Complete Market Research Guide for Amazon Sellers

Most Amazon sellers decide which product to launch by looking at numbers: search volume, sales estimates, the number of competing listings, the margin left after fees. Those numbers say whether a category is large and crowded. They do not say what the people buying in that category are unhappy about, which is usually the most valuable thing a new seller can know.

That information is sitting in the reviews. Every competing product in a category has reviews that describe what buyers expected, what disappointed them, what they wished the product did and what they compared it with. Read across a whole category, those reviews show the need that no current product fully meets. Collecting them before entering a category, and organizing them into findings, is category research in its purest form.

This guide shows how a small or mid-sized Amazon seller, whether selling its own brand or other brands' products, can turn those reviews into real market research about its own products, its competitors and the category it sells in.

Key Research Questions

Before collecting anything, it helps to know what the research is for. For an Amazon seller, the questions fall into three groups: questions about your own products, questions about the products you compete with, and questions about the category as a whole.

Your Products

If you already sell, the first questions are about how your products are received. Which features do buyers praise, and which do they mention only when something goes wrong? Do the complaints point to the product itself, to the listing that set the wrong expectations or to packaging and delivery? Which variations, sizes or colors attract the most negative feedback?

If you are preparing a launch, the same questions apply to the product you are planning. What would buyers in this category expect from it, and which of their current complaints could it solve?

Your Competitors

On Amazon, your competitors are the products shown next to yours in search results. What do buyers praise in the best-reviewed products, and what do they keep complaining about even there? Which complaints are shared across many competing products, and which belong to one brand? A complaint that appears across the whole category is an opening; a complaint that belongs to one competitor is a weakness you can avoid.

The Category

The third group concerns the category. Who buys these products and why? Which needs come up repeatedly without a product that meets them? What do buyers compare these products with, including products from other categories or a different way of solving the same problem? Are expectations changing, for example around materials, size, durability or ease of cleaning?

Sources of Evidence

Evidence for an Amazon seller comes from four places: what your own buyers already tell you, what buyers of competing products write in public, what you can ask buyers directly, and what you can observe on the listings yourself. Each answers the questions above from a different angle, and the strongest research combines them.

Your Own Reviews and Records

Reviews of your own products are the starting point, together with what buyers tell you privately: messages, return requests and the reasons given for them. Returns and messages often name a problem before it appears in a public review. Collected in one spreadsheet, with columns for the product, the variation, the date and the source, they become the beginning of a customer feedback analysis practice.

Competitor Reviews in Your Category

Reviews of competing products are the core of category research. A single product's reviews describe that product; reviews across ten or twenty products in the same category describe the category. If the products you compete with are also sold on Walmart or Costco, reviews there add more buyers and sometimes different expectations.

Surveys and Conversations with Buyers

Traditional methods still have a role. A short follow-up message to buyers can ask one open question about what they would change. A few conversations with people who use products in your category, including people who chose a competitor, show how they search, compare and decide. Survey answers and conversation notes are qualitative research data, and they need the same structured analysis as reviews to become more than a collection of opinions.

Listings and Pricing

Two practical checks sit slightly outside feedback research. A listing check looks at how competing products are presented: the benefits they lead with, the questions their images answer, the claims they make about size or materials. A pricing check compares prices across the products you compete with, including bundles and variations. Reviews then add what a listing cannot: whether buyers felt the product delivered what the listing promised and whether it was worth the price.

Researching Amazon Products with Kimola

Kimola is an AI platform that turns customer feedback into enterprise-grade market research. For an Amazon seller, that means taking the reviews of your products and of the competing products in your category and turning them into a structured study of what buyers want, what the category fails to deliver and where your product can stand out. The steps below follow the order a researcher would work in, and you can start with a free account, without a credit card.

Building the Dataset

After signing up, go to the Create your report area on the Home screen and paste the link to a product page on Amazon. With the Add Multiple option you can paste several links, one per line, and Kimola collects the reviews from all of them into a single dataset. For category research, add the main competing products in the category, and your own product if you already sell one. If the same products are sold on Walmart or Costco, their product pages can go into the same report. The support article on creating a report by adding links walks through each screen.

Your own data, such as return reasons or buyer messages, can be uploaded as an .xls, .xlsx, .csv or .tsv file. If the file contains columns like product, variation, brand or month, Kimola reads them, and you can filter and compare the results by those values. The support center's tutorial on creating a consumer research report for a product category shows a complete category study built this way, using product attributes as columns.

