Skip to content

· 7 min read

A9 Amazon Ranking and the Seven Numbers to Check First

Joel Turcotte Gaucher

Joel Turcotte Gaucher · Founder

Flapen cover for A9 Amazon Ranking and the Seven Numbers to Check First: choosing the hero sample for a shoot against a color card

A9 is the name sellers use for Amazon's product search ranking system. No public formula sits behind it for you to optimize against, so treat the name as a label rather than a lever. What moves is what a shopper does on your listing, the click, the purchase, and the return, so fix those numbers first.

The short version

  • A9 is a name, not a control panel. Sellers use it for whatever orders Amazon search results, and no published formula sits behind the word.
  • Conversion is the part you own. A visitor who reaches your listing and buys is the outcome every hour of the work aims at.
  • Our free audit reads seven areas inside 48 hours. It covers listing quality, primary image click-through rate, conversion rate, ad performance, traffic channel activation, pricing, and return rate.
  • The product sets the ceiling. We build for 0.2 stars above the niche average, because the complaints that decide a rating are already written in competitor reviews.
  • Buy the measurement, not the word. Ask any vendor which of the seven areas is your constraint, and what number would prove the work paid.

Amazon A9 in plain terms

A9 is what sellers call the system that orders Amazon search results. The name travels much further than any published detail about it, so most of what circulates is inference presented as fact.

I treat it the way I treat any system I cannot open. A shopper types a phrase, scans a grid of tiles, clicks one, and either buys or leaves.

Those four moments produce numbers on your account, and those numbers are the only things you can move. So the useful question is never how A9 weighs anything. It is which of your numbers sit below the listings above you.

Seven checks to run before you buy ranking work

A post-launch seller arrives saying I'm spending money on ads but don't know if it's working. The diagnosis is the same whether the invoice reads A9, search, or SEO. Run these seven in order.

  1. Listing quality. The title, the bullets, and the A+ content carry the words a buyer would use. Done properly means a stranger names the product, its size, and one reason to pick it in ten seconds.
  2. Primary image click-through rate. This is the share of shoppers who see your tile in the grid and click it. Done properly means the tile reads at phone size and shows one visible difference from its neighbors.
  3. Conversion rate. Of the shoppers who reach the page, this is the share who buy. Done properly means you know the number per product and how it stands against the listings above you.
  4. Ad performance. Traffic bought against a page that does not convert buys the same failure faster and at a higher price. Done properly means the ad review follows the conversion review, never precedes it.
  5. Traffic channel activation. Sessions arrive from more places than search, and most accounts run fewer than the seller thinks. Done properly means you can name every source sending sessions and its cost of customer acquisition.
  6. Pricing. Price sits inside the tile next to the rating, so it is read before your listing opens. Done properly means the price is set against the shelf you compete on and revisited when it moves.
  7. Return rate. Returns erode margin quietly, and they feed the rating that decides the next shopper's click. Done properly means the rate is tracked per product against the under 8% bar we use in validation.

Run the seven yourself first, because a proposal that cannot name your worst check has not read your account.

What I checked before I hired anyone for search work

Before Flapen I ran data and technology at BRANDED and Moonshot Brands (YC W21), two large Amazon aggregators. Auditing and scaling 60+ acquired brands at $5M to $10M each meant reading somebody else's ranking work constantly. I was the buyer, and the word A9 sat in almost every proposal.

The proposals that leaned on the word never named a number. The ones worth signing did the opposite.

What the proposal said What I asked for instead What we publish against it
We optimize for A9 The conversion rate today and the target Seven audit areas in writing inside 48 hours, no charge
We will widen keyword coverage Which of the seven areas is the constraint 90+ data points behind every launch decision
We know what the algorithm rewards One threshold of theirs that was wrong 193,753 niches scored at the 2026-08-26 capture, 4.8% pass
We will get you to page one What would make them tell me to stop Kill criteria on four signals inside 60 to 90 days

Flapen figures as of September 2026. The niche cohort is the 2026-08-26 capture.

Read the right-hand column as a demand you make of anyone, ours included. We run about 70 brands by hand with 50 operators, and the majority reach profitability inside their first year.

Silver Aid sells pet wound care for dogs and horses, and we run the brand on Amazon at a 16.9% conversion rate, roughly double a typical category median. No line item there was ever sold as A9 work.

The four signals in the last row are the ones that end an argument: rating trend, return rate, conversion rate, and cost of customer acquisition trajectory.

What most agencies will not tell you

Three things stay out of the pitch, and on a careless day that includes ours. Each has a version of done properly you can hold a vendor to.

  1. Nobody selling ranking work can name a weight. No formula is published, so an ordered list of ranking factors is one company's reading of its own experience. Done properly means the vendor names which of your seven numbers they will move.
  2. The product can be the ceiling, and listing work never raises it. When the shelf sits at 4.2 stars and one complaint repeats across competitors, the answer is a better product built for 0.2 stars above the niche average. Done properly means the vendor reads those reviews before quoting.
  3. Ranking work on an unvalidated product is a bet with your money. Phase 1 runs 200 units on $5,000 to $10,000, and it exits only on rating, conversion rate, and cost of customer acquisition. Done properly means nobody sells you scale work until those three are proven.

Hold us to all three. If we quote ranking work without showing which of the seven numbers is losing, do not buy it.

One free thing this week, and it costs 20 minutes. If you run one to five products between $5K and $30K a month, open your best keyword and screenshot the top row.

Write down the price, the rating, and the review count for your listing and the three above it. Then put your own conversion rate beside that row. The worst column is the check to buy work against.

To have all seven read on your own account, request the free written audit that comes back inside 48 hours at Flapen.

Share this post
Joel Turcotte Gaucher

About the Author

Joel Turcotte Gaucher

Joel has spent 10 years in Amazon and ecommerce. He ran data and technology at BRANDED and Moonshot Brands, two of the largest Amazon aggregators. There he audited and scaled 60+ acquired brands. He co-founded Flapen to give sellers the data-driven tools and insights they need to compete. His expertise spans product research, listing optimization, PPC advertising, and international expansion.

FAQ

Questions sellers ask

The Flapen Weekly Product Research report, an Amazon niche shortlist scored 0–100 with its score radar on the cover

The weekly niche report

Product research, in your inbox

Every niche that cleared the bar this week. What it sells for, what it costs to enter, and why it passed. When we get one wrong, we publish the correction.