# A/B and outcomes

> Measure size-related returns with and without TryYourSize, with a holdout group and the orders and returns you send.

Source: https://docs.tryyoursize.com/products/ab-outcomes/

TryYourSize's job is fewer size-related returns. To show it with your own numbers:

1. **A/B holdout.** In Studio, start an experiment: a share of visitors (10–50 %) sees no TryYourSize element at all. The SDK assigns each visitor with a random first-party ID; [`tys.experiment()`](https://docs.tryyoursize.com/reference/web/client-experiment.md) tells you the arm.
2. **Orders.** Your server sends each order with [`POST /v1/orders`](https://docs.tryyoursize.com/reference/node/orders-create.md), with the visitor ID saved with the cart. No customer data is accepted.
3. **Returns.** Send returns with [`POST /v1/returns`](https://docs.tryyoursize.com/reference/node/returns-create.md) and their reason (`size_too_small`, `size_too_large`, …).
4. **Results.** Studio → Results compares size-related return rates between the arms. Aggregates only.

Shopify stores get orders and returns automatically from the app.

- [A/B holdout](https://docs.tryyoursize.com/guides/ab-holdout.md): Set it up and keep the holdout clean.
- [Orders and returns](https://docs.tryyoursize.com/guides/outcomes.md): What to send, from your server or as CSV.