By default, Kimola finds the themes in the data itself. If you prefer a classifier built for this kind of feedback, the Advanced option lets you choose the E-Commerce Feedback Classifier, one of Kimola's pretrained models.

Choosing Dimensions

Dimensions are the research questions Kimola asks of every piece of feedback. You choose them when you create the report or add them later, and the guide to Dimensions explains each one. For category research on Amazon, six are especially useful.

Unmet Needs asks what buyers want and cannot find. In a category report, these are the comments that describe the product nobody sells yet: a larger capacity, a quieter motor, a version that is easier to clean, a size that fits a specific space. Pain Points asks what goes wrong, which shows the problems the whole category shares and the ones that belong to particular products.

Motivations asks why buyers chose a product in the first place, which tells you what your listing should lead with. Personas groups buyers by who they are and how they use the product, such as a parent, a small business owner or someone buying a gift, and shows that the same product can satisfy one group and disappoint another.

Substitutions asks what buyers compare the product with or switched from, including other products in the category and entirely different solutions. Co-Occurrences looks at which topics appear together in the same reviews, for example whether complaints about size tend to come with complaints about instructions, which helps separate a single problem from a cluster of related ones.

Reading the Report

The Overview shows the main themes in the feedback and how sentiment is distributed. Kimola finds those themes in the data itself instead of sorting comments into a fixed list of categories, so a review can be tagged with something as specific as a lid that leaks when the bottle is laid flat or a charging port that loosens after a few weeks.

Each review can mention several themes, and each theme gets its own sentiment. A buyer who praises the design but complains that the product broke within a month contributes a positive mention of the design and a negative mention of durability, which a single star rating would hide.

The Performance screen ranks themes with a Satisfaction Score and a Dissatisfaction Score. These scores account for how many reviews mention a theme as well as the percentage, so a theme backed by five reviews does not outrank one backed by fifty. When some products in your report have far fewer reviews than others, this keeps the ranking honest.

Signals

Signals are findings that have passed a statistical test. Kimola crosses every theme and Dimension with the columns in your dataset, such as product, brand, star rating or date, and checks whether a topic is discussed more often or more negatively in one group than in the others (Differences) or more often recently than before (Trends). Only significant results become Signals, and they are ranked so the most important ones come first. The explanation of how Signals are tested covers the method in more detail.

A Signal might show that complaints about durability are concentrated in the products from one brand, that a particular unmet need is mentioned far more often in reviews of the cheaper products, or that complaints about a component have increased in recent months. Signals need enough feedback and at least some selected Dimensions. When the data does not contain a pattern strong enough to report, the report shows none, which is an honest result.

Compare

Compare puts segments of a report side by side: different products, different brands, or one-star and five-star reviews. Up to five can be compared at once, without using extra queries. In a category report, Compare is the natural way to see competitors side by side: the volume of feedback, the sentiment, the themes they share and the themes that appear for only one of them. Themes that appear for every product are category problems. Themes that appear for only one product show what that product does differently, for better or worse.

Action Plans

Action Plans turn the findings into steps. You start by choosing or creating a Profile that describes your business or brand, what it sells and to whom, then select a perspective: Our own offering, A rival or Our market. These match the three groups of questions at the beginning of this guide. For a category report built before a launch, Our market with the Product Development goal turns the findings into product ideas. For a report about a product you already sell, Our own offering with the Positioning goal focuses on how to present it against the competition.

Kimola builds hypotheses from the report, tests each one against the feedback and drops the actions the evidence does not support. Each remaining action comes with a confidence level and explains why it matters, who it is for, when and where to act, the risk involved and how to see that it worked, along with the evidence behind it. A plan contains up to 12 actions. Plans work best when the report already has Signals, as the support article on how actions are built explains.

Kimola also publishes category research made on the platform, with its methodology. The published report on electric toothbrushes shows what a category study built from Amazon and other retail reviews looks like.

From Findings to Decisions

Research is only useful if it changes a decision. For an Amazon seller the decisions are concrete, and each one connects to a specific part of the report.

Product Selection

A category is worth entering when its reviews show a clear, repeated need that current products do not meet, and when that need matters to a large enough group of buyers. Unmet Needs and Pain Points show the need, Personas shows who has it, and the Performance screen shows how strongly it is felt. A category where the reviews are uniformly positive may be large, but it offers a new seller little to improve on.

