supply behind its demand than forced-proximity does. It does not say 170 readers are
waiting for each reverse-harem book.
It is also not comparable across dates. The 7 August version of this report showed found-family at 40× and reverse-harem at 125×. Our demand corpus grew 88.8% between then and now (974,544 → 1,839,816 dated signals) while the title registry grew far more slowly. That collection growth, not a change in reader behaviour, is most of the movement to 74× and 170×. Quote the ranking; date the multiple.
Two of our own findings pull in opposite directions here, and both are true. Reverse-harem has the thinnest published supply behind its demand — and it is also the clearest cooling trope in romance, having lost 46% of its share of reader conversation in a year on a fixed cohort of the same 649 books. A thin shelf is not automatically an opportunity.
It is also not comparable across dates. The 7 August version of this report showed found-family at 40× and reverse-harem at 125×. Our demand corpus grew 88.8% between then and now (974,544 → 1,839,816 dated signals) while the title registry grew far more slowly. That collection growth, not a change in reader behaviour, is most of the movement to 74× and 170×. Quote the ranking; date the multiple.
Two of our own findings pull in opposite directions here, and both are true. Reverse-harem has the thinnest published supply behind its demand — and it is also the clearest cooling trope in romance, having lost 46% of its share of reader conversation in a year on a fixed cohort of the same 649 books. A thin shelf is not automatically an opportunity.
Your unfair advantage
The hero who protects is gaining share of the conversation, and the tropes in section 03 have the least published supply behind the demand we can see. Where those two overlap is the honest place to aim — and knowing which is which beats guessing at a trend.
This is a dated snapshot, and every figure on this page carries its measurement date. The data moves every week.
See what YOUR next book should hit →TropeSmith tracks it live — a per-book map of what's rising, cooling, and thinly supplied in your subgenre, from the same 22.3-billion-play dataset behind this report.
Methodology. Every figure on this page carries the date it was measured. Nothing here is modelled, estimated or extrapolated; each number is a count you could re-run.
1. Plays — one video, one play total. Our BookTok panel holds one row per video, keyed on the TikTok video id, carrying that video's play count as at the moment we last captured it. The headline figure is the sum of those play counts. A video that appears under five hashtags is counted once; a video we have scanned forty times is counted once. This is the count published across PlotProse and TropeSmith, so a single number is live everywhere:
2. Per-hashtag plays. The same panel, grouped by hashtag, with the video count and median so the base is visible. This is our panel's plays under a tag, not the tag's lifetime reach — no public source gives us that:
4. Demand vs supply. Reader demand signals in our corpus that name a trope, against titles in our registry tagged with it, as at 22 August 2026 04:00 UTC. See the note in section 03 for what that ratio can and cannot be used for. Supply is drawn from a retailer title registry that carries no history, so no supply figure here is ever differenced against an earlier date.
Corrections. This report was corrected on 22 August 2026; the changes are listed at the top of the page. Corrections are published, not silently applied. A snapshot moves week to week — TropeSmith tracks it live.
Citing this report. Free to quote: cite any figure as “the PlotProse 2026 Romance Demand Report” with a link to this page. Published by PlotProse (Coral Hart Group).
1. Plays — one video, one play total. Our BookTok panel holds one row per video, keyed on the TikTok video id, carrying that video's play count as at the moment we last captured it. The headline figure is the sum of those play counts. A video that appears under five hashtags is counted once; a video we have scanned forty times is counted once. This is the count published across PlotProse and TropeSmith, so a single number is live everywhere:
SELECT count(*) AS videos,
sum((payload->>'playCount')::bigint) AS plays
FROM (SELECT DISTINCT ON (video_id) video_id, payload
FROM raw_booktok_videos
WHERE video_id IS NOT NULL
AND payload->>'playCount' IS NOT NULL
ORDER BY video_id, scraped_at DESC, id DESC) v;
22,293,880,514 plays across 72,641 videos as measured 21 August 2026; re-run on 22 August it returns 22,652,648,965 across 73,802 videos, because the panel grows hourly. We considered and rejected a second method — summing each hashtag's latest reported total, which gives about 15.6 billion. It double-counts any video carrying more than one hashtag, and the per-hashtag total is itself only the videos one scan returned, so it is not a stable measure of anything. The per-video count has neither problem, so it is the one we use everywhere.
2. Per-hashtag plays. The same panel, grouped by hashtag, with the video count and median so the base is visible. This is our panel's plays under a tag, not the tag's lifetime reach — no public source gives us that:
SELECT hashtag, count(*) AS videos,
sum((payload->>'playCount')::bigint) AS plays,
percentile_disc(0.5) WITHIN GROUP (
ORDER BY (payload->>'playCount')::bigint) AS median_plays
FROM raw_booktok_videos
WHERE payload->>'playCount' IS NOT NULL
GROUP BY hashtag ORDER BY plays DESC;
3. Trends — share, on the reader's clock. Each trope's mentions in a quarter divided by that quarter's total mentions across all romance lanes, Q2 2026 against Q1 2026. Two rules make this a market measurement rather than a measurement of us: we window on when the reader posted, never on when our pipeline parsed it; and we compare shares, never raw counts, because our archive is growing fast enough that raw counts would show almost everything “rising”. Only tropes with at least 250 mentions in each quarter are shown.
4. Demand vs supply. Reader demand signals in our corpus that name a trope, against titles in our registry tagged with it, as at 22 August 2026 04:00 UTC. See the note in section 03 for what that ratio can and cannot be used for. Supply is drawn from a retailer title registry that carries no history, so no supply figure here is ever differenced against an earlier date.
Corrections. This report was corrected on 22 August 2026; the changes are listed at the top of the page. Corrections are published, not silently applied. A snapshot moves week to week — TropeSmith tracks it live.
Citing this report. Free to quote: cite any figure as “the PlotProse 2026 Romance Demand Report” with a link to this page. Published by PlotProse (Coral Hart Group).