How to Run an Amazon Price Test Without a Split Test

Amazon gives you a split test for your title, images and A+ Content. It gives you no split test for price. Here is how to test price on one ASIN. You also get the math to read the result before you change the price.

Manage Your Experiments, Amazon's own split test tool, covers images, titles, bullet points, descriptions and A+ Content. So you test price over time on one ASIN. Make one price move. Change nothing else. Run the same number of weeks on each side. Read the result against a break-even number you set before the test.

Why doesn't Amazon let you A/B test price?

Amazon built Manage Your Experiments for listing content. It splits shoppers between two versions of product images, titles, bullet points, descriptions or A+ Content, including Brand Story. Amazon's own tool page lists those five options. A price test runs outside the tool, on a timeline you control.

The tool also has entry rules. You need a Professional selling account, Brand Representative access in Brand Registry, and enough recent traffic on the ASIN. Amazon recommends 8 to 10 weeks per experiment. Its "to significance" setting sometimes returns a result in four.

Open the experiment page in your own account. It shows a finished main image test. Eight weeks and a declared winner. On the same listing, the price field still shows the launch-day price.

The image got eight weeks of structured testing. The price got a guess.

Amazon decides which tools you get. You still decide what you test. A launch-day price with no test behind it sits on the listing for months. Price moves more numbers on the page at once than any listing change. Every month it sits untested, you have no number to compare it with.

Why does price move ACOS, conversion and margin at once?

Price sits inside all three of these numbers at once. It sets revenue per click. ACOS moves with it. Shoppers compare it in the search grid. Conversion rate moves with it. Most of each price dollar is margin. The referral fee takes its share first. No image or title change touches all three.

First, revenue per click. ACOS = CPC ÷ (CVR × price). Hold the click cost and the conversion rate flat. Then change the price. ACOS moves on its own.

Second, conversion rate. Shoppers see your price next to other offers in the search grid before they read a single bullet. Search Query Performance (SQP) shows both prices for each query. In Seller Central, go to Marketing, then Brand Analytics, then Search Analytics, then Search Query Performance. Switch to the ASIN view. Set the reporting range to weekly. Open "Customise columns". Add two columns to the Clicks, Basket Adds and Purchases groups. Price (median) is the median offer price on the query. Brand price (median) is the median price of your own ASINs. If shoppers buy at a median well under your brand price on your top queries, price is the first thing losing you the sale.

Third, the margin on each unit. Every dollar you take off the price comes off what each sale keeps, less the referral fee on each dollar. A raise works the same way in reverse.

So a price test needs its math done before the test starts.

What does a $2 price cut need to do to pay for itself?

A $2 cut pays for itself only when the extra units bring back more margin than the cut removes. On a product keeping $7.50 a unit before ads, a $2 cut leaves $5.50. Units then have to rise 36.4% before the product earns what it earned at the old price.

Write this formula at the top of the test sheet:

Break-even unit lift = old margin per unit ÷ new margin per unit − 1

Margin here means the dollars left from a sale after you pay every cost except ads. Take the price. Subtract product cost, the referral fee, the FBA fee, and your returns and storage cost for each unit.

I built the inputs below for this article as a labelled example.

• Price: $24.99
• Margin per unit before ads: $7.50
• Organic units a month: 300
• Ad click: $0.90 CPC at 15% CVR
• Ad cost per order (CPA): $0.90 ÷ 0.15 = $6.00
• Profit per ad order: $7.50 − $6.00 = $1.50
• ACOS: $0.90 ÷ (0.15 × $24.99) = 24.0%

The table moves the price four ways. Each row shows the units and the conversion rate a move needs to earn the same as before:

The example moves margin by the full price change to keep the math clean. In your own sheet, the referral fee moves with price. The real margin change is a little smaller.

Read the −$2 row slowly. Conversion has to climb from 15% to 22.5%, a 50% jump, before each ad order earns what it earned at the old price. If it does, ACOS falls from 24.0% to 17.4%. The dashboard calls it a win. Profit per ad order is exactly where it started.

Now read the +$2 row. Conversion has room to fall from 15% to 11.3%. ACOS climbs to 29.6%. The dashboard calls it a loss. Each ad order still earns $1.50.

ACOS moves with price even when profit stays flat. Judge a price test in dollars.

How do you run a price test on Amazon today?

You run it as a timed test on one ASIN. Record two to four weeks at the current price. Move the price once. Freeze everything else. Run the same number of weeks again. Then compare units times margin, minus ad spend, for each period. Here is the method, in order.

