Buy 3 Get 1 Free in the Cart Drawer: 2.3% Took the Offer, Revenue per Visitor -2.8%
[ SUMMARIZE WITH AI ]
[ FREE CRO TEARDOWN ]
Find the 3 biggest revenue leaks on your store.
Every day a conversion leak goes unfixed, you're paying for traffic that doesn't buy. Get a 5-minute Loom through your PDP, cart, and checkout, with mockups of the fixes. No pitch.
Get My TeardownA buy 3 get 1 free offer sounds like an easy add-on. In this cart drawer test, almost nobody took it, and revenue did not move.
A golf equipment brand sells clubs and training gear, plus its own golf ball. The ball already had a buy 3 get 1 free deal on its product page. The idea: bring that buy 3 get 1 free offer into the cart drawer, so every shopper buying a club sees it right before checkout.
After 19 days, 2.3% of shoppers who saw the offer added it. For shoppers who opened the cart, revenue per visitor was -2.8% against control, with a 27% probability to beat it. We recommended stopping the test.
The Offer We Put in the Cart Drawer
- Where: cart drawer, below the cart items
- Setup: control plus one variation, in Intelligems
- Length: 19 days, about 1,400 to 1,500 orders per version
| Version | Below the cart items |
|---|---|
| Control | The store's existing upsell app widget |
| Variation 1 | A "Complete your setup" block: ball photo, a one-line pitch, three benefit bullets, "Buy 3 dozen, get 1 dozen free", the struck-through price, a per-ball price and one add to cart button. The upsell widget was hidden so the two did not stack |
One click added the 3-dozen pack, and the store's own bundle logic added the free dozen, the same way it works on the product page.
94 of 4,078 Shoppers Took the Deal
| Metric | Result |
|---|---|
| Shoppers who saw the offer and added it | 2.3% (94 of 4,078) |
| Revenue per visitor, cart openers | -2.8% vs control |
| Probability to beat control | 27% |
Earlier in the test the gap was bigger: -8.5% at 8% probability to beat. It narrowed as orders came in, but it never turned positive. With this much volume, a flat-to-negative read is a real answer, not a lack of data.
Why the Buy 3 Get 1 Free Offer Didn't Land Here
Three reasons, from our read of the data:
- Wrong moment for a big add-on. 3 dozen balls is a real purchase on its own. Asking for it on top of a club order is asking for a second decision at checkout
- The deal was not new. The same offer already lived on the ball's product page. Shoppers who wanted balls had a path to it
- It replaced something. Hiding the existing upsell widget meant the test measured "this offer vs that widget", not "offer vs nothing". The widget was doing some work already
The take rate says the most. When fewer than 1 in 40 shoppers who see an offer act on it, the offer is not the lever for that page.
What This Means for Bundle Offers in the Cart
Bundle deals work best when the shopper is already shopping for that product. A buy 3 get 1 free offer on the product page meets someone deciding how many to buy. The same offer in the cart drawer meets someone who came for something else.
We have seen the same pattern with a different brand: accessories in the cart drawer, ranked by what people buy together, cut revenue per visitor by 12%. The next test on this account moves the ball offer to a one-time vs subscription choice, still in the cart, to see if a different deal changes the take rate.
Buy 3 Get 1 Free Offer Questions
Does a buy 3 get 1 free offer work in the cart drawer?
Not in this test. Only 2.3% of shoppers who saw it added it, and revenue per visitor came in at -2.8% with a 27% probability to beat control, across about 1,400 to 1,500 orders per version.
Where should bundle offers go instead?
On the product page of the product being bundled, where the shopper is already choosing a quantity. In the cart, offers compete with the checkout decision.
Is 2.3% a good take rate for a cart cross-sell?
It is low for a deal this strong. It told us shoppers buying a club were not in the market for a large ball order at that moment.
Why stop a test that is only -2.8%?
Because it had enough orders to show there was no upside coming. A test that sits near flat after 1,400+ orders per version is taking a slot another test could use.