How to A/B Test Your Product Prices

Price testing is the process of showing different prices to different groups of customers and measuring which price generates the best results. Most ecommerce sellers set their prices once using cost-plus math or competitive benchmarking, then never test whether a different price would be more profitable. A systematic approach to price testing typically reveals that optimal prices differ from initial guesses by 10% to 30%, which translates directly to thousands or tens of thousands of dollars in additional annual profit for even modest-volume sellers.

Before You Start

Price testing is different from testing a headline, button color, or product image. When customers discover they were shown a different price than another customer for the same product at the same time, the backlash can be severe. Amazon faced significant public criticism in 2000 when customers discovered the company was testing different prices on the same products for different users. The ethical and practical constraints of price testing mean you need to be more careful about methodology than with typical conversion rate optimization experiments.

The safest approach is sequential testing (testing one price for a period, then another price for the next period) rather than simultaneous split testing (showing different prices to different visitors at the same time). Sequential testing eliminates the risk of two customers comparing prices and feeling unfairly treated. The downside is that it introduces time-based variables (one period might have different traffic quality, seasonality, or external events than another), but for most ecommerce sellers, this tradeoff is worth the reduced customer trust risk.

Step-by-Step Price Testing Process

Step 1: Define what you are optimizing for.
The "right" price depends on your goal. If you optimize for conversion rate, you will find a lower price. If you optimize for revenue per unit, you will find a higher price. The metric that matters most for business health is total profit per visitor, which is conversion rate multiplied by profit per unit. A product at $29.99 with a 3.0% conversion rate and $12 profit per sale generates $0.36 profit per visitor. The same product at $34.99 with a 2.6% conversion rate and $17 profit per sale generates $0.44 profit per visitor. The higher price is more profitable despite fewer sales. Always calculate profit per visitor, not just conversion rate or revenue per sale.

Step 2: Choose your test prices.
Start with two to three price points spaced far enough apart to produce measurably different outcomes. Testing $24.99 versus $25.49 is unlikely to produce a detectable difference in conversion rate because the change is only 2%. Testing $24.99 versus $29.99 (a 20% difference) will produce clearly different conversion rates that reveal price sensitivity. For your first test, try your current price, a price 15% to 20% higher, and optionally a price 15% to 20% lower. The higher and lower prices bracket the range and help you determine the direction of your optimal price. If the higher price generates more profit per visitor, test even higher in the next round.

Step 3: Calculate the sample size needed.
Price changes produce smaller conversion rate differences than most other A/B test variables, so you need larger sample sizes for reliable results. Use a sample size calculator (tools like Evan Miller's calculator or Optimizely's calculator are free online) with your baseline conversion rate and the minimum detectable effect. For a product with a 3% baseline conversion rate and a goal of detecting a 15% relative change (3.0% vs 2.55%), you need approximately 7,000 visitors per variation at 95% confidence. For a product with a 5% conversion rate detecting a 20% relative change, you need approximately 2,500 visitors per variation. If your product page gets 100 visitors per day, a two-variation test at the 3% baseline would need 140 days (70 days per variation), which is too long. In that case, focus price testing on your highest-traffic products where results come faster.

Step 4: Run the test for the full calculated duration.
For sequential testing, run Price A for the required period, then switch to Price B for an equal period. Make sure both periods cover at least one full week to account for day-of-week variations (weekday shoppers may differ from weekend shoppers). Avoid running tests during major sales events, holidays, or promotional periods that introduce confounding variables. Record daily metrics (visitors, conversions, revenue, profit) so you can analyze trends and spot anomalies.

Step 5: Analyze results using profit per visitor.
After both test periods have run for the full duration, calculate total profit per visitor for each price point. Account for all variable costs including platform fees (Amazon referral fees change with price), shipping costs, and expected return rates (which may differ at different price points). Compare the profit per visitor numbers and check whether the difference is statistically significant. If Price B generates $0.48 profit per visitor versus Price A at $0.41, and the difference is statistically significant at 95% confidence, implement Price B. If the difference is not statistically significant, it means both prices perform similarly and you should keep whichever gives you the better margin or test a different range.

