Read the release


Test automatic discounts against no offer, or against each other. Each group gets its own offer at cart and checkout, and you see the lift next to what it cost.

No offer versus 10% off

Sample data
MetricControlVariant
Revenue per visitor$7.47$8.07
Conversion rate3.00%3.60%
Average order value$249.00$224.10
Profit per visitor$4.17$4.11

How it works


Each offer is a Shopify automatic discount, so your existing discounts stay untouched.

  1. 01Draft

    Create your offer test

    Pick an offer for each group: amount off, threshold, volume, free shipping or a free gift. Control can keep no offer.

  2. 02Launch

    Launch with one discount

    ABConvert builds one automatic discount for the test. Preview each group's cart and checkout before launch.

  3. 03Read

    Read the result

    See revenue per visitor, conversion, AOV and profit per visitor for each group, and whether the difference is significant.

Four offer experiments

A discount can win an order that would not have happened. It can also give margin to a shopper who would have paid in full. The test shows which one is happening. Each experiment compares one offer with no offer or with another offer, on Tockle's store: the $249 coffee grinder, $110 in product cost, and its accessories.

A 10% discount wins enough extra orders to pay for itself.

Start with the plainest question: does the discount you plan to run earn more than no discount at all? Control keeps the normal price, so the comparison counts the shoppers who would have bought anyway and now pay less.

Variable being tested

Offer: none → 10% off the grinder.
ControlNo offer
TOCKLECart

Your cart

1

Tockle grinderCream

$249.00

Subtotal$249.00

Total$249.00

Check out
Variant10% off
TOCKLECart

Your cart

1

Tockle grinderCream

$249.00

Subtotal$249.00

10% off−$24.90

Total$224.10

Check out

Keep fixed. Same products, prices, shipping rates, page content and audience. The discount applies only in the variant's cart and checkout.

Preview both versions
Add the grinder in both groups. Confirm $249 in Control and $224.10 in the variant's cart and checkout.
Worked example with sample data
Gross profit per order falls from $139 to $114.10. At equal traffic, 100 orders without the offer need 122 with it to break even. See the full result ↓
When to roll it out
Keep the discount only if profit per visitor rises, not just conversion.

Worked result

+20%orders with 10% off

The discount won 60 more orders and $5,976 more revenue, and still earned less: the 300 shoppers who would have bought anyway each received $24.90 off.

No offer versus 10% off

Sample data
MetricControlVariant
Revenue per visitor$7.47$8.07
Conversion rate3.00%3.60%
Average order value$249.00$224.10
Profit per visitor$4.17$4.11

10,000 visitors per group. Control: 300 orders, $74,700 revenue, $41,700 gross profit. Variant: 360 orders, $80,676 revenue, $41,076 gross profit. Profit uses $110 product cost. All figures are illustrative.

What we do


  • AI and MCP

    Ask Claude or ChatGPT how a test is doing or draft the next one. Actions depend on access you grant.

  • Audience targeting

    Run a test only where it matters: by country, device, traffic source or customer type.

  • Analytics

    See profit per visitor, conversion, AOV and significance for each variant, with product costs included.

  • Visual editor

    Change copy, images, and templates on any page without code. Preview before launch.

Requirements

Every offer is a Shopify automatic discount built with Shopify Functions. Your existing discounts stay untouched, and an offer applies only to visitors in its test group. Offer tests are on every ABConvert plan.

Each group gets one offer type: amount off products, amount off order, shipping discount, volume discount, threshold discount or free gift. Control can keep no offer, or carry its own to compare two promotions.

The test is one automatic discount, so its discount combinations apply to every group. Shopify allows 25 active automatic discounts per store, shared with your other discounts. Offers can't change after launch; close the test and start a new one.