Every Monday we make the same three-way call on every live offer: keep it, change it, or kill it. This series puts you in the coach's seat. The store and the numbers are illustrative, modelled to look like a normal week on a mid-sized pet brand. The decision is real: it's the one founders get wrong most often.
The basket
The buyer just paid €42 for a forest-green trail harness. On the post-purchase page, before the Thank you page, they see one offer: the matching lead, same colour, same width. Normal price €18, offered at 15% off: €15.30. One tap adds it to the same order.
The week
Here's the scorecard for seven days on harness orders paid by card or Shop Pay, the ones that can see a one-click offer.
| Week 3 · harness orders | Number |
|---|---|
| Eligible harness orders | 470 |
| Held out (10%, saw no offer) | 47 |
| Saw the lead offer | 423 |
| Accepted | 14 |
| Take rate (14 ÷ 423) | 3.3% |
| Add-on revenue (14 × €15.30) | €214.20 |
Illustrative example, modelled figures. Not a client result.
3.3% is below the 4–6.5% we plan with. The lead costs €7.20 to make (40% of €18), so each one sold still adds €8.10 before refunds. Not a disaster. Not a winner either.
Keep, change or kill?
Pick one. No sign-up. Your first pick is final, and we record which one you chose.
Change. Not the offer. The audience.
Split the 423 orders by what was already in the basket and the picture flips:
| Harness orders that saw the offer | Saw it | Took it | Take rate |
|---|---|---|---|
| No lead in the basket | 290 | 14 | 4.8% |
| Already had a lead in the basket | 133 | 0 | 0% |
| All | 423 | 14 | 3.3% |
Illustrative example, modelled figures.
Nearly a third of the offers went to people who had just bought a lead. Of course they declined. For everyone else the offer does 4.8%, inside the range we'd keep. The offer isn't weak. It's aimed at the wrong people a third of the time, and that drags the average down.
The fix is one exclusion rule: if a lead is in the basket, don't show the lead. Show the next thing a new harness owner needs instead. Here, a treat pouch at €9, full price.
Before and after
| Per week | Before | After (modelled) |
|---|---|---|
| No lead in basket: lead offer, 290 × 4.8% | 14 × €15.30 = €214.20 | 14 × €15.30 = €214.20 |
| Lead in basket: 133 orders | Lead offer: 0 × €15.30 = €0 | Treat pouch, 5%: 6 × €9 = €54.00 |
| Add-on revenue | €214.20 | €268.20 |
| Contribution (lead €8.10, pouch €5.40 each) | 14 × €8.10 = €113.40 | €113.40 + 6 × €5.40 = €145.80 |
Illustrative example, modelled figures. 5% of 133 is 6.65; rounded down to 6 to stay conservative. Pouch cost assumed at 40% of €9 (€3.60).
That's +€54 a week in add-on revenue, about €2,808 a year (€54 × 52), from one rule on one product. And the lead offer's reported take rate goes from 3.3% to 4.8%, because it stops being shown to people who can't use it.
Why not keep?
Because a third of the offers were wasted, every week, forever. “It's making money” is how losing offers survive.
Why not kill?
Because the blended 3.3% hides a 4.8% offer. Kill it and you throw away the €214.20 a week it earns from the right buyers. Segment before you kill.
One honest caveat: 47 held-out orders is too few to read lift in a single week. The holdout gets read over several weeks. This call is about who sees the offer, which the basket data answers on its own. More on that in holdout testing for upsells.
The series
Every edition: one basket, one offer, one call. Send us yours. A bestseller, the offer you show after it, and roughly how it's doing. We'll pick one, anonymise it, and make the call in public. Your store name never appears.
Three rules decide every call:
- Keep when it beats the holdout and the offer it replaced, and nothing in the segments looks off.
- Change when it works for one group and not another, or the price or discount is wrong. Most calls are this one.
- Kill when contribution per add-on is negative, or it shows no lift after enough orders. Check it in the AOV calculator first.
Your offer, our call.
Leave your email and bring the offer to a free 15-minute audit. You'll leave with keep, change or kill, and the first test we'd run next.
No setup fee. No retainer. You only pay from the profit we add.
The lesson for your store
Pull last month's orders for your bestseller. How many already had the item you offer after it in the basket? If the answer is more than a handful, you're running this exact offer. Add the exclusion rule today; it's the cheapest fix in this game. More rules like it in Shopify post-purchase upsells, and what a pet store would test next in post-purchase offers for pet brands.
Stop at “thank you”? Or sell one more.
We build, run and test your post-purchase offers for you, and make this call on every offer, every Monday. Live within 48 hours of access and approval.
Free 15-minute call. For Shopify brands doing €100k+ a month.