How to find a profitable KDP niche with AI - a research workflow

Quick answer: The workflow is: Amazon Best Sellers category browsing → Perplexity trend validation → competitor review gap analysis → Claude synthesis. Done well, this takes 2-3 hours and tells you whether a KDP niche has real buyer demand before you create anything.

The most expensive mistake in KDP publishing is spending weeks writing or designing a book for a niche that is either too competitive or too small to support a new entrant. AI tools compress the research phase from weeks of manual browsing to a structured 2-3 hour session.

This guide covers the actual workflow.

What makes a KDP niche profitable

A profitable KDP niche has three things:

1. Consistent search demand — buyers are actively searching for it on Amazon, not just on Google
2. Manageable competition — the top results are not all from established publishers with thousands of reviews
3. A quality gap — something the existing books are not doing well that you can do better

For low-content books (journals, planners, activity books, colouring books), a fourth factor matters:

4. Visual differentiation — the cover and interior design are meaningfully better than the current alternatives

Step 1: Browse Amazon Best Sellers (30 minutes)

Start where buyers are. Go to Amazon → Books → Best Sellers → drill into specific subcategories.

The subcategories to pay attention to for low-content and niche books:

• Books > Self-Help > Journals & Activities
• Books > Crafts, Hobbies & Home > Crafts & Hobbies
• Books > Education & Teaching > Schools & Teaching
• Books > Children's Books (for activity and colouring books)
• Kindle Store > Kindle eBooks > [any category]

What you are looking for:

• Books in the top 100 of a subcategory that have under 100 reviews — recent success without an entrenched incumbent
• Subcategories where the top sellers have very different styles (inconsistent quality = opportunity)
• Seasonal books that dominate a subcategory only part of the year (a gap when off-season)
• Any search result page where the first few results look clearly worse than what you could produce

Build a list of 3-5 candidate niches from this browsing session.

Step 2: Validate trends with Perplexity (30 minutes)

Perplexity adds real-time web context to your hypothesis. For each candidate niche, ask:

Is [niche] a growing trend for books and planners in 2026? 
What are people searching for in this space that is not well served yet?
What formats (journal, workbook, activity book) are most popular?

Follow up:

Are there any emerging subtopics within [niche] that have growing interest 
but fewer published books than the main category?

Perplexity will cite sources — click through. Social media trend reports, Pinterest annual trend reports, and Amazon editorial category reports are all useful signals.

What to look for: Evidence that the niche is growing (not peaking or declining), and specific sub-niches that are underserved. A trend article from 6 months ago saying "[niche] is booming" is weak evidence. A current search trend or a newly-popular subtopic is stronger.

Step 3: Competitor review analysis (30 minutes)

For each candidate niche that survived Perplexity validation, open the top 5-10 Amazon listings and read the reviews — specifically the 3-star and below ones.

What negative reviews tell you:

• "Not enough pages" — buyers want more content than the current books provide
• "Boring design" — visual differentiation is the opportunity
• "Too basic / too advanced" — there is a specificity gap
• "Does not work for [specific use case]" — a niche within the niche
• "Wished it had [feature]" — a concrete product improvement

Every complaint is a product brief. A journal niche where the top sellers all have 3-star reviews saying "the prompts are generic" is a direct signal: write better, more specific prompts and you have a differentiated product.

Also note: what is the average review count of the top 10? Under 50 reviews each = accessible niche. Over 500 reviews each = established competition, higher barrier.

Step 4: Synthesise with Claude (20 minutes)

Feed everything to Claude:

I am researching KDP niches. Here is my research:

CANDIDATE NICHES:
[list your 3-5 candidates with Amazon subcategory and approx review counts]

TREND SIGNALS (from Perplexity):
[paste what you found]

COMPETITOR GAPS (from review analysis):
[paste the complaints you found in reviews]

Please:
1. Rank these niches from most to least promising for a new KDP publisher
2. For the top niche, describe the specific product gap in one sentence
3. Suggest 3 specific product ideas with a clear differentiator for each
4. Name 2 risks I should know about before committing to this niche

Claude will not tell you what will definitely sell. But it will surface patterns across your research and help you articulate the gap before you start creating.

Step 5: Validate the cover concept (optional, 30 minutes)

Before committing to a niche, validate that you can create covers that stand out visually. The cover is the primary conversion lever for KDP books.

Open Midjourney and generate 4-6 cover concepts for your proposed book. Compare them side by side with the current top sellers in your category.

If your Midjourney output is clearly better than what is ranking — cleaner, more on-trend, more professional — that is a strong execution signal. If it looks similar or worse, you need either a stronger design brief or a niche where visual design is less of a differentiator.

For low-content books that need consistent interior illustrations, Recraft is useful at this validation stage — it can show you whether you can generate a matching set of interior elements at the quality the niche demands.

What this research cannot tell you

No research fully predicts Amazon's algorithm. The platform weights factors you cannot observe externally — click-through rates, conversion rate benchmarks by category, A+ content quality signals.

Signals that are reliable:

• Low review counts on top sellers = accessible niche
• Specific, recurring complaints in reviews = product gap exists
• Growing trend signal + underserved subtopic = timing opportunity

Signals that mislead:

• Amazon Best Seller rank (BSR) without context — a book can spike to #1 with a small number of sales in a tiny subcategory
• Seasonal spikes that look like permanent trends (Christmas activity books, Valentine's planners)
• A category that looks low-competition because it is too small to support a business

Run this workflow for 3-5 niches and compare before committing. The niche that looks best after you have seen three alternatives is almost always a better bet than the first one that seems promising in isolation.