A good Amazon product research workflow is a fixed sequence of checks that every candidate product must pass before you spend money on stock. It runs in five stages: shortlist, validate demand, assess competition, model margin after every cost, and monitor after launch. Do it the same way each time and you stop choosing products on instinct. Below is the full process, with UK-specific points on fees, VAT and data quality.
TL;DR
- Run every candidate through the same four tests: demand, competition, margin and legality. Drop it at the first failure.
- Model profit at SKU level, including referral, fulfilment, storage, returns and ad spend. Top-line revenue misleads.
- In the UK, treat modelled sales estimates with caution and check observed price and rank history, plus how the seller count has changed.
- Research does not end at launch. Feed live Amazon data back into your next decision.
- A scorecard beats a spreadsheet maze. Automate the repetitive parts so you can spend time on judgement.
Table of Contents
- Why a workflow beats hunting for a perfect product
- The five-stage workflow at a glance
- Stage 1: Build a shortlist from several angles
- Stage 2: Validate demand without trusting a single number
- Stage 3: Assess the competition honestly
- Stage 4: Model margin after every cost, then check compliance
- Stage 5: Keep researching after the launch
- Where spreadsheets break and automation helps
- Our perspective: research is a loop, not a launch task
- A worked example: scoring three candidates
- Putting it into practice with Osellpa
- FAQ
- Sources
Why a workflow beats hunting for a perfect product
Most failed launches do not fail because the product was unusual. They fail because a step was skipped. Industry commentary on UK research notes that new sellers often underestimate basic cost analysis, VAT compliance and logistics. It also stresses that rigorous validation matters more than finding a “perfect” product.
The context is not forgiving. In its 2025 report, SmartScout found sellers facing rising fulfilment fees and heavier competition. Over half reported lower profitability than before. Competing on price alone is a weak plan, so your research has to find products where you can build an edge: a bundle, a better listing, stronger operations or a brand.
Expectations also need calibrating. Stories of $10,000 months are common, but one 2026 statistics roundup estimates median seller revenue at $40,000 to $60,000 a year. Reaching high monthly revenue usually takes a data-backed approach to inventory, pricing and PPC, not luck.
The five-stage workflow at a glance
| Stage | Question it answers | Main output |
|---|---|---|
| 1. Shortlist | Which ideas deserve a proper look? | 10 to 30 candidates |
| 2. Demand | Do enough people buy this, steadily? | Demand score |
| 3. Competition | Can you realistically enter and win? | Competition score |
| 4. Margin and compliance | Does profit survive every cost, and can you legally sell it? | Landed margin per unit |
| 5. Monitor | Is reality matching the model? | Weekly SKU-level review |
Keep one row per candidate and one score per stage. A product that scores well on demand but fails on margin is a rejection, not a “maybe”.
Stage 1: Build a shortlist from several angles
Start wide, then narrow. Good sources for ideas include:
- Amazon’s own search pages. Autocomplete suggestions, “customers also bought” and the best sellers lists in subcategories show what UK shoppers already buy.
- Amazon’s native research tool. Product Opportunity Explorer analyses trends in searches, purchases, reviews and pricing. Check which marketplaces your seller account can access, as it is listed on Amazon’s US seller site.
- Your own catalogue. Variations, bundles and adjacent products to items that already sell are often the lowest-risk ideas, because you already have real data.
- Customer reviews. Read the one- and two-star reviews on category leaders. Repeated complaints are product briefs.
At this stage, avoid deep analysis. You are collecting names, ASINs and a one-line reason for each. Aim for more candidates than you need, because most will fail.
Stage 2: Validate demand without trusting a single number
Demand has three parts: size, stability and direction.
- Size. How many units per month does the product plausibly sell? Use best seller rank as a proxy and search volume for the main keywords.
- Stability. Is demand steady across the year, or does one season carry it? A product that makes 70% of its sales in six weeks needs a different cash plan.
- Direction. Is the trend rising, flat or falling over at least 12 months?
The UK-specific trap is relying on modelled revenue. Practitioners warn that third-party sales estimates can mislead in the UK, because the calibration models were originally built on US data. Experienced sellers lean on observed price and rank history as a better signal of stability.
