Open your email platform and a notepad. Leave every dashboard closed for now.

Ten checks, 0 to 2 points each. Score 0 if it failed or you can't answer it, 1 if it's partly true, 2 if it's clean. Don't read the scoring bands until you've scored all ten.

Before you score, skim the ten check names and write down the two you expect to fail. The same checks fail at a furniture showroom, an inherited retail file and a regulated cannabis brand. Category won't protect you.

Check 1 runs first, from memory, because it tests what your team carries in its head. A number that lives only in a report isn't steering any decision, so the gap between the guesses and the file is the finding.

If you had to go looking for the answer, score it zero.

The ten checks

1. The one-and-done rate, from memory · 4 minLook at: Nothing yet. Ask the founder, the growth lead and whoever runs email what share of customers have bought once and never again, and write the answers down before anyone opens a report.
   Good: Three answers within five points of each other and of the file.
   Cost if wrong: In an invented example, the room guesses 40% and the file says 70%. This year's acquisition budget was set for a business that doesn't exist.
   Read next: Your One-Time Buyers
2. Your mail arrives · 6 minLook at: Google Postmaster Tools for domain reputation and user-reported spam rate, your spam and unsubscribe rates, and click rate by recipient domain, never open rate. After that, send one flow message to seed addresses at Gmail, Outlook, Yahoo and iCloud.
   Good: Nothing in spam or junk at any provider. Gmail in Primary, Promotions or Updates, all three of which are inbox. Postmaster Tools reputation High or Medium, spam rate under 0.1%. Click rate within a few points across Gmail, Yahoo, Outlook and Apple domains.
   Cost if wrong: Mail in the spam folder still counts as delivered, so a program can report steady opens while Gmail users never see it. Since September 2021, Apple Mail can load your images whether or not anyone opens the email, so open rates overstate engagement; judge engagement on clicks, site visits and orders.
   Read next: Prove the Mail Arrives
3. Repeat rate by first product exists, with n · 4 minLook at: A table of second-order rate by the product on each customer's first order, with the customer count beside every rate.
   Good: It exists, every row has its n, and someone has read it this quarter.
   Cost if wrong: You rank entry offers on cost per first order and never see who came back.
   Read next: Cost per Returner
4. Duplicate and overlapping flows · 4 minLook at: Every live flow with its trigger and its first three subject lines, sorted by subject line.
   Good: One flow per trigger, and no two flows that can fire on the same event for the same person. Each flow excludes anyone already in an overlapping one.
   Cost if wrong: A $21.8K-a-month brand ran eight flows, and four of them sent the identical 17-email sequence on different triggers. Nothing stopped one customer from entering more than one.
   Read next: Structural Share
5. Audiences with no consumer · 4 minLook at: Every list and segment by size, with the flow or campaign that uses it written beside it. Filter flows to Draft as well, and note any whose trigger is a populated list or segment.
   Good: Every list or segment over 1,000 profiles has a named consumer or gets deleted. No draft flow holds a live audience: publish it or detach the audience.
   Cost if wrong: At the same brand, the largest segment held 18,184 profiles and fed a flow that had never been switched on. Another 10,732 subscribers sat on orphaned lists, receiving nothing.
   Read next: The Data Contract
6. Flows ranked per recipient · 5 minLook at: Is revenue per recipient, on a click-based attribution window, the default sort in the report your team reviews? Check the window setting first.
   Good: No marketing flow with more than 1,000 sends and zero attributed revenue, unless its job isn't revenue. Sunset, re-permission, how-to-use, review-request and gift-check flows are judged on their own outcome: suppression carried out, reviews collected, a return to the site. Everything else earns or gets turned off.
   Cost if wrong: At the same brand, email earned $1,416 of $21,812 in monthly revenue, or 6.5%. One post-purchase flow brought in 96.5% of that, and no campaign had ever been sent.
   Read next: Attribution Isn't Proof
7. Winback before replenishment · 3 minLook at: The trigger delay on your winback flow and on your replenishment flow, beside the median days between a customer's first and second order.
   Good: Replenishment fires first, and both delays come from the order file, never a platform default.
   Cost if wrong: The winback tells a healthy customer they've lapsed and hands them a discount on an order they were about to place at full price.
   Read next: The Kept Interval
8. Discount depth · 4 minLook at: Discount depth by month: total discounts as a share of gross sales, and the share of orders carrying any discount (code, automatic, markdown or subscription price), each type on its own line. The trend, not the average.
   Good: Both lines flat or falling, with full-price orders still arriving between sales.
   Cost if wrong: When the discounted share climbs and order count doesn't, each sale is lowering the price on orders that were coming anyway.
   Read next: Margin Is a Retention Metric
9. Subscription share of consumable orders · 3 minLook at: Of orders containing a consumable, the share placed on subscription. Divide by consumable orders, not all orders, or your durable catalog hides the gap.
   Good: You know the figure to the point. Every consumable with a predictable runout offers subscribe, listed first and never selected by default.
   Cost if wrong: A product with a predictable runout, left on manual reorder, turns every second order into a memory test. Customers who fail it don't complain; they stop ordering.
   Read next: Subscriptions
10. A randomized holdout exists · 3 minLook at: A group of customers, chosen at random and fixed, who get no marketing email or text, and a date on which someone reads it.
    Good: It exists and holds 5–10% of customers, or more on a small file. A leak check shows nobody in it was mailed last month.
    Cost if wrong: Every result your platform reports is attribution, and you can't say how many of those orders would have come anyway.
    Read next: Attribution Isn't Proof

Read your score

Fix the low checks in check order, because the early ones produce what the later ones read. A flow can't be ranked per recipient until you know the mail arrives, and a discount can't be priced until you know who came back. Score all ten again each quarter, with the same three people answering check 1.

Do this

• Score all ten. Write the total and your two lowest checks at the top of one page. Hand it to whoever owns the second order before Monday.