Short answer: installing the app is the easy part. The rest is weekly work: picking offers, setting rules, checking payment methods, running tests and measuring them against a holdout. Do it yourself if you're under €100k a month, sell a simple catalogue, or have someone in-house who loves testing. Otherwise hand it to someone whose only job it is.
Why upsell apps stall after week one
The pattern is familiar. You install the app, pick an offer, switch it on. The first week the dashboard lights up with accepted offers. A month later the line is flat, and nobody has opened the settings since launch day. The app didn't break. It just stopped being anyone's job.
Four things cause the stall, almost every time:
1. Offer fatigue
Your returning customers see the same offer on every order. The buyer who declined the matching lead in March declines it again in April, and May. The offer that was relevant once becomes wallpaper. Without rotation, the people most likely to buy again are the ones you stop selling to.
2. One generic offer
The setup wizard asks for one product, so the bestseller goes in. But the bestseller doesn't belong with every basket. Someone who just bought a tent needs the footprint for that tent, not your most popular mug. One offer for everything means the wrong offer for most orders. (The difference matters: upsell vs cross-sell.)
3. Nobody owns it
Marketing thinks it's an ecom setting. Ecom thinks it's a marketing test. The founder installed it and moved on. No owner means no tests, no rules, no one noticing when a product goes out of stock and the offer quietly stops showing.
4. No holdout, so nobody knows if it works
App dashboards count every accepted offer as revenue the app created. Some of those buyers would have bought the item next month anyway. Without a group of orders that never sees an offer, you can't tell lift from noise, so you can't tell a winning offer from a losing one. Which means you can't improve it. Method: holdout testing for upsells.
The seven jobs, and who does them
Here's what running post-purchase offers properly takes, every week. The hours are our rough estimate for one person running one test a week on a mid-sized catalogue. Launch weeks take longer. Your mileage will vary; the jobs won't.
| Job | DIY: you or your team | Managed: us |
|---|---|---|
| Pick the offers | Match an add-on to each bestseller. Rough: 1 h a week | We do it from your order data |
| Write the copy | Headline, product line, price display per offer. Rough: 30 min | We write it |
| Set rules and suppression | No lead offer if a lead is in the basket; out-of-stock checks. Rough: 30 min | We set and maintain them |
| QA on payment methods | Test orders on card, Shop Pay, PayPal after every change. Rough: 30 min | We test before anything goes live |
| Orders paid with iDEAL, Klarna, Apple Pay | Usually skipped | Offered on the thank-you page with a quick second payment |
| Run weekly tests | Plan the next offer, price or sequence; launch it. Rough: 1.5 h | A new test every week |
| Measure with a holdout | Keep 10% of eligible orders out; compare revenue per order. Rough: 1 h | Built in from day one |
| Report | Write up what won and what's next. Rough: 30 min | Scorecard in your Slack every Monday |
| Total | Roughly 5.5 hours a week | 5 minutes to approve, then read the scorecard |
Rough estimate, not a measured benchmark. Assumes one test a week.
5.5 hours a week is about 24 hours a month (5.5 × 52 ÷ 12 = 23.8). Not a lot for one dedicated person. A lot for a founder who also runs ads, hiring and the warehouse. The hours are rarely the problem. The problem is that the hours get skipped, and a skipped week is a test that never ran.
Already running an offer? Bring it.
We look at your current offer, your payment mix and your bestsellers, and show you the first test we'd run against what you have today.
No setup fee. No retainer. You only pay from the profit we add.
When DIY is the right call
We'd rather tell you now than waste your call. Run it yourself when one of these is true:
- You're under €100k a month. At that volume there aren't enough orders to call a weekly test with any confidence. A holdout of 10% on a few hundred eligible orders is mostly noise. Pick one solid offer, set suppression rules, check it monthly. That's enough.
- Your catalogue is simple. One hero product with one obvious add-on? There's not much to test beyond price. Set it, check it, leave it.
- Someone in-house loves testing. If you have a growth person who reads holdout results for fun and has the hours, give them the job and our guides. You'll keep the learning in-house.
If that's you, here's the DIY kit: how post-purchase upsells work, why your upsell isn't showing, how to run a holdout, and the AOV calculator to check an offer makes money before it goes live.
Managed wins when the opposite is true: €100k+ a month, a catalogue with a real “next item” for many products, and a team with no one whose job it is. That's most of the brands we talk to.
What each one really costs
DIY
- The app subscription. Whatever your plan costs, every month, whether the offers earn or not.
- Your team's time. The roughly 5.5 hours a week above, from someone whose hours are worth something elsewhere.
- The tests that never run. The quietest cost and the biggest. Every week without a test is a week the offer doesn't get better.
- The orders that never see an offer. Most setups skip buyers who paid with iDEAL, Klarna or Apple Pay entirely.
Managed
- No setup fee. No retainer. The software is included.
- A share of the extra revenue our offers add, measured against the holdout, agreed per store on the audit call and invoiced monthly.
- If the holdout shows no lift, there's nothing to share. You only pay from the profit we add.
The difference in one line: DIY is a fixed cost with an uncertain return. Managed is a variable cost that only exists when there's a return to share. Neither is free. One of them can't lose you money on fees.
How to switch without losing your baseline
The worst way to switch is to rip out the old offer, launch new ones and compare this month to last month. Seasonality, ad spend and stock all move at the same time, and you learn nothing. Do it like this instead:
- Write down where you are. Your current offer, its price and discount, and the numbers your app shows for the last 30 days. Export or screenshot them; check what your app keeps once it's switched off.
- Know the constraint. Shopify lets one app own the post-purchase page at a time (Settings → Checkout). So the old and new setups can't run side by side on that page.
- Make your current offer the control. We rebuild your incumbent offer and run it as the baseline arm. Every new offer has to beat it, not beat nothing.
- Hold out 10% from day one. One in ten eligible orders sees no offer at all, so we can show what the offers add in total, not just which one wins.
- Give it enough orders. In a low-volume week the holdout can be inconclusive. We say so on the scorecard and keep running rather than calling it early.
Kickoff is within 24 hours of the audit call; first offers are live within 48 hours of store access and your approval. Month to month, 30 days' notice. If the numbers don't add up for your store, we'll tell you on the call.
FAQ
Can I run my current app and a managed setup side by side?
Not on the post-purchase page. Shopify lets one app own that page at a time. Your current offer becomes the baseline instead: it runs as the control while new offers are tested against it, with a 10% holdout that sees no offer at all.
How much of my time does managed take?
A 15-minute onboarding call, one look at the offers before they go live, and a weekly scorecard in Slack every Monday. No calls unless you want them.
What does managed cost compared with DIY?
DIY costs the app subscription plus your team's time. Managed has no setup fee and no retainer: a share of the extra revenue the offers add, measured against the holdout, agreed per store on the audit call and invoiced monthly.
When should I keep doing it myself?
Under €100k a month in revenue, with a simple catalogue, or with someone in-house who enjoys testing and has a few hours a week for it. Our guides cover the whole method.
What happens to orders paid with iDEAL, Klarna or Apple Pay?
They can't see Shopify's one-click post-purchase page. Most setups skip them. We don't: those buyers get an offer on the thank-you page and pay for it with a quick new payment. More in our Thank you page guide.
Am I locked in?
No. Month to month, with 30 days' notice.
Stop at “thank you”? Or sell one more.
We build, run and test your post-purchase offers for you. Live within 48 hours of access and approval. No setup fee, no retainer.
Free 15-minute call. For Shopify brands doing €100k+ a month.