A practical rule: treat any single estimate as a range, and trust a product only when two independent signals agree. For example, rank history that is steady for a year plus rising search interest.
One UK checklist suggests benchmarks such as 200 to 300 units a month and a best seller rank below 10,000 in main categories. Use figures like these as starting filters rather than laws. A subcategory with a higher rank can still be very profitable.
Stage 3: Assess the competition honestly
Demand tells you the prize. Competition tells you the price of entry.
What to check on the page
- Review counts and ratings of the top ten. If every result has thousands of reviews and 4.6 stars, your launch costs will be high.
- Listing quality. Weak images, thin bullets and no A+ content mean a better listing can win share.
- Price spread. A tight price band suggests a commodity. A wide one suggests room to position.
- Brand dominance. If one brand holds most of the first page, the category may be locked.
What to check over time
Look beyond rank to the history of the number of sellers. As the same UK checklist points out, a product with high sales and a fast-rising seller count often signals a race to the bottom. It is a pitfall new sellers miss when they look only at current demand.
Also check which sellers hold the Buy Box. If Amazon itself or a large private-label brand wins it most of the time, your share of sales will be small even with a good listing.
Estimate your advertising burden
Competition shows up in your cost per click. Before committing, search the main keywords and note how many sponsored placements sit above the organic results. A crowded top of page means a higher advertising cost of sale at launch. You will carry that cost in the margin model next.
Stage 4: Model margin after every cost, then check compliance
This is the stage that kills the most candidates, and it should. Many sellers fail by underestimating the cumulative effect of referral fees, fulfilment, storage and ad spend, along with tax compliance costs.
Build the unit economics
Start with the Amazon UK revenue calculator, which helps you project profit across different products, variations and fulfilment methods. Then list every cost per unit:
- Product cost (COGS) and inbound freight
- Import duty and the VAT cash-flow effect, where relevant
- Referral fee: Amazon UK lists 8% for items up to £10 and 15% above £10 in many categories, with a minimum per-item fee, so check your exact category
- FBA fulfilment fee, which depends on size and weight (see our guide to dimensional weight and the 5,000 rule)
- Storage, including long-term storage risk for slow sellers
- Returns and damages
- Advertising spend, modelled as a percentage of sales
Set a margin floor before you look
Decide your minimum acceptable margin up front, so a tempting product cannot talk you out of it. Professional sellers often target a net margin of 15 to 20% after fees, COGS and ad spend. The same source reports that 58% of sellers turn a profit in their first year and 22% do not.
Run a stress test
Re-run the model with three changes: the price drops 10%, the ad cost of sale is 5 points higher than planned, and the return rate doubles. If the product still clears your floor, it is robust. If one change wipes out the profit, you have found the real risk.
Check that you can legally sell it
The fourth test is not about profit. Confirm that the product category is not restricted for your account, that any required safety marking and documentation exist for the UK market, and that your VAT position is clear. VAT in particular catches new sellers out, so speak to an accountant who works with online sellers before you order a first container. Do this before you commit capital, not after the stock arrives.
Stage 5: Keep researching after the launch
The model you built is a forecast. Real data is the correction. Once the product is live, compare the forecast with the actuals each week:
- Is the unit conversion rate in line with your assumption?
- Is the advertising cost of sale falling as the listing gathers reviews?
- Is your actual profit per unit within a couple of points of the model?
- Are replenishment and restock limits putting your stock position at risk? Our guide to calculating FBA headroom explains how to check it, and automating replenishment covers how to stop fees eroding margin.
- Is advertising simply replacing organic sales? A 30-day TACoS check helps you see it.
This is where Amazon’s Profit Analytics is useful. It gives a SKU-level view that combines sales, returns, fulfilment fees, advertising spend and your own cost data. The lesson is to judge a product by SKU-level margin, not by revenue.
Where spreadsheets break and automation helps
A spreadsheet is fine for ten candidates. It stops being fine at fifty SKUs with live ad campaigns, because every export is stale the moment you download it, formulas drift and nobody trusts the numbers. Practitioners increasingly see automation as a necessity rather than a nice-to-have for a growing business. The most effective workflows pull direct Amazon API data for real-time visibility, as Amazon’s own guidance on profit tracking suggests.
