I’ve been selling on Amazon for almost 4 years, focusing primarily on the baby and home categories. In my first two years, I made countless product research mistakes and lost nearly $2800 before I developed this field-tested product research workflow. Today, every product I pick consistently delivers a net profit margin of 40% or higher. I’m sharing the full process today, and I welcome feedback and input from fellow sellers.

I split product research into three core stages, which filter out 90% of high-risk products before you invest any significant time or money.

Stage 1: Identify Genuine User Needs

Many sellers start product research by judging if a product looks nice or if the supplier price is low enough. I take the opposite approach: I set aside all product features first and focus on finding the core, real user pain point. For example, the core value of a pregnancy pillow isn’t the pillow itself – it solves the problem of pregnant people struggling to find a comfortable sleeping position during pregnancy. If you nail that core pain point before you adjust product features, you’re very unlikely to miss the mark.

I use three main methods to find unmet needs:

  • Browse in-site category rankings. I go through every second and third-tier subcategory of my target niche, cross-referencing products on both the New Releases (NR) list and Best Sellers (BS) list. If an NR product also appears on the BS list, I break down what the competitor did right, and brainstorm small innovations I can make to differentiate my offering. If an NR product has no equivalent on the BS list, I first assess if it addresses a new, unmet demand, and if I can capture market share quickly with a price or feature advantage.

  • Find products reverse-engineered from keywords. I regularly compile rising search terms with growing search volume. If a term has reached meaningful search volume but no mature matching product exists, I check if I can adjust an existing mature product to fit the demand and launch it at a premium price point. I also keep an eye on long-tail terms related to materials or use cases – for example, material terms like muslin cotton have dozens of matching niche product opportunities, making it easy to find underserved demand.

  • Reference trending content on external platforms. When I browse Instagram and TikTok, I intentionally look at content with a product-focused mindset. If I find a product featured in a video with over 100,000 views, fewer than 50 product-related comments, and fewer than 3 top-ranked listings in its category, I reach out to suppliers for quotes immediately. In 2023, I found a baby cooling mat using this exact method, and it reached the top 20 of its category within two months of launch, holding a 45% net profit margin ever since.

Have you ever encountered a product that looked viral on the surface but turned out to be fake demand? Feel free to share your bad experiences in the comments.

Stage 2: Initial Research and Filtering

After identifying a potential demand, I run a quick initial screening to rule out products not worth investing in, saving time for higher-potential opportunities. I look at 5 core criteria for this stage:

  • Prioritize net profit margin, and eliminate any product with below 40% margin immediately. I get quotes from public B2B sourcing platforms first, and if the net profit margin (not counting ad spend) is below 40%, I almost never move forward, to avoid losing principal if ad costs rise later.

  • Check how new-seller friendly the category is. If no new listing has broken into the top 100 of the category in the past 3 months, or if there is no room for small, innovative differentiation, I move on.

  • Run a quick patent check to eliminate products with clear infringement risk.

  • Eliminate any category where average Cost Per Click (CPC) is above $2, to avoid overly high ad spend cutting into return on investment later.

  • Check keyword search trends: if search volume has been declining for 3 consecutive months, I drop the product, only keeping niches with stable or rising search volume.

Stage 3: In-Depth Research and Validation

For products that pass the initial screening, I run a second round of detailed research to confirm if it’s worth launching:

  • First complete competitor analysis and demand validation. I categorize all products on the BS list that address the same demand, export all reviews from the first 10 pages of relevant keyword results, and compile the most common user pain points to define our differentiation angle. Last year, I used this exact method to find that most pregnancy pillows on the market lacked sufficient support. After adjusting the fill material, my listing reached the top 50 of its category within 3 months of launch.

  • Run a second patent check to confirm no risk of design or trademark infringement.

  • Confirm product lifecycle and seasonality: use Google Trends and keyword trends to judge if the product is seasonal, if launching now aligns with the sales window, and if demand has long-term growth potential.

  • Check market size and monopoly risk: use Amazon Brand Analytics (ABA) data to judge total market demand, confirm that top sellers do not control more than 60% of the market share, and estimate realistic target sales based on the performance of competitors at different tier levels.

  • Assess overall competitor strength: if top sellers are established niche specialists with over a dozen relevant listings in the category, I carefully weigh if it’s worth entering the market.

Beyond this workflow, I have a set of non-negotiable product research rules: I drop any product that breaks even one of these rules:

  • I do not sell in apparel, consumer electronics, Amazon policy-restricted categories, or categories where average CPC is consistently above $2

  • Product price must be at least $20, and gross profit margin (after ad spend, calculated at the lowest comparable price point) must be at least 35%

  • Total modification plus production lead time cannot exceed 30 days, to avoid missing the sales window

  • Total upfront investment per SKU cannot exceed $4200, to control trial and error costs

  • If modification requires new tooling, mold costs cannot exceed $420

I’ve used this workflow for 4 years, and it has helped me avoid at least 80% of common product research mistakes. Fellow sellers have also shared supplementary demand analysis methods that I’ve tested and found extremely useful: you can use the Jobs To Be Done (JTBD) framework and Kano Model to prioritize demand, first meeting mandatory user needs before optimizing for expected needs and delight features, which maximizes the success rate of new launches.

What does your current product research workflow look like? Do you have any additional tips to share? I’d love to hear your thoughts in the comments.