Table of Contents
| The $1,200 Question You Only Need to Ask Once |
| The Problem With Amazon Research Tools |
| What We Built |
| How It Works |
| The Gemini Filter and Why Human Review Matters |
| What the Excel Report Contains |
| A Real Example: Intimate Washes for Women, Amazon India |
| Honest Limitations |
| When to Use This vs. When to Pay for the Subscription |
| How to Get Started |
| FAQs |
The $1,200 Question You Only Need to Ask Once
A founder wanted to know one thing before launching into a new category: who’s already selling this, at what price, and how crowded is it really? That’s it. One question, asked once, before a go or no-go decision.
To answer it properly, the standard advice is to sign up for a $99 to $199 a month research tool, learn the interface, pull the numbers, and cancel the subscription a month later once the question has been answered. Most founders do this, use the tool two or three times, and let it sit unused for the rest of the year while the bill keeps arriving.
That’s not a research problem. That’s a pricing mismatch. And it’s the reason we built something else.
The Problem With Amazon Research Tools
Helium 10 starts at $99 per month. The plan most serious brands actually need, the Platinum tier, runs $199 per month. That works out to roughly $1,200 to $2,400 per year for a subscription that most D2C brands doing market research use two or three times before it collects dust.
The pitch makes sense if you’re a full-time Amazon seller constantly tracking rankings, monitoring competitors by week, and running PPC campaigns that need keyword search volume data. For those teams, the subscription pays for itself.
But that’s not most of the brands we work with. What we see more often is a founder or brand manager who wants to enter a new category, understand who’s selling what, what price points are working, and how crowded the space is. That’s a one-time question. You don’t need an annual subscription to answer it.
What We Built
/amazon-research is a Claude Code slash command that runs a full Amazon market research pipeline using the Smacient MCP connector. You type one command, something like/amazon-research "intimate washes for women" --marketplace in, and it returns an 8-tab Excel report and a Markdown summary.
The whole run costs about 8 Smacient credits, which works out to a fraction of a dollar. Not $99 a month. Per run.
It doesn’t replace Helium 10 for everything. But for the specific job of understanding a category before you enter it, it does the work.
The skill is available at https://github.com/smacient/marketing-skills/tree/main/skills/amazon-research.
How It Works
The pipeline has five stages, and Claude Code drives all of them.
First, it generates 8 search keywords for your category. These aren’t generic terms. They’re specific to the category you provide, covering different angles: primary terms, format-specific terms, audience-specific terms.
Second, it runs each keyword search against Amazon through the Smacient connector. Each search returns up to 50 products. Across 8 keywords, you end up with a raw pool of several hundred product listings, duplicates and overlaps included.
Third, it hands the full product list to Gemini AI, which classifies every product as RELEVANT, BORDERLINE, or IRRELEVANT. This step matters more than it sounds.
Fourth, and this is the part we like most, before dropping borderline products, Claude presents them to you in the chat and asks for your call. Not every borderline case is the same.
Fifth, once you’ve confirmed the final product list, the script extracts brand names, product formats, and pricing from each listing and then runs the analysis.
The Gemini Filter and Why Human Review Matters
When we tested this on intimate washes for women on Amazon India, the raw search results included men’s grooming products, generic soaps, and a handful of baby care items that slipped through on certain keywords. Gemini caught most of them automatically. But there were a few products where the answer wasn’t obvious: moisturising washes that could serve either segment, pH care products without clear gender targeting.
Those borderline cases came back to us in the chat. We reviewed each one and gave a yes or no. It took about two minutes.
Without that step, you’d either be too loose, with garbage data skewing your market size estimates, or too strict, dropping legitimate products and undercounting real competitors. The human-in-the-loop review keeps the data clean without requiring you to manually sort through 300 raw listings yourself.
What the Excel Report Contains
The output is an 8-tab Excel file: Market Overview, Brand Analysis, Revenue Estimates, Top Products, Format Breakdown, Pricing Analysis, Competitive Flags, and Raw Data.
Market Overview gives you the headline numbers: total products analysed, unique brands, estimated total monthly revenue.
Brand Analysis is where the real work lives, every brand is ranked by estimated monthly revenue, product count and average selling price per brand. That’s where you see who actually owns the space, not just who shows up in search.
Revenue Estimates breaks the data down at the product level, so you can see the spread between the top 10 listings and everything else.
Top Products lists the highest-revenue individual listings with their prices, ratings, and review counts, useful for understanding what a winning listing actually looks like in the category.
Format Breakdown groups products by type. In the intimate wash example, that meant gels vs. foams vs. sprays vs. wipes. You can see instantly what format the market gravitates toward and where there might be gaps.
Pricing Analysis cuts the market into budget, mid-range, and premium tiers and shows how revenue and volume are split across them.
Competitive Flags surfaces signals worth a second look: products with high review counts but low ratings, a category where one brand takes a disproportionate share of estimated revenue, or price gaps between tiers.
Raw Data is the full cleaned product list, every product that made it through the filter, with all the fields, so you can sort it however you want.
The Excel file is saved to your outputs folder alongside the markdown summary that shows up in the chat.
A Real Example: Intimate Washes for Women, Amazon India
When we ran this on the intimate wash category for women on Amazon India, here’s what the data showed.
91 products made it through the filter across 57 unique brands. The extrapolated market size came to INR 1.27 crore per month.
VWash Plus leads the category by a wide margin: INR 24.56 lakh per month across 3 products, at an average selling price of INR 268. Namyaa comes in second at INR 3.83 lakh per month across 5 products, at a slightly higher average of INR 319.
