Offer tests with Freebies Engine

Offer tests compare offers you build in Freebies Engine, so they need Freebies Engine on the same store. Create the offers there first; they then appear in the test.

• Which offers do you want to compare? Choose an offer for each version, or No offer for the Original.
• A draft or paused offer is test only: it runs only for shoppers in its version while the test runs. An offer that is live for everyone can't be in a test, because the test couldn't limit it to one version: pause it first, or choose a copy.
• Gift, discount and free-shipping offers can be tested. Offers with discount codes, bundles and upsells can't yet; they show as not available.
• Each version needs a different offer, and an offer can be in one test at a time.
• When you add product costs, the gift cost counts toward profit. Offer tests are included in Experiment Engine Pro.

When the test ends with a winner, the winning offer goes live in Freebies Engine for everyone. The other offers go back: drafts stay drafts, and an offer that was live is paused. Ending with The original ships no offer and puts them all back the same way. If your Freebies Engine plan has no free live offer slot, the test keeps running: pause a live offer or upgrade, then end it again. Pausing or archiving an offer in a running test pauses the test, and you are asked first; editing it makes results from before not comparable. Only admins can change such an offer, and the test's activity records it.

Test a post-purchase offer

A Post-purchase offer test compares versions of one live post-purchase upsell. The offer keeps running in Freebies Engine, and the test decides which version each order gets. It needs both apps, the Pro plan in each, and a live offer that no other test holds.

• A version changes the product (a whole product, so shoppers choose a variant, or one variant), the discount and its name, the headline and the description. It never changes the product's own price, who sees the offer, or its limits and schedule.
• Amount off only works on orders in your store's own currency. Other orders see the original offer.
• Each order is assigned by its own reference number, on our server, with the shares you set, so the same order always sees the same version. A returning customer can see a different version on a different order. An order is counted once it was shown the offer. Orders that Shopify skips, such as wallet payments, are never counted.
• The goal is Added profit per order shown when your product costs are known, and Added revenue per order shown when they are not. Accept rate is shown next to it, but it shouldn't be the goal on its own, because a very deep discount always wins it.
• A version needs 100 orders shown and 10 accepts before it is compared. The table lists orders shown, accepts, declines, orders that couldn't be completed and the added money per order shown, net of the item's discount, refunds of that line and cancelled orders. Money updates within about 10 minutes, and test orders are left out.
• Pausing the test, uninstalling Experiment Engine or turning on Pause all tests gives every order the original offer. Shoppers never see an error from a test.
• Ending the test with a winner copies that version's product, discount and copy into the offer in Freebies Engine, all at once. If the offer can't be saved, the test keeps running and nothing changes. The change is recorded in the offer's history.