Amazon product research is the single most important step in the entire Amazon FBA process. Pick well and the rest is execution. Pick badly and no amount of ad spend, photography, or copywriting fixes the underlying math. In 2026 the tools have gotten better, the competition has gotten faster, and the criteria that worked three years ago no longer all work. Below is the actual research process I run before I commit a single dollar to a new Amazon product, with the tools, the kill criteria, and the green-light criteria.
The tools that matter
You do not need a dozen subscriptions. You need three core tools and a couple of free ones to fill specific gaps.
Keepa ($19/month). Historical price and sales rank charts for every Amazon product. The single most useful research tool for spotting demand stability, seasonality, and the difference between a viral spike and a real evergreen winner.
Helium 10 or Jungle Scout ($35-$100/month depending on plan). Product database with estimated monthly sales, revenue, and competitor analysis. Pick one. The data is roughly equivalent. Helium 10 has better Chrome extensions, Jungle Scout has a slightly easier interface.
Seller Assistant App (free Chrome extension with paid tier). Quick analysis directly on Amazon product pages. Shows estimated fees, profit margins, restrictions, and historical data without leaving the page.
Free supplements: Google Trends for broader demand validation, Reddit search for honest customer pain points, AliExpress and Alibaba for sourcing checks, US Patent and Trademark Office search for IP risk.
The four-stage process
The research process I run breaks into four stages. Each stage filters down the pool from "every product on Amazon" to "the one I am ordering samples for". Most beginners try to skip stages or do them in the wrong order. The order matters.
Stage 1 - category and price filter
Start in Helium 10 or Jungle Scout with very wide filters and narrow down.
Filters to set:
- Price range: $20-$70 for a first product
- Monthly sales: at least 200 units for the top listing in the category
- BSR range: top listing under 5,000 in its main category
- Number of reviews: top listing 100-2,000 reviews. Below 100 means category is too new and risky. Above 2,000 means competition is entrenched.
- Category: anything except restricted (gated) categories for now
The output is a list of 50-200 product types that pass the basic filter. This is your starting universe.
Stage 2 - the unit economics filter
For each candidate product type, run the FBA fee calculator. The math has to work BEFORE you go deeper.
The formula: retail price minus Amazon referral fee (15% in most categories) minus FBA pick-and-pack fee minus estimated COGS minus inbound shipping minus duty minus expected PPC cost minus refund reserve. The result should be at least $7 per unit, ideally $10+.
Use Helium 10's profitability calculator or Seller Assistant's overlay for fast checking. Most product candidates fail at this stage because the unit economics are too thin. Eliminate ruthlessly.
The wider the price range you can absorb, the more room you have to compete. A product with $15 profit per unit can spend $5 on ads and still be profitable. A product with $4 profit per unit cannot spend $2 on ads without going negative.
Stage 3 - the Keepa stability check
The candidates that pass Stage 2 get the Keepa treatment. Open Keepa, look at the top 3-5 listings in the category. Three things to verify.
One - the BSR (Best Seller Rank) trend should be relatively stable over the last 6-12 months. Not climbing wildly (over-saturated or seasonal spike), not falling off a cliff (category dying). A flat-ish BSR with normal seasonal patterns is what you want.
Two - the price has been relatively stable. Massive price drops suggest sellers are racing to the bottom because demand collapsed. Stable prices suggest steady demand.
Three - the seller count has been stable. A category where 5 new sellers join each month is going to be saturated by the time you launch.
This stage eliminates products that look attractive in Helium 10 but are actually in late-stage commoditisation. Most beginners skip Keepa and miss this signal.
Stage 4 - the differentiation check
For the few remaining candidates, the final question: can you do anything meaningfully different from the existing top sellers?
Read the negative reviews of the top 5 listings in the category. Look for repeating complaints. "Strap broke after 3 weeks". "Color was different from photo". "Missing a key feature". Each complaint is a differentiation opportunity. Can your product fix one or more of those issues?
Look at the listings themselves. Are the photos generic? Could you produce better photos? Are the bullets uninspired? Could you write better ones? Is there room for a brand identity in a category dominated by no-name sellers?
If you cannot identify a clear differentiation angle, the product is wrong. Even if the unit economics work, you will compete on price alone and slowly lose. Detail in how to brand a product from scratch.
Competition just means customers exist in this niche. You just need to do a little better.
The kill criteria
Hard nos that should eliminate any candidate immediately:
Restricted or gated category that requires an existing approved business history. Skip.
Patent or trademark on the basic product type that limits your variants. Check USPTO. Skip if encumbered.
Top 5 sellers are all under one private-label brand with 50,000+ reviews. The category is owned. Skip.
Returns rate above 8% in apparel-adjacent categories. Returns will eat your margin. Skip.
Seasonality where 80% of sales happen in 2 months a year. Cash flow becomes brutal. Skip for first product. Detail in real cases like the nipple covers seasonal launch that died with the season.
Product weight that pushes you above the Amazon "small standard" tier. FBA fees jump dramatically. Skip unless margin is very high.
The green-light criteria
What I want to see before committing:
- Top 5 sellers each doing 300+ units per month
- Price range $25-$60 with $10+ margin per unit
- Reviews of top 5 in 200-1,500 range (proves market is open)
- Keepa BSR stable in the top 10K of category
- At least one clear differentiation angle I can execute on
- Product weight under 1 pound to keep in small-standard FBA tier
- No IP encumbrance
- Year-round demand (some seasonal swing is fine, total seasonality is not)
A product that hits all 8 is a strong candidate. Most products hit 5-6 and require a judgement call.
How long this should take
For a first product, the full research process from "I want to sell on Amazon" to "I have committed to a specific product" should take 2-3 weeks. The mistake is to compress it (rushed picks fail) or to extend it indefinitely (researching forever to avoid commitment).
Day 1-3: tool setup, broad category exploration. Building the candidate list.
Day 4-10: applying filters, eliminating bad candidates. Down to 10-15 strong candidates.
Day 11-14: deep Keepa and differentiation analysis on the final 10-15. Down to 2-3 finalists.
Day 15-21: order samples from 2-3 finalists. Photograph them. Stress-test the sample quality. Pick the one to commit to.
The mistakes to avoid
Searching for "winning products" on YouTube. By the time a YouTuber names a product, it is over-saturated.
Picking based on "I like this product personally". Your personal taste is one data point. The math is the deciding signal.
Skipping the Keepa stability check. The single most expensive omission. Many products that pass other filters fail here.
Researching too long. After 4 weeks of research with no commitment, you are stalling, not researching. Pick a product and order samples even if you are not 100% sure. The market teaches faster than your spreadsheet does.
What happens after the pick
Once you have committed, the research role shifts to monitoring. Keepa watches the category over the next 90 days. You revisit the unit economics monthly. New competitors get analysed quickly.
The skill of product research compounds. By your fifth product, you can do the four-stage filter in 2-3 days instead of 2-3 weeks. By the tenth product, you have an intuition for which categories are open and which are closed without running the tools.
For the launch playbook that follows, read Amazon FBA for beginners and how to write Amazon listings that convert. The full Amazon module - including the spreadsheet I use for stage-by-stage scoring - lives in the course. The candidate list is built in week one. The commitment is in week three.
