What is the rank today for product B07XXXXXXX across all my keywords?
why ask: a snapshot of where one product stands today. Useful before any conversation about where to push next.
Helium 10 charges $229 a month. Data Dive charges $150. Amazon will hand you the same data for free — you just have to ask. This is how to wire it up in one evening, even if you've never written a line of code.
Every keyword research tool charges a "tracker quota." Helium 10, Data Dive, Sellersprite — they all cap how many keywords you can watch, then bill you monthly for the privilege. They all do the same simple thing: search Amazon for your keyword, write down where you ranked, repeat tomorrow.
Here is the part nobody mentions. Amazon has had a developer tool sitting in plain sight for years called the Catalog API. You ask it "show me the top results for [keyword]," and it gives you back a list — the same list a shopper would see if they searched. Find your product in that list, and that is your rank.
Amazon will hand you that exact same data. For free. With no quota.
Does it match the real Amazon search results? A seller tested it side-by-side on 10 keywords. 9 out of 10 matched perfectly. The one off-by-one was a flip between #4 and #5. For watching how your rank moves over time — which is the only thing these tools are actually good for — that is more than good enough.
Before you build anything, decide which keywords you actually care about. For most accounts, 50 to 200 keywords across your catalog is plenty. Tracking 5,000 just buries the signal under noise.
The keywords worth tracking are the ones where rank actually moves the needle: terms that drive your sales, terms you'd happily pay ads for, terms competitors are camping on. Where to find them: your own search-term reports, Brand Analytics if you have Brand Registry, Helium 10 / Cerebro exports, or just the keywords you already obsess over.
The format does not matter. A text file with one ASIN and one keyword per line. A two-column Google Sheet. A note app. The script reads whatever you give it. The only thing it needs is a list of (product, keyword) pairs. That is the whole input.
Start with your top 20 ASINs by revenue, and 5 keywords each. That's 100 pairs and covers the keywords that pay your rent. You can always add more later. The point isn't volume. The point is watching the right ones.
Once a day, the system makes six stops. You don't have to understand any of them — Claude Code writes all the code. But here is the route, for the curious:
What the database holds: one row per keyword per day per product. Seven columns — date, product, keyword, rank (or blank if outside top 100), top competitor, total results, captured at. After 5 years the file is still under 10 megabytes.
Gather these before you start. Don't worry about understanding them yet — Claude Code will tell you exactly where to paste each one.
claude.ai/code. The engine that writes the code so you don't have to.The "API keys" thing isn't as scary as it sounds. They're passwords for software to talk to Amazon on your behalf. You'll generate maybe 8 of them total, paste each one when prompted, and never touch them again. Claude Code holds your hand through the whole flow. There is nothing to memorise.
Open Claude Code in any folder on your computer. Then paste the message below into the chat. Don't worry about the technical words — Claude Code reads them natively.
I want to build a daily Amazon keyword rank tracker. Here's what it should do:
keywords.csv in the project folder.Use the API keys in my ~/.secrets/.env file. Verify the database has rows in it before you tell me it's done.
The prompt tells Claude Code what you want, not how. The clearer the contract, the less the AI drifts.
Claude Code will do this for you. You'll see it print a line for each keyword as it goes — something like B0XXXXXXXX 'your-keyword-here' rank=#11. If most come back with a number, you're good. If they all error out, your API keys probably aren't right yet — paste the error back to Claude Code and it'll diagnose.
Download a free tool called DB Browser for SQLite. Open the database file the script created. You'll see a table with one row per keyword, today's date, and the rank you saw. That is your historical log starting on day one.
Just ask Claude Code: "is the daily schedule set up correctly on my server?" It'll log in, check, and tell you yes or no. The next morning, ask: "did last night's rank tracker run?" If yes — congratulations, you have a rank tracker. If no, paste any error back and Claude Code will fix it.
| Tool | Tracked Cap | Per Month | You Build |
|---|---|---|---|
| Helium 10 (Diamond) | ~2,500 | $229 | $0 |
| Data Dive Rank Radar | 1,000 | ~$150 | $0 |
| Sellersprite | 200–1,500 | $50–$200 | $0 |
| This script | Unlimited | $0 | + $5/mo VPS |
The kicker isn't the dollar savings. At the high end you save a few thousand a year. Real, but not life-changing.
The kicker is data ownership.— what you actually walk away with
Your rank history sits in a SQLite file you control. You can plug it into any other system you build later — a PPC bid-adjustment agent, a morning Slack digest, an alert engine, an end-of-quarter deep-dive analysis. None of that is possible when your history is locked behind another vendor's dashboard. The third-party tools all hold your history hostage. Yours stays yours.
The first day you'll only have one day of data. Obviously. The "rank changed by 5 positions overnight" alerts only become useful starting day two. Just let it run.
The rank you see won't perfectly match what you see when you search Amazon yourself. Real shoppers see slightly personalized results based on Prime membership, location, and shopping history. Your tracker shows the "neutral" view. For trend tracking — which is the whole point — that's actually what you want.
The whole point of owning your data is being able to ask any question you want. You don't write code or SQL — you talk to Claude Code. Open it in the folder where the database lives and ask things like the questions below. It writes the technical query, runs it, hands you the answer.
Rank tracking became a commodity the moment AI got good enough to write the code for you. The data was always sitting inside Amazon. The math was always simple. What was missing was a programmer to glue it together — and that is not missing anymore.
The other point: what you really want from a rank tracker isn't the tracker. It's the decisions. Should I cut ad spend on this keyword? Is this product slipping? Where should I push next? Those decisions are easier when the data is yours and not locked behind someone else's login screen.
Build the boring foundation once. Own it forever. Then go build the interesting stuff on top.— see you back on the beach