The pricing data was telling. The budget tier, products under INR 300, dominated in volume. That’s where most of the sales are happening. But the mid-range and premium tiers aren’t empty. There’s enough revenue there to suggest that buyers do trade up if the product positioning is right.
One thing the data made clear: this category is not as fragmented as it looks. 57 brands sound like a crowded space, but the revenue is concentrated. The top two brands account for a significant chunk of the estimated market. Most of the 57 are doing negligible volume.
If you were evaluating whether to enter this category, you’d know within 20 minutes of running the command: who you’d be going up against, what price point the market gravitates toward, and which formats are doing well.
Honest Limitations
This is not Helium 10. There are things it genuinely cannot do, and it’s worth being direct about that.
The biggest one: only 29 of the 91 products in the intimate wash analysis had sales data available. That’s 31.9%. Amazon doesn’t expose sales data for every listing, and we’re working with what’s visible. Revenue estimates for the rest are extrapolated from the products that do have data, which introduces error. The market size number is indicative, not precise.
There’s no keyword search volume data. You’ll know what products exist and roughly how they sell, but you won’t know which search terms drive traffic or how competitive the bidding is on any given keyword.
There’s no rank history. You’re seeing a point-in-time snapshot. A product that ranks well today might have been in the top 10 for six months or six weeks. You can’t tell from a single pull.
There’s no PPC data. You won’t see what competitors are spending on ads or which keywords they’re bidding on.
If you need any of those things, rank tracking over time, keyword search volumes, competitor PPC analysis, you need Helium 10 or a comparable tool. This skill isn’t a substitute for ongoing category monitoring.
When to Use This vs. When to Pay for the Subscription
Use /amazon-research When: you’re evaluating whether to enter a category, you want a competitive snapshot before a brand meeting, you’re helping a client understand their Amazon competitive set, or you need a rough market size estimate for a new product.
Pay for Helium 10, or a similar tool, when you’re actively selling on Amazon and need to track your rankings week over week, you’re running PPC campaigns and need search volume data to build keyword lists, or you need historical rank data to understand how competitors respond to seasonality.
The two aren’t in competition. One is for episodic research. One is for ongoing operations.
How to Get Started
What you need
- Claude Code, Anthropic’s AI coding tool. Requires a Claude subscription (Pro, Max, Team, or Enterprise) or a Console account. Available as a CLI, a desktop app, and extensions for VS Code and JetBrains.
- The Smacient Claude Connector is the data layer that gives Claude access to Amazon search data and drives the Gemini classification step. Available at smacient.com.
- A Google Gemini API key is used for the product filter. The free tier is sufficient for most category runs.
- A Python virtual environment is set up in your project folder since the analysis script runs locally.
Installation
- Get the skill from https://github.com/smacient/marketing-skills/tree/main/skills/amazon-research, or clone the full repo and copy the folder you need:
git clone https://github.com/smacient/marketing-skillscp -r marketing-skills/skills/amazon-research path/to/your/project/.claude/skills/amazon-research - Set your Gemini API key as the GEMINI_API_KEY environment variable in your Claude Code settings.
- Set up a Python virtual environment in your project root and install the dependencies
- Set your Gemini API key as the
GEMINI_API_KEYEnvironment variable in your Claude Code settings. - Setup takes about 15 minutes the first time.
Running your first analysis
- Open a Claude Code session in your project folder with the Smacient Claude Connector active.
- Type
/amazon-research "your category" --marketplace <country code>, for example/amazon-research "intimate washes for women" --marketplace in - Claude generates the keyword set, runs the searches, and hands the results to Gemini for classification.
- Review any borderline products Claude flags for you, a yes or no on each, usually a two-minute step.
The script extracts brand, format, and pricing data and builds the full analysis.
Your report lands as an 8 tab Excel file in your outputs folder, alongside a markdown summary in the chat
Each run takes 5 to 10 minutes depending on category size, and costs about 8 Smacient credits.
Get the skill: https://github.com/smacient/marketing-skills/tree/main/skills/amazon-research
Connect the Smacient Claude Connector: smacient.com
FAQs
/amazon-research skill? It’s a free, open-source Claude Code slash command that runs a full Amazon category research pipeline: keyword generation, product search, AI classification with a human review step, and a full revenue and pricing breakdown. The output is an 8-tab Excel report plus a Markdown summary.
The skill itself is open source and available at github.com/smacient/marketing-skills. You need a Claude Code session, the Smacient Cloud Connector, and a Gemini API key to run it. Each run costs a small number of Smacient credits, a fraction of a dollar, rather than a monthly subscription.
Those tools are built for ongoing operations: rank tracking, keyword search volume, PPC monitoring, week over week. This skill is built for a one-time question: who sells in this category, at what price, and how concentrated is the market. It doesn’t do rank history or search volume data. If you need those, you still need a subscription tool.
After Gemini classifies every product as relevant, borderline, or irrelevant, Claude shows you the borderline cases directly in the chat and asks for a yes or no on each. In our test run, this was a two-minute step that kept the data clean without requiring you to sort through hundreds of raw listings yourself.
They’re indicative, not precise. Amazon doesn’t expose sales data for every listing, so revenue for products without visible sales data is extrapolated from the products that do have it. In our intimate wash example, only about a third of products had sales data available, which is worth keeping in mind when reading the market size number.
Claude Code, free from Anthropic, the Smacient Cloud Connector, a Google Gemini API key, and a Python virtual environment in your project folder. Full setup is in the “How to Get Started” section above.
