How to see Shopify Rollouts experiment results by revenue
Short answer: The built-in Rollouts experiment view reports rates — conversion rate, bounce rate, add-to-cart rate and reached-checkout rate. It doesn't show net sales, average order value or revenue per session for each variant, and the metrics can't be customized. To compare variants by money, you need a view that splits sales by variant. Upshot adds that view to every experiment: net sales, discounts, profit and revenue per session, side by side. For discount experiments, Shopify doesn't attribute revenue to variants at all, so Upshot shows a clearly labeled estimate when the setup allows it.
What the built-in view shows (as of October 2026)
| Experiment type | Metrics shown |
|---|---|
| Experiment (A/B test) | Conversion rate over time · Bounce rate · Reached checkout rate · Add to cart rate |
| Checkout changes | Checkout conversion rate |
Shopify's documentation notes that you can't customize the available metrics, and that discount-specific metrics aren't available — even for rollouts that include a discount.
Why revenue per session matters more than conversion rate
Conversion rate counts how many visitors bought. It doesn't count how much they spent or how much you gave away to get the sale.
- A variant with a bigger discount can convert more visitors and still bring in less money.
- A variant with fewer but larger orders can lose on conversion rate and win on revenue.
Revenue per session (net sales ÷ sessions) combines both. It's the number most teams use to decide which variant to ship.
What to compare for each variant
- Sessions and conversion rate — is there enough traffic to trust the result?
- Net sales and average order value — did people spend more or less?
- Discounts — how much did each variant give away?
- Profit — what was left after product costs and discounts?
- Revenue per session — the single number to rank variants.
With Upshot
On Upshot's Pro plan, open any experiment and these five are already there for every variant — as a labeled estimate for discount experiments, where Shopify doesn't break revenue down by variant — with a plain-language verdict (“Too early to call. About N more days needed.”, “B is likely converting better”) and, once there's enough data, the probability that a variant converts better. Orders not yet matched to a variant are shown separately, so you know when the data is incomplete.