One question, right after the box lands.
Most post-purchase flows are built to extract: a review request, a cross-sell, a referral ask. Open ten accounts and you’ll find those three in various orders. What you almost never find is the one email whose only job is to listen — a short note, just after the order arrives, asking whether it arrived right.
That email is the cheapest insurance a brand can buy, and hardly anyone sends it.
The check-back
Every decent restaurant runs this play. Two minutes after the plates land, someone comes by: how is everything? Not at the end of the meal, when the damage is done and the tip is decided — two minutes in, while the cold plate can still go back to the kitchen.
The timing is the whole trick. Ask too early and there’s nothing to report. Ask too late and the problem has already hardened into an opinion. Restaurants learned the window the hard way: the diner who was never asked doesn’t complain to the waiter. They complain in the review, where the kitchen can’t fix anything.
Orders work the same way. Between the box landing and the customer’s next public statement about your brand there is a window — usually a day or two — where a problem is still a private matter between you and them. The delivery check-in is the check-back, run inside that window.
Why the silent customer is the expensive one
The customer who emails you about a bad order is doing you a favour. They’re giving you the chance to fix it while it’s nobody else’s business. I’ve written before about what happens in public when that chance is missed — the short version is that the same defect becomes either loyalty or a permanent warning, and response speed is the only variable.
But the check-in exists for the customer who wasn’t going to email. Most unhappy customers don’t complain to the brand at all. They quietly don’t come back, and some of them describe why in places you’ll never be notified about. A one-line email that says we want to know moves a measurable share of those silent exits into conversations you can actually resolve.
The build
The flow is deliberately boring. Its power is in existing, not in cleverness.
Trigger: the delivery event, not the order event. Shopify fulfilment events carry it if your carrier updates tracking; apps like AfterShip or Wonderment send a clean Delivered metric into Klaviyo. Triggering on “shipped + N days” is the fallback when no delivery data exists, but real delivery data is worth the app.
Delay: one day after delivery. Same logic as the waiter — after the first bite, before the opinion sets.
The email: plain text. No template, no hero image, no products. Sender is a person, not “noreply”. One question: “Did everything arrive the way it should? If anything’s off, just reply — a real person reads these.” Sign it with a name. The email should look like it took ninety seconds to write, because emails that look designed read as marketing, and marketing is exactly what this must not be.
Exclusions: anyone with an open support ticket (sync a profile property from your helpdesk), anyone who already got one in the last 30 days, and obvious non-events like digital goods.
The part everyone skips: the replies
The flow is the easy half. The build fails at the inbox, so decide three things before switching it on:
Who owns the replies. A named person, checking at least daily. If replies land in an unwatched inbox, the flow is worse than nothing — you asked, then went silent, which is the restaurant asking about the meal and walking away mid-answer.
What they’re allowed to do. The owner needs authority to fix without escalating: replace, refund, or credit up to a limit, same day, no interrogation and no receipts-and-photos obstacle course for a first offence. Set the limit against customer lifetime value, not unit cost — arguing over a $30 replacement with a customer who orders monthly is the most expensive $30 the brand will ever save.
What gets recorded. Tag the profile: issue type, resolution, date. Over months, this becomes a defect log your ops team will actually use, sourced from the people who matter most.
Measuring a flow that attribution can’t see
This flow will look like a failure in every revenue report, because its wins are invisible to click-through attribution: the refund that didn’t become a one-star review, the quiet churner who got caught and kept. Measure it on its own terms:
- Reply rate, and the share of replies flagging a real problem — that share is your defect rate, measured better than any survey.
- Repeat-purchase rate of recovered customers against customers who reported nothing. In my experience the recovered group holds its own and sometimes wins — a fixed problem is proof the brand stands behind the product, which no flawless order ever demonstrates.
- Review trajectory after launch. The flow’s revenue shows up as the bad reviews you stopped getting.
The take
Before the loyalty program, before the referral engine, before another campaign: send the one-line email that asks. It’s an afternoon to build — one trigger, one delay, one plain-text send — and a standing decision about who answers. The check-back has kept restaurants in business for a century on exactly this economics: the cold plate is cheap while it’s still on the table, and very expensive once the diner has left.