Price Testing on Amazon

Amazon provides a built-in tool called Manage Your Experiments (available to Brand Registry sellers) that supports A/B testing for product titles, images, bullet points, and A+ Content. As of 2026, Amazon does not offer a native price A/B testing tool, so you need to use sequential testing. Change your price for a defined period (minimum two weeks, ideally four to six weeks per price point), measure the impact on unit sessions percentage (Amazon's conversion rate metric), units ordered, and total profit, then switch to the test price for an equal period.

Be aware that Amazon's algorithm responds to price changes in ways that affect the test. Lowering your price may temporarily boost your organic ranking because the improved price-to-value ratio increases conversion rate, which Amazon's algorithm rewards with better search placement. This means a lower price might appear to perform better than it actually would at steady state because the algorithm is giving it a temporary visibility boost. To account for this, allow a 3 to 5 day stabilization period after each price change before starting your measurement window.

Also factor in Amazon's referral fee, which is a percentage of the selling price (typically 15%). When you raise your price from $24.99 to $29.99, your referral fee increases from $3.75 to $4.50. This $0.75 increase in fees partially offsets the $5.00 price increase, so your margin improvement is $4.25, not $5.00. Always use after-fee profit in your analysis, not just revenue.

Price Testing on Shopify and WooCommerce

Shopify sellers can use apps like Neat A/B Testing, Dexter, or Intelligems to run simultaneous price A/B tests. These tools show different prices to different visitors and track conversion rates for each variation. Intelligems is specifically designed for price testing and handles the complications that general A/B testing tools do not, like ensuring customers see consistent prices throughout their session, handling cart and checkout pricing correctly, and accounting for the revenue and margin differences inherent in price tests.

WooCommerce sellers can use plugins like Nelio A/B Testing or manual sequential testing. The manual approach is simpler: change the price, run for two to four weeks, change to the next test price, run for an equal period, and compare. Track daily visitors, conversion rate, revenue, and profit in a spreadsheet. The manual approach is less rigorous statistically but costs nothing and works for sellers who do not have the traffic volume for simultaneous split testing.

If your store gets fewer than 500 visitors per day to the product you want to test, sequential testing over longer periods (four to six weeks per price point) is your best option. Simultaneous split testing with low traffic takes too long to reach statistical significance and the tools add cost without speeding up the process. Focus on testing your top 3 to 5 products by traffic volume, where results come fastest and the profit impact is largest.

Common Price Testing Mistakes

The most common mistake is ending a test too early based on preliminary results. After three days, one price might look clearly better, but early results are noisy and frequently reverse as more data accumulates. Commit to the full test duration calculated in Step 3 and do not peek at results and make decisions before the data is sufficient. Premature test ending is the single biggest source of incorrect pricing decisions from testing.

Another mistake is testing too many price points simultaneously. Each additional variation requires proportionally more traffic to reach statistical significance. Testing three prices simultaneously triples the required traffic compared to testing two prices. Start with two price points (current price and a test price), get a result, then test the winner against a new price in the direction the first test pointed. This iterative approach finds the optimal price in two to three rounds of testing, each with manageable sample sizes.

Ignoring external factors that affect conversion rate independently of price is also common. If you test a higher price during a period when a competitor was out of stock (artificially inflating your conversion rate), you will conclude the higher price works better than it actually does under normal competitive conditions. Record significant external events (competitor stockouts, your own advertising changes, seasonal trends, viral social media mentions) alongside your test data so you can account for their influence when analyzing results.

Beyond Simple A/B Testing

Once you have established a baseline through simple two-price tests, more advanced techniques can fine-tune your pricing further. Price elasticity analysis measures how much your sales volume changes for each percentage change in price. If a 10% price increase causes only a 5% volume decrease, your product is relatively price-inelastic and can support a higher price. If a 10% price increase causes a 15% volume decrease, your product is elastic and the current price may already be near optimal or too high.

Segmented pricing tests different prices for different customer segments rather than testing one price against another for all customers. New customers versus returning customers, customers from paid ads versus organic visitors, and customers from different geographic regions may all have different price sensitivities. If your data shows that paid ad traffic converts better at a lower price while organic traffic supports a higher price, you might consider running different promotions for different traffic sources rather than applying one price to all visitors.