A/B testing for e-commerce brands
How DTC brands should A/B test: ad landing pages, bundles, subscribe and save and first-order offers, the traffic each test needs, and why revenue decides.
Updated 28 September 2026 · 4 min read
Start where paid traffic lands: the ad landing page, the product page and the offer. Judge every test on revenue per visitor from your store's orders, not on ad clicks, email signups or add-to-cart rate. A brand with 200,000 visitors a month can confirm a 10% revenue lift on a store-wide change in about a month; the same change on one product page takes most of a year.
A direct-to-consumer brand buys most of its traffic, which changes what you should test. Every visitor from an ad has already cost you money, so the page they land on and the offer they see decide whether that spend comes back. The test that matters is the one that moves revenue per visitor on paid traffic.
Start with the ad landing page, then the offer, and judge everything on the revenue your store records.
Why testing is different for DTC brands
Traffic is expensive and concentrated. Most of it arrives on a few pages from a few ads. That's good for testing, because those pages get enough visitors to measure, and bad for guessing, because a weak landing page burns ad budget every day.
Offers move two numbers at once. A 3-pack, a free gift above $70 or a subscribe and save discount changes both how many people buy and how much they spend. Conversion rate alone misses half of that. Revenue per visitor catches both.
Some revenue comes later. Subscribe and save orders repeat, and first-time buyers come back. A test window of four weeks sees the first order, not the second, so tests that change repeat behavior need a longer read afterwards.
Margins vary by offer. Revenue per visitor tells you which version brought in more money. Whether that money was profitable after discounts and shipping is a check you run yourself on the winner.
The tests to run first
1. A dedicated landing page for your top ad
Send traffic from your biggest ad to a page built around that ad's promise, against the standard product page. Measure revenue per visitor from that ad. The landing page guide covers what to put on it.
2. The 3-pack shown first
On a consumable product, show the multi-pack as the first option with its per-unit price, instead of the single. Measure revenue per visitor and units per order.
3. Subscribe and save as the default
Pre-select the subscription option, with the one-time purchase clearly visible next to it. Measure revenue per visitor, then keep reading subscription revenue for 60 days after the test, since the value is in the second and third orders.
4. First-order offer
Test a 10% welcome discount against a free gift of similar cost, or against no popup at all. Measure revenue per visitor, not email signups.
5. What it does, above the fold
Move the section that explains how the product works (ingredients, materials, sizing) above the fold on mobile. Measure revenue per visitor.
6. Ad copy angle
In Meta ads, test copy that leads with the problem the product solves against copy that leads with the product. Measure revenue per ad visitor from your store, not click-through rate. See Meta ads A/B testing.
7. Your guarantee next to the button
If you offer a returns or money-back guarantee, state it in one line under the add-to-cart button. Measure revenue per visitor. Don't test a guarantee you don't honor.
How much traffic you need
A made-up example. Your store gets 200,000 visitors a month, about 6,700 a day. 2.5% of visitors order, the average order is $80, and most orders fall between $20 and $140.
A store-wide test, such as a new product page template, uses all of that traffic. To detect a 10% lift in revenue per visitor at 95% confidence and 80% power, you need about 96,500 visitors per version, 193,000 in total. That is about 29 days.
Now run the same change on one hero product page that gets 20,000 visitors a month. The same 193,000 visitors would take about 290 days.
| Where the test runs | Visitors a month | Days to detect a 10% revenue lift |
|---|---|---|
| Store-wide template | 200,000 | about 29 |
| One product page | 20,000 | about 290 |
Judging on conversion rate alone would need about 64,200 visitors per version for the same lift. The revenue test needs roughly 1.5 times that, because order values vary. It is still the right test, because conversion rate can rise while revenue falls. Check your own results with the revenue per visitor calculator, and see what a win is worth with the split test ROI calculator.
Common mistakes
- Judging ad tests on click-through rate or cost per click. A cheap click that doesn't buy is not a win.
- Using the ad platform's reported purchases instead of your store's orders. The store is the source of truth for revenue.
- Shipping a discount that won on revenue and lost on margin.
- Running a test across a sale week, then applying the result to full-price weeks.
- Pre-selecting a subscription without making the one-time option obvious. It can raise revenue in the test and refunds and chargebacks after it.
- Testing on one product page when the template would reach every product.
How Outtest fits
Outtest connects read-only to your store (Shopify, or a payment tool such as Stripe or Paddle) and your analytics, finds the pages and ads where the most money leaks, and builds and runs tests like these. It installs through the Shopify, WordPress, Webflow or Wix apps, Google Tag Manager, or one pasted line.
Each test is judged on revenue per visitor from your orders. The Builder agent never invents claims, discounts or reviews, and any change to pricing or checkout waits for your approval. A winner needs at least 7 days, a 90% chance of beating the original and a lift of at least 10% (defaults you can adjust). After a win, 5% of visitors keep seeing the original, so you can see what the winner actually earned over the following months.
Questions people ask
What should a DTC brand A/B test first?+
The page your paid traffic lands on, then the offer. Test a dedicated landing page against the product page for your main ad, then test bundles and a subscribe and save default. These change what people buy and how much, so they can move revenue enough to measure.
Should e-commerce tests be judged on conversion rate?+
Judge them on revenue per visitor. Bundles, thresholds and offers often move order value and conversion rate in opposite directions, and only revenue per visitor shows the net effect. Check margin separately for any test that changes a price or discount.
How do I test a first-order discount popup?+
Split visitors between the popup and no popup, or between two offers, and judge on revenue per visitor over the test. A bigger discount usually brings more email signups, which says nothing about whether the discount paid for itself.
How long should an e-commerce A/B test run?+
At least one full week to cover weekday and weekend shoppers, and until you reach the sample size your traffic needs. Avoid running a test across a big sale, since sale shoppers behave differently from everyone else.
Read next
Let Outtest run your split tests
AI agents read your analytics and payments, find where you lose the most money, build the fix and test it. Every test is judged on revenue, not clicks. Plans from $29 a month.