Upsells vs no upsells: +1.86% revenue per visitor, not significant
Test run byKasper DolkMaison Commerce
Commentary byLiftArchiveOur analysis, not theirsKasper Dolk tested adding upsells against a no-upsell control, and revenue per visitor rose 1.86%, from $1.46 to $1.49. Profit per visitor rose 1.80%, estimated at +$17,775 per month. The confidence ranges in the screenshot still include a loss, so this is a lean, not a proven win.
- Revenue per visitor: $1.46 to $1.49 (+1.86%), range -3% to +7%. Estimated +$26,471 per month.
- Profit per visitor: $1.02 to $1.03 (+1.80%), range -3% to +7%. Estimated +$17,775 per month.
- Conversion rate: 1.498% to 1.524% (+1.73%), range -2% to +5%. That is 5,974 vs 6,076 orders.
- Average order value was flat: $97.75 vs $97.88.
- Each range includes zero, so the lift may not be real.
The change
Upsell offers added; the real placement was not shared, so the cart is a stand-in.
Simplified mockup drawn by LiftArchive from the shared screenshots.
Control
Shoppers saw no upsell offers. The test group was labeled No Upsells.
Variant
Upsell offers were added. The post does not show where they appeared or which products were offered.
Why it likely won
Our reading of the result, not the practitioner's claim.
- the gain came from slightly more orders, not bigger ones, since AOV barely moved.
- upsells that add little to the basket may not shift order value enough to show a clear effect.
- judging on profit per visitor guards against upsells that win on revenue but lose on margin.
Caveats: The test ran on about 398,600 visitors per arm, but all three confidence ranges cross zero. Duration and upsell placement were not shared.
Who ran it
Kasper Dolk (@kasper_dolk) shared this test on X. The numbers and screenshots are theirs.
View the original post on X
How to run this test on your store
- Hypothesis
- If we add upsell offers, then profit per visitor will rise because some shoppers add an extra item without hurting conversion.
- Primary metric
- Profit per visitor
- Guardrails
- Conversion rate
- Average order value
- Refund rate
- Who should test it
- Stores with complementary, decent-margin add-ons and enough traffic to detect a 1-2% change. Low-traffic stores will struggle to reach a clear answer on an effect this size.
Questions
Do upsells increase revenue?
In this test, upsells lifted revenue per visitor 1.86% across about 797,000 visitors. The confidence range ran from -3% to +7%, so the gain is not proven.
Do upsells raise average order value?
Not here: AOV was $97.75 without upsells and $97.88 with them. The small lift came from slightly more orders.
What metric should you use to judge an upsell test?
Kasper Dolk used profit per visitor as the main metric, which rose 1.80%. It catches upsells that add revenue but cost margin.
@joshtCRO