AB price testing means showing different prices, or different ways of presenting a price, to randomly assigned shoppers and measuring which one earns more. Done well, it’s one of the most valuable tests a store can run, because price touches every order. Done badly, it can annoy customers, confuse your support team, and give you a “winner” that quietly lowers profit. The safest path is to start with how prices are framed, move to offers and shipping thresholds, and only test actual list prices once your setup is solid.
Let’s walk through that in order.
Two very different kinds of price test
People lump these together, but they carry very different risks.
Presentation tests keep the price the same and change how it’s shown. Think “$60” versus “$2 a day,” a crossed-out compare-at price versus no anchor, or showing the bundle saving in dollars versus a percentage. Every shopper pays the same amount. These are low risk and often surprisingly powerful.
Actual price tests charge different groups different amounts for the same product. Group A sees $48, group B sees $54, and you measure orders and profit. These can answer the biggest question in your business, but they need care, both technically and ethically.
There’s a middle ground too: offer tests. Same base price, different offer structure. Free shipping over $50 versus over $75. Buy two get 10% off versus a free gift. These often teach you more than raw price tests, with less risk.
Start with presentation and offers
If you’ve never run a pricing experiment, begin here. You’ll learn how sensitive your shoppers are without changing what anyone pays.
A good place to look is shipping. Baymard Institute’s cart abandonment research found that among shoppers who abandoned for reasons other than just browsing, 40% said extra costs like shipping, taxes, and fees were too high. Another 12% said they couldn’t see or calculate the total cost up front. That suggests two easy tests: showing estimated shipping earlier on the product page, and testing where your free shipping threshold sits.
Threshold tests are a nice example of why you need the right metric. Raising the threshold might lower conversion rate slightly but raise average order value. Only revenue per visitor, or better, gross profit per visitor, tells you whether the trade was worth it.
Other presentation ideas worth testing:
- Per-use or per-day pricing on consumables and subscriptions.
- Showing the bundle saving as a dollar amount versus a percentage.
- Placing installment messaging near the price for higher-ticket items.
- Leading with the subscription price versus the one-time price.
If you sell software or plans, there’s a separate set of questions around tiers and defaults, covered well in this guide to A/B testing ideas for pricing pages.
Before you test actual prices
When you’re ready to test real price points, a few guardrails matter more than the tool.
Keep each shopper’s price consistent. A visitor who sees $48 today and $54 tomorrow will feel cheated, and they’d be right. Assignment should stick to the visitor across sessions, and ideally across devices for logged-in customers.
Make sure the price they see is the price they pay. This sounds obvious, but some setups only change the displayed price while the cart and checkout show the original. That’s worse than not testing. Confirm the tested price flows through cart, checkout, order confirmation, and emails.
Check the law where you sell. Consumer protection and price display rules differ by country and state, especially around reference prices and “was” prices. If you’re unsure, get advice before launching. It’s a small cost compared to a complaint.
Brief your support team. Someone will eventually notice a different price on a friend’s phone. Decide your policy in advance. Many stores simply honor the lower price when asked, which is a cheap way to keep goodwill.
Avoid personal-data pricing. Testing price points across random groups is one thing. Charging people more because of who they are or what device they use is another, and it’s the kind of thing that ends up in news stories. Keep assignments random.
Running the test
Price tests usually need more data than layout tests, because the effect on conversion can be small while the effect on profit is large. Use a sample size calculator before you start, run for full weeks, and don’t call the result early. Evan Miller’s classic post on peeking at A/B tests shows how checking results repeatedly inflates false positives, and that risk hurts more when the decision is your price.
Pick your primary metric carefully. I’d use gross profit per visitor if you can calculate it, since revenue alone can reward discounts that destroy margin. Track conversion rate, average order value, and refund rate as supporting numbers. If you can, look at 60-day repeat purchases for each group too, because a lower price that brings in one-time bargain hunters may look good now and worse later.
Avoid major sale periods. Shoppers during a holiday promotion behave differently, so a price test running through one won’t tell you much about normal weeks.
A quick worked example
Here’s a hypothetical to show why the metric matters. Imagine a skincare serum that costs $14 to make and ship. At $48, the control group converts at 3.0%. At $54, the test group converts at 2.7%.
On conversion rate alone, the higher price looks like a loss. Now do the profit math per 1,000 visitors. The control makes 30 orders with $34 of gross profit each, which is $1,020. The test makes 27 orders with $40 each, which is $1,080. The “losing” price actually earns more.
Flip the numbers slightly and the answer changes, which is exactly why you test instead of guessing. It’s also why you should set the metric before launch. If your team agrees on profit per visitor up front, nobody can argue for conversion rate after the fact just because they preferred the lower price.
Reading the result
Say the higher price lowered conversion rate by a little but raised profit per visitor. Most teams would ship it. Before you do, check segments. If the drop came entirely from new visitors arriving through paid ads, you might keep the higher price for returning customers and revisit acquisition offers separately.
And if there’s no meaningful difference? That’s genuinely useful. It suggests your shoppers aren’t very price-sensitive in that range, which gives you room to test a bolder change next time.
Tools that can handle it
Most ecommerce A/B testing software can handle presentation tests, but not every tool can run real price tests. Many can only change what’s displayed. If you’re on Shopify, check that the tool changes the price through the cart and checkout, and ask whether it works with any third-party checkout you use. A Shopify CRO app that supports price and offer tests will save you a lot of custom development.
CustomFit.ai is one of the platforms built for this. It’s a no-code A/B testing, personalization and CRO platform for D2C and Shopify brands, and it supports price, discount and shipping tests natively on Shopify, so tested prices carry through to the cart rather than living only on the product page.
Key takeaways
Start AB price testing with presentation and offers, not list prices. Use shipping thresholds, anchors, and per-unit framing to learn how price-sensitive your shoppers are. When you do test real prices, keep each visitor’s price consistent, make sure the cart matches the page, check local rules, and brief support. Judge results on profit per visitor, run full weeks, and resist peeking. Price is the most powerful lever in your store, so treat it with the most care.
