Most people pick a KDP niche the same way: a gut feeling, a YouTube video about "hot niches", and a quick look at Amazon. Then they spend two weeks making a book nobody searches for.
We do it differently. Before we make anything, our publishing system, Pressroom, scores every niche from 0 to 100 using Amazon's own data. So far it has analysed 170 search phrases and 336 books across four research runs. This guide walks you through the exact method, with the real numbers, so you can use it yourself, with or without code.
Short version: collect the phrases buyers type, look at the top 10 books for each one, and score six things: demand, competition, margin, series dominance, freshness, and title gap. Then check for red flags that override the score.
Step 1: Collect the phrases buyers actually type
Start with Amazon's search autocomplete, in the Books department. Type a seed phrase like "word search for seniors" and write down every suggestion. Then add one letter after the seed ("word search for seniors a", "…b") to dig deeper.
Autocomplete is gold for two reasons. Every suggestion is something real buyers typed, and the order hints at how popular it is. Pressroom records each phrase with a popularity value from autocomplete, which it later uses in the score.
Before you look at any results, set Amazon's Deliver to to a US ZIP code (10001 works). Otherwise prices show in your local currency and your margin math will be wrong.
Step 2: Look at the top 10 books for each phrase
Search each phrase and look at page 1. Ignore the sponsored results; you want the organic top 10, because that's what you'd compete with for free.
For each of those 10 books, note:
- Best Sellers Rank (BSR), from the product page
- Number of reviews
- Price and page count
- Publication date
- Series or brand, if it's part of one
That's the raw material. Everything below is calculated from it.
Step 3: Score six things
Each part gets a value from 0 to 1, and the weighted total becomes the niche score out of 100.
| Part | Weight | What it measures |
|---|---|---|
| Demand | 35% | How well books that target this phrase sell (BSR) |
| Competition | 25% | How many top books have few reviews |
| Margin | 15% | Your royalty per book after print costs |
| Concentration | 10% | Whether one series owns the results |
| Freshness | 10% | Whether new books are breaking in |
| Gap | 5% | How many top titles don't even target the phrase |
Demand (35%): only count books that target the phrase
This is the mistake most niche tools make. Amazon fills page 1 with generic bestsellers, so a phrase can look hot just because a famous book shows up next to it.
Pressroom measures demand from the books whose titles actually contain the phrase's specific words. A median BSR of 20,000 or better scores full marks; 400,000 or worse scores zero. If fewer than two top books really target the phrase, the angle is marked untested and its demand score is cut hard.
For a rough sense of what BSR means in sales (paperbacks, estimates only):
| BSR | About this many sales a day |
|---|---|
| 1,000 | 60 |
| 10,000 | 12 |
| 50,000 | 3 |
| 100,000 | 1.5 |
| 500,000 | 0.15 |
Competition (25%): reviews are the real barrier
A new book can rank, but it can't fake 1,000 reviews. So competition looks at two numbers: the share of the top 10 with fewer than 100 reviews, and the median review count.
Compare these two from our data:
- "jumbo word search for seniors": median of 19 reviews in the top 10
- "large print word search books for seniors": median of 1,310 reviews
Both phrases have buyers. Only one of them lets a new book in.
Margin (15%): what you actually earn per book
KDP pays 60% of the list price, minus the print cost. For a black-and-white 8.5×11 paperback (KDP's 2025 US schedule; check their calculator, as it changes):
- up to 108 pages: about $2.84
- over 108 pages: about $1.00 + $0.017 per page
So a 150-page large-print book at $11.49 earns:
royalty = 0.60 × 11.49 − (1.00 + 150 × 0.017)
= 6.89 − 3.55
= $3.34 per book
A royalty under $1 gets flagged. Our data flagged "fall word search puzzles" for exactly that: a median price of $6.32 left about $0.94 per book.
Concentration (10%): is one series in charge?
If one series holds 40% or more of the top 10, buyers already have a favourite and Amazon keeps recommending it. "Cozy coloring book" scored well on demand, but a single series held half of the top results.
Freshness (10%): are new books breaking in?
Count the top-10 books published in the last 12 months that sell better than BSR 150,000. If new books are selling, the door is open. "Murder mystery word search puzzle book" had 7 of them, the most of any phrase we checked.
Gap (5%): do the top books even target the phrase?
If most top titles don't contain the phrase's specific words, there's room for a book that matches the search exactly. It's a small weight because it's a nice-to-have, not proof of demand.
Step 4: Check the red flags
Some problems override any score:
- Brand or trademark searches. "Brain games word search" scored 71.6, but Brain Games is a brand, so you can't use it in your title or keywords. Same for "large print word search kappa". Pressroom checks a list of brands and also spots words shared by the leaders' brand or publisher names, then marks the phrase not targetable.
- Seasonal phrases. "Halloween word search for adults large print" scored 79.0 in late September, at the season's peak. The score is real, but the window is short. Launch 6–8 weeks before the season.
- Low price levels. If the market sells at $6, your royalty is tiny no matter how good the book is.
What the scores told us
Here's a slice of our real results. The data was captured between September 28 and October 13, 2026; BSR moves daily, so treat it as a snapshot.
| Phrase | Score | Verdict |
|---|---|---|
| murder mystery word search puzzle book | 90.8 | Strong, build it |
| jumbo word search for seniors | 89.8 | Strong, build it |
| cozy mystery word search | 86.9 | Strong, build it |
| cozy fall coloring book | 85.6 | Strong (seasonal) |
| bold and easy christmas coloring book | 78.4 | Strong (seasonal) |
| word search for seniors | 76.6 | Strong, build it |
| large print word search books for seniors | 63.5 | Viable, niche down |
| easy word search for seniors with dementia | 62.5 | Viable, niche down |
| brain games word search | 71.6 | Not targetable (brand) |
Three of the books we’ve published sit in niches from the top of this list: a murder mystery word search, a jumbo word search for seniors, and a bold and easy Christmas coloring book. You can see them, and the system that made them, on the Pressroom page.
How to read a score
- 75 and up: strong. Build it.
- 60 to 74: viable. Niche down further, or win on a better cover and interior.
- 45 to 59: weak. Only with ads or a much sharper angle.
- Under 45: avoid.
Doing this without code
You don't need our system to use the method. A spreadsheet works:
- One row per phrase from autocomplete.
- For each phrase, copy the top 10 organic books' BSR, reviews, price, pages, and publication date into a second sheet.
- Calculate the median BSR of the books that target the phrase, the share with under 100 reviews, the median royalty, and the count of fresh books selling.
- Score each part 0 to 1 against the thresholds above, apply the weights, and sort.
It takes about 30 minutes per phrase by hand. That's why we automated it, but the thinking is the same.
The honest limits
- BSR-to-sales numbers are estimates. Amazon doesn't publish sales. Use them to compare niches, not to forecast income.
- Scores are snapshots. Re-check a niche before you build, especially seasonal ones.
- A good niche doesn't sell the book for you. The cover, the interior, and the listing still have to be better than what's there. The score just tells you where that effort is worth it.
Want the next one? We publish a new guide from the builds every week, plus one product opportunity backed by data, in the newsletter. The walkthrough video for this guide is coming to YouTube.