Find out if anyone will buy it.
Before you write it.

No — and anyone who says they can is selling you something. Book sales are a lottery with a very long tail, and most of what makes a hit isn't testable by anybody. What a market test does reliably is remove the bottom half of the outcomes: it tells you when a category is dead, when a cover doesn't belong on the shelf, and when a blurb loses people in its first three lines. We can't promise you a hit. We can stop you publishing into a category that sells three copies a day.

Because you only get to do that once. The first thirty days set your ranking, your also-boughts and your visibility, and a weak launch follows that listing around permanently — you can't cleanly relaunch the same book. You also only learn one thing: that it sold badly. You don't learn whether a different title would have sold ten times as much, because there's nothing to compare it against.

The instinct is right — you want variation, and something to compare. But that version varies the most expensive thing in the system, the whole book, in order to learn about the cheapest, the packaging. Pen names also start cold, with no platform, list or reviews, so what you measure is the floor rather than your book's real prospects. Hold the book still and vary the cover, title and blurb instead. That's this test, for the price of an afternoon rather than a few years.

It's the best possible time. Everything this test measures — category, cover, title, blurb, price — exists before chapter one does. Finding out the category is dead costs you an afternoon now, or a year later. And if the verdict is go, we can write the book for you (£49) and put it in front of five real readers before you publish (£149).

Then you got what you paid for. A market test that only ever said "looks great" would be worthless, and we'd rather tell you a category is dead than take another £49 to write into it. A negative verdict isn't a failed order — it's usually the most valuable one, and it normally comes with the version of the idea that would work.

Mostly in what gets compared. Running ads for one cover tells you how that cover did; it tells you nothing about whether another would have done three times better. We run 3 versions against each other on the same audience and the same budget, so the difference between them is the cover and the title rather than luck or timing. Covers routinely differ two- to fivefold in click-through, which is a big enough gap to see clearly on £100 of spend.