1. Pick the ASIN. Steady sessions, in stock for the whole window, no Lightning Deal, coupon or Prime event inside the test dates. A product with thin traffic gives you noise.
2. Record the baseline in whole weeks. In Seller Central, open Reports, then Business Reports, then Detail Page Sales and Traffic by ASIN. Note sessions, Unit Session Percentage, units, sales and your Buy Box share. Some guides call this column Buy Box Percentage, others Featured Offer Percentage. In SQP (ASIN view, weekly range), note Price (median) and Brand price (median) for your top five queries. From the Sponsored Products campaign report, note CPC, CVR, spend and orders.
3. Write the break-even number in the test sheet before you touch anything. For a cut, the unit lift you need. For a raise, the unit drop you have room for. Date the line.
4. Move the price once. Change nothing else. No bid changes, no new images, no new coupon, no budget shifts. Every extra change turns the result into a guess.
5. Run the same number of weeks at the new price. Check your Buy Box share every day. A lost Buy Box ends the test.
6. Read the result in dollars. Units times margin, minus ad spend, for each period. Compare the two totals.
7. Keep a sister ASIN untouched as a control if you have one in the same category. If its sessions fall 10% over the same weeks, treat the 10% as the market. Take the same 10% off the test ASIN's change before you compare.

In my operator guide, price floors sit in the red zone: the decisions AI never makes alone. AI does the test sheet math: the break-even lift, the weekly totals, the gap to the query median. The price decision stays with the operator. Automate what you measure. Protect what you decide.

When in a product's life should you test price?

Test price at any stage with a clear question. After two weeks of steady sessions, a new product's test picks its price band: premium, middle or budget. While sales grow, it checks whether a cut pays for itself in units. At scale it finds the highest price before conversion rate drops. At end of life it sets the clearance pace.

The stage changes what you measure. An early test reads conversion on the ad clicks you pay for. A scale test reads the allowed unit drop. At scale, each extra cent of margin repeats on every unit you sell. An end-of-life test reads how fast units leave before storage fees take the rest of the margin.

What to do this week

1. Pick your top ASIN by ad spend. Write down its margin per unit before ads.
2. Work out the break-even unit lift for a $1 and a $2 cut. Then work out the allowed unit drop for a $1 and a $2 raise.
3. Open Search Query Performance in the ASIN view. For your top five queries, add Price (median) and Brand price (median) to the Purchases group. Note the gap.
4. Pull four weeks of sessions and Unit Session Percentage for the ASIN from Detail Page Sales and Traffic by ASIN. This is your baseline.
5. Check the calendar. If no event, deal or coupon falls in the next four to eight weeks, schedule one price move. Freeze everything else for the full test.

FAQ

Does Manage Your Experiments test price?

No. Amazon lists product images, titles, bullet points, descriptions and A+ Content, including Brand Story. Test price over time on one ASIN, baseline period against test period.

How long should an Amazon price test run?

Use the same number of whole weeks on each side. Two to four weeks per side covers weekday swings and matches the weekly range in SQP. Avoid Prime events, deals and Q4 weeks.

Does lowering price lower ACOS?

Only when conversion rises enough. If CPC and CVR stay flat, ACOS goes up when price goes down, because each click brings less revenue. A cut lowers ACOS only when the new CVR times the new price beats the old CVR times the old price. In the example, an 8.0% cut from $24.99 to $22.99 needs CVR to rise more than 8.7% before ACOS drops.

Should I test a price raise or a price cut first?

Start with the side your SQP gap points to. If your brand price sits well above the purchase median on your top queries, test a cut. If it sits at or below the median, test a raise. The allowed unit drop shows how far units fall before the raise loses money.

The main image test got eight weeks and a declared winner. The price got a guess on launch day. Give the price two weeks on each side, one change and a break-even number you write down first. Then it gets the same discipline as the image.

Sources

• Amazon, "Manage Your Experiments" (sell.amazon.com, checked 7 Nov 2026): what the tool tests, eligibility and recommended test length. https://sell.amazon.com/tools/manage-your-experiments/
• Amalytix, "Amazon Search Query Performance report" (updated 20 Oct 2026): click path, brand and ASIN views, weekly range, Price (median) and Brand price (median) columns. https://amalytix.com/en/knowledge/controlling/amazon-search-query-performance-report
• Novadata, "Amazon Buy Box Guide 2026" (updated 8 Oct 2026, checked 7 Nov 2026): Reports, then Business Reports, then Detail Page Sales and Traffic by ASIN, with a Buy Box Percentage column. https://novadata.io/resources/blog/amazon-buy-box-guide
• Novadata, "What is the Amazon Sales and Traffic report?" (26 Jul 2026, checked 7 Nov 2026): sessions and Unit Session Percentage per ASIN (units ordered ÷ sessions). https://novadata.io/resources/answers/sales-and-traffic-report-explained
• AccelerList help (updated 15 Jul 2026): labels the same Business Reports column Featured Offer Percentage, so the article describes the column instead of quoting one label. https://intercom.help/myaccelerlist/en/articles/15949402-how-to-check-buy-box-percentage-in-seller-central
• All other figures come from labelled worked examples built from the formulas in this article.