Post-Purchase Offers: Measure What They Actually Add
Unico Commerce, FounderPost-Purchase Offers: Measure What They Actually Add
Post-purchase offers convert absurdly well. A shopper who has just paid is the highest-intent person on your site, so any offer shown right then posts a take rate that makes the rest of your funnel look broken.
That's the trap. The audience is exceptional — not the offer. Measure post-purchase by take rate or acceptor AOV and you'll conclude it's your best channel and scale it. The truth is usually smaller.
Why take rate lies
Comparing acceptors to decliners proves nothing. Acceptors were always going to spend more — that's selection bias, not lift.
The only number that matters is incremental margin per eligible order (IMEO): the extra contribution the offer creates across everyone who saw it, measured against a control group that saw the same page with no offer.
A worked example
Say 10,000 orders per arm. The offer is a $25 add-on at 60% margin → $15 contribution per accept.
- Treatment: 12% accept → 1,200 accepts.
- Control: no offer; 2% buy the same item through the normal catalog → 200.
- Incremental accepts: 1,000.
Now net it out — including the effects most reports ignore:
| Line | Amount |
|---|---|
| Gross incremental contribution (1,000 × $15) | +$15,000 |
| Upsell returns (10% of 1,200 = 120 × $15) | −$1,800 |
| Base-order refund lift (+0.4pp × 10,000 orders × $80 × 40% margin) | −$1,280 |
| Net incremental contribution | +$11,920 |
| Per eligible order | $1.19 |
Still a good offer — but it's $1.19 per eligible order, not the "12% take rate" headline. And note the third line: post-purchase offers can move whole-order behavior, not just the add-on.
The refund adjustment everyone skips
Post-purchase offers create a specific distortion: they can trigger whole-order friction — extra shipping decisions, buyer's remorse, payment confusion. In the example, the offer's own returns cost $1,800 and the base-order refund lift cost $1,280. Skip either and you overstate IMEO.
Always compute per eligible order, net of refunds. Never per acceptor.
Three traps to pre-register
- Payment friction. One-click accept vs. re-enter-card behave very differently. Randomize the mechanism, not just the creative.
- Reloads. Thank-you page refreshes must not double-count accepts or re-randomize. Dedupe by order ID.
- A refund guardrail. Pre-register a kill threshold — e.g., pause if the base refund rate rises 0.5pp or more. The offer isn't free if it destabilizes the main order.
What to do
- Design for the post-purchase moment: low-risk add-ons, warranties, replenishments — not a deep discount that cheapens what they just bought.
- Keep a persistent 5–10% holdout so "does it still work?" is always answered.
- Report IMEO net of refunds as the headline. Take rate is a vanity metric.
The takeaway
Post-purchase is one of the best places to make money and one of the easiest to fool yourself. A clean test separates the two in weeks.
Aim the offer at the order that just converted, prove it against a control, and measure the contribution you kept — not the one acceptors grabbed.