Product Development

Once you know the gap, the reviews describe the specification. Pain Points shared across competing products become a list of things your product must not repeat. Unmet Needs becomes a list of things it could add. Co-Occurrences shows which problems come together, so you fix the cause instead of a symptom. Variations deserve the same attention: if complaints concentrate on one size or color, the issue may be the variation rather than the product.

Listing Content

Motivations tells you what buyers care about first, which is what your title, first image and opening bullet should address. Many negative reviews come from expectations the listing created, such as a size that looked larger in photos or a feature buyers assumed was included. When Pain Points shows that kind of mismatch for competitors, your listing can answer it directly with measurements, comparison images or clear statements of what is and is not included.

Pricing and Positioning

Compare shows how buyers describe value across products at different prices. If buyers of cheaper products complain about durability and buyers of expensive products complain about features they never use, there may be room for a product positioned between them. Your price is supported by what buyers feel they received, which reviews show more directly than sales estimates.

Differentiation

Differentiation means solving a problem the category shares, not adding a feature nobody asked for. The themes that appear across every competing product in Compare, combined with the unmet needs buyers describe, show where a product can be meaningfully different. Substitutions adds a further check: if buyers keep comparing the category with a different kind of product, the strengths of that product are worth studying too.

Reviews After Launch

After launch, your own reviews become a new dataset. Adding them to the same report as your competitors shows whether buyers notice the improvement you designed for. Trends Signals show whether a new complaint is growing, which is often the first sign of a supplier or quality issue, and early action on it protects both the product and your rating.

Ongoing Research

A single report gives you a snapshot, and an Amazon category changes as new products launch and existing ones are updated. Repeating the research at regular intervals, for example every quarter or after a significant change to your product, shows whether the changes you made actually moved the themes they were meant to move.

On paid plans, Feeds keep the research running continuously. A Feed collects public mentions from social media, news sites, blogs, forums, YouTube and review sites, based on the keywords and links you define. Alerts email your team when something important happens, and Mailings send a regular summary. Over time, Trends Signals show which topics are rising and which are fading.

Brands and Multi-Channel Sellers

Brands with many products, and sellers active on several marketplaces and their own website, face the same questions at a larger scale, with one more added: why does the same product perform differently from one channel, market or product line to another? On the Enterprise plan, each channel, product line or feedback source can become a Space, a continuously growing source of feedback classified with the same Funnel, which is a shared set of custom models and Dimensions. Because every channel is analyzed the same way, results stay comparable over time and across the business. Spaces also calculate an NPS from what customers write, on a scale from -100 to 100. It is derived from the positive and negative share of written feedback, so it is not the same as a Net Promoter Score from a 0 to 10 survey, and NPS targets can be set for each area.

When a finding needs action, Tasks assign it to a team member with the underlying feedback attached as evidence, and Tasks can be sent to Jira. After a Task is closed, Kimola measures feedback on the same topic again two and four weeks later, so a product change or a new supplier can be checked against what buyers say afterwards. Executive summary screens give leadership a single view of what customers are saying across the organization. Enterprise accounts are set up together with a dedicated Account Manager, and you can request a demo to discuss your setup.

Frequently Asked Questions

A few practical questions come up when an Amazon seller starts researching with customer feedback for the first time.

How many competing products should a category report include?

Enough to describe the category rather than one product. Including the products that appear most often in search results, across different price levels, usually gives a balanced picture. Compare then lets you look at up to five of them side by side.

Can I research a category before I sell anything in it?

Yes. Category research is most useful before a launch. Any product page on Amazon, Walmart or Costco can be added to a report, whether or not you sell there.

How many reviews do I need before the findings mean anything?

There is no fixed threshold, and the ranking on the Performance screen is designed to protect you from small samples by weighing how many reviews support each theme. Adding more competing products to the same report is the easiest way to reach a dataset large enough for Signals.

Getting Started

The buyers in your category have already described what they want and what current products fail to give them. Create a free Kimola account, add the main competing products in a category to a single report, and read what buyers are asking for. If you promote products rather than sell them, or sell books on Amazon, the market research guides for Amazon affiliates and bookstores selling on Amazon apply the same approach, and the market research guide for small businesses covers the method for any industry.