The same logic applies to the supply side. If you are a brand that also wants to sell through retailers, wholesale buyer discovery software can reduce manual prospecting in the same way that API-connected tools reduce manual reporting on Amazon.
What should you automate?
- Profit tracking by SKU, so your margin model is checked against real fees.
- PPC management, so bids and wasted spend are handled continuously rather than monthly.
- Alerts for stock, price and rank changes on your products and your competitors.
- Reporting, so profit and loss is ready when you need it, not after an afternoon of copying columns.
What should stay manual? The judgement calls: whether a niche fits your brand, whether the product can be differentiated, and whether the risk is worth the capital.
Our perspective: research is a loop, not a launch task
We build Osellpa for Amazon sellers, and the pattern we see is consistent. Sellers put their effort into choosing the product, then run the business from scattered exports. The research is only as good as the data that feeds the next decision.
That is why we connect directly to Amazon’s API. Profit tracking, PPC optimisation, competitor keyword tracking, inventory management, reimbursement tracking and product research management sit in one place. The numbers you used to approve a product are the numbers you use to judge it six weeks later.
Our view on tools is simple. A tool should remove the repetitive work, show you the cost of each SKU honestly and leave the decisions to you. We also think advanced tooling should not be reserved for large sellers. A part-time seller with a handful of SKUs needs accurate profit visibility just as much as a large one.
If you take one idea from this article, make it this: a product is not “validated” at launch. It is validated when its real margin holds up month after month.
— Harry
A worked example: scoring three candidates
Here is a simplified illustration, not real market data. Imagine you score each candidate from 1 to 5 on each stage, and you reject anything that scores 2 or below on any stage.
| Candidate | Demand | Competition | Margin | Compliance | Decision |
|---|---|---|---|---|---|
| Silicone baking mat set | 4 | 2 | 3 | 5 | Reject: crowded, low differentiation |
| Refillable pet water bottle | 3 | 4 | 4 | 4 | Proceed to sample |
| Seasonal garden lights | 5 | 3 | 2 | 3 | Reject: margin fails stress test |
The second product wins not because it has the biggest demand but because it passes every stage. That is the point of a workflow: the winner is the product with no fatal weakness.
Putting it into practice with Osellpa
Once a product is live, Osellpa helps you check whether the research held up. You get profit tracking, automatic and manual PPC optimisation, AI-powered listing enhancement, competitor keyword tracking and automated review requests in a single subscription. Plans are priced by your Amazon sales volume: Launch is £10 per month or £96 per year, Scale is £20 per month or £192 per year, and Advanced is £40 per month or £384 per year. Compare them on the Osellpa pricing page, or see the Osellpa homepage for an overview of the tools.
FAQ
How do I do product research for Amazon?
Follow a fixed sequence: build a shortlist, validate demand, assess competition, model margin after every cost and confirm you can legally sell the product. Score each candidate at each stage and reject it at the first serious failure. Then monitor real results after launch.
Is FBA still profitable?
It can be, but margins are tighter. One 2026 statistics summary reports that most sellers use FBA and that 58% are profitable within a year, while 22% are not. Rising fees mean you should model profit per SKU before buying stock.
Can you make $10,000 a month selling on Amazon?
Some sellers do, but it is not typical. Median annual revenue is estimated at $40,000 to $60,000, and the top 10% exceed $1 million. Reaching high monthly revenue usually takes disciplined inventory, pricing and PPC management.
What is the best tool for Amazon product research?
It depends on what you need to answer. Amazon’s native tools cover trends and fee modelling, third-party tools add history and keyword data, and API-connected software tracks real profit after launch. Pick the combination that covers demand, competition and margin without forcing you into spreadsheets.
Sources
- Amazon Product Research UK: Tools, VAT and the Checklist
- SmartScout’s Voice of the Amazon Seller 2025
- Amazon Statistics 2026: 25 Facts Every Seller Must Know
- How Many Amazon Sellers Are There: 2026 Statistics
- Amazon Profit Analytics (Seller Central UK)
- Amazon FBA Revenue Calculator UK
- Amazon UK pricing and referral fees
- Product Opportunity Explorer
