Randio shows a lot of numbers. Four of them decide what you do next; the rest are context. This article says which is which, where each one lives, and how to tell when sending more has stopped paying.

Where the numbers are

  • The campaign page shows the run itself: how many it reached, how many emails were sent and are waiting, and the orders it produced.
  • Orders lists every paid order that used a Randio discount, with the campaign, the customer and the saving. Filter by campaign, group, product or date.
  • Analytics summarises orders over a date range, with a scope switch between all Randio orders and one campaign. Three tabs: Orders, Products and Customers.
  • Products turns it around: per product, what sold through Randio, which variants and options, and the discount depth it sold best at.

Everything counts paid orders only and shows revenue before refunds. A refund later does not take the order out; Shopify is the record for that.

The four that matter

Orders per catalog sent. Not the conversion rate of a discount; the yield of a run. Take a campaign's paid orders from Analytics and the catalogs it sent from the campaign page. This is the number to compare across groups and across runs of the same group, because it already contains everything: who opened, who clicked, what they saw, what it cost to show them.

Average order value, per campaign. It tells you whether the per-product limit and the range are doing their job: a personal offer should lift what a buyer spends, not just whether they buy. If it falls below your store's normal order value, the catalog is selling the discount, not the products.

Discount dependency. The Analytics chart of order volume by the discount applied. It is the only direct evidence of what your range is doing. Orders clustered at the ceiling mean the floor was decoration; orders spread across the range mean the floor earns its keep. Read it after every change to a strategy.

Discount depth by product, on the Products tab and the Products page. The depth each product actually sold at is the best guide to next run's pools: products that sold shallow go in Guaranteed pools at fixed, low percentages; products that only sold deep go in Limited pools, capped.

The ones to treat with care

Opens. Mail clients open emails on the customer's behalf, so the open count overstates attention. Use clicks and the add-to-cart step of the funnel instead; opens are a ceiling, not a signal.

Catalog views. A view is one device loading the catalog page in a day. Views of the storefront block are not counted, so a storefront campaign's views understate its reach.

Unattributed orders. An order marked "Randio discount · unattributed" used a discount that could not be linked to a run. It is listed on Orders and left out of Analytics. A run of them usually means the cart embed was off at the time.

Revenue as a total. It is gross, it is paid orders only, and it is the campaign's, not the store's. Compare it to the same campaign's last run, not to a sales week.

When sending more stops paying

There is no curve to look up, because there is no data about your store yet. There is a way to find the edge in two or three runs.

An Outreach run sends to the most engaged members of a group first, inside a daily limit, and a larger group adds its less engaged half. So the yield of the next hundred catalogs is always lower than the yield of the last hundred. Run the same strategy to your top group, then to the next group down in your group order, and compare orders per catalog sent. When a group's yield drops to a fraction of the one above it, that is the edge of where this strategy pays, and the catalogs below it are better spent on a different strategy for a different kind of customer.

The same test works within one group by capping it. Cap it at half, run, lift the cap, run again, and compare the two yields. Group caps and ordering are explained in Segment before you randomize; the rotation that lets you repeat the comparison on a schedule is in Campaign cadence.