What I'm checking

What the evidence shows: Yes, with archived captures, though coverage is patchy. For one book, the per-day numbers showed when it grew fastest, and a halving method narrows it further when growth is lumpy. The Ratings Rewind on this site does the sums. Read NB-002

What the evidence shows: Heated Rivalry had a small group writing about it for years before its show. Next, I want to follow books with a high fan writing rate, adapted or not, and see whether that pattern repeats. I'll publish the result either way. Read CS-001

What the evidence shows: Part of the Launch Study.

What the evidence shows: Not in the file I studied, which holds 325,610 posts. The 1,000 most-played posts came from 788 different accounts, and the 15 most-played accounts got 6.4% of all plays. It's a large sample, not all of BookTok. Read NB-001

What the evidence shows: Yes. One in five of the most-played 1% of posts came from accounts under 10,000 followers. Big accounts still have an edge: accounts with 100,000+ followers made half of those posts. Read NB-001

What the evidence shows: Not how much each factor counts. Amazon names some factors, including sales history, Best Sellers Rank and keywords. It also says what authors control: up to seven keywords and three categories. And it says rank weighs recent activity more heavily. Read CS-002

What the evidence shows: A model is an attempt to work out Amazon's rules from the outside by watching what happens. It can be trusted only as far as its data and method can be checked, and none of the six guides showed enough to check. The piece lists seven questions to ask any guide, starting with how much data it really used. Read CS-002

What the evidence shows: Partly. Both sites ranked the three books in the same order, but the gaps were very different: Heated Rivalry's rate was 316 times The Deal's on AO3 and about 7 times on Wattpad. Part of the reason may be what each site's tag holds, because a tag isn't a topic. Read CS-001

What the evidence shows: Probably not the only reason. Red, White & Royal Blue is M/M too, with a ratings count much closer to The Deal's. Its fan writing rate was far above The Deal's, but less than a third of Heated Rivalry's. Read CS-001

What the evidence shows: By the mainstream measures, Off Campus. The Deal had about a million Goodreads ratings against about 42,000 for Heated Rivalry, and its show was publicly announced first. Fan writing told a different story: on AO3, Heated Rivalry had 13.6 fan works for every 1,000 ratings, against 0.02 for The Deal. Read CS-001

What the evidence shows: At least 72 hours. Amazon estimates up to 72 hours for keyword, title and description updates to appear, and up to a week for rank after a category change. Read CS-002

What the evidence shows: No, and Amazon says so: rank "is not an accurate way to track your book's activity". One sale can move a quiet book's rank a long way. Orders and pages read are the better measure. Read CS-002

What the evidence shows: Nothing in the file connects a post to a sale. TikTok says about 59 million print sales in 2024 could be tied to BookTok, but that is TikTok's own figure, and it doesn't explain how a sale gets tied. I found no independent research on whether big or small creators move more copies. Read NB-001

What the evidence shows: Free books appear to be more common than payment. Only 0.25% of posts carry a paid-ad hashtag, which is almost certainly an undercount. Creators told The Bookseller that free and early copies are far more common than payment, but that is one UK trade article. Read NB-001

What the evidence shows: My theory is that stories tend to be open or closed, and that a high fan writing rate may partly measure how much room a story leaves. If that holds, leaving room is a choice a writer can make. Nothing yet tests it. Read CS-001

What the evidence shows: Not always. Of the 819 tags with 300 or more posts, 49 are half or more one account. A tag's size shows how many videos there are, not how many people are talking. Read NB-001

What the evidence shows: Sometimes. The Internet Archive saves dated copies of Goodreads pages, and each one is a real point on a book's arc. Coverage is uneven, and some captures are blocked. Read NB-002 for a step-by-step guide.

What the evidence shows: No. Four of the six I compared lean on sales velocity, and two put conversion first. Half are written for sellers of physical products, some of the numbers aren't weights at all, and none showed enough data to check. Read CS-002

What the evidence shows: No. Amazon says newer reviews carry more weight, and so do reviews from verified purchases. Read CS-002

What the evidence shows: My plan is to change one thing at a time and write down when. I wait for Amazon's stated update window before counting, then judge by orders and pages read, not rank. After a few rounds, it should show how my own book responds, which no general guide can. Read CS-002

What the evidence shows: Mostly by trope, in the tags I checked. The bigger trope tags are spread across many accounts: #darkromance holds 9,031 posts from 3,280 accounts. For a writer, that makes the most useful BookTok question about a manuscript a craft question: which tropes does it truly deliver? Read NB-001

What the evidence shows: Two tags show two shapes. #booktropes spiked and faded, though the spike more likely reflects how the file was collected. #romantasy built slowly, to about 25 of every 1,000 posts in 2025. A spike shows what was loud that month, not what lasts. Read NB-001

What the evidence shows: I count totals on public tag pages and never look at who wrote anything. Next, I want to learn whether counts like these are useful, and what fan writers consider fair. Read CS-001

What the evidence shows: Launch guides give numbers, but none I've found can be checked. Part of the Launch Study.

What the evidence shows: Part of the Launch Study.