Thirty-four percent. That was the median lift in monthly units across 61 backlist titles after we did nothing but rewrite their metadata — categories, keywords, subtitle, description. No new covers. No ads. No promotion. Same books, same authors, same prices.
I lead with that number because authors consistently treat metadata as paperwork and covers as strategy. The data says the reverse is closer to true.
Why the mechanism works
Retail discovery is a category game. Your book competes inside a bracket, and the bracket you choose determines both how much traffic you can receive and how much competition stands between you and it. Choose a bracket that's too broad and you're invisible at rank 40,000. Choose one that's too narrow and you're visible to almost no one. The ceiling is set before a single reader sees your cover.
Think of it as two multiplied numbers: the demand flowing through a category, and your share of the attention inside it. A title sitting sixth in a modest category frequently outsells the same title sitting ninety-fourth in a glamorous one. Authors optimise for the prestige of the shelf. The money is on the shelf where you are visible.
The audit, in order of payoff
- Categories. Three to five candidate brackets. For each, note the units-per-day implied by the rank of the #10 book. That's your realistic ceiling — not the #1 book's.
- Subtitle. It is a search field, not a poem. It should contain the phrase a stranger would type.
- Keywords. Seven slots, seven phrases, no repeated words, no words already in your title or category name. Repetition wastes a slot.
- Description. First two lines carry the click; everything below the fold carries the buy.
- Series and read-through. Discounting a standalone is arson, not marketing. Discount book one of three and the math changes entirely.
The read-through arithmetic
This is the part that surprises people, so here is the whole calculation. Say book one earns you £2 and 55% of its readers continue to book two, then 70% of those continue to book three. Every book-one buyer is worth roughly £2 + (0.55 × £2) + (0.55 × 0.70 × £2) ≈ £3.87. That is why a series author can afford to pay £1.50 to acquire a reader and a standalone author cannot. Same market, same ad platform, completely different economics — driven entirely by whether the catalogue is linked and the linkage is visible in the metadata.
What to measure
Change one variable, wait fourteen days, compare units — not rank. Rank is relative to everyone else's week. Units are yours. Keep a plain spreadsheet: date, what you changed, units for the fourteen days before, units for the fourteen days after. Six entries in and you will know more about your own market than any course will tell you.
Most "algorithm secrets" are category math with better branding. Do the math. It takes an hour.





