Almost all of the loss happened before shoppers reached the basket.
Investigate and test the view-to-basket stage before optimising checkout.
A team can spend time improving checkout without reaching the people who never put anything in their basket. I wanted to locate the largest break in the observed journey before recommending a change.
I used SQL to follow each visitor from a product view to a later basket action and transaction. That distinction matters: counting repeated clicks as different shoppers would give a different answer.
Only 2.35% of product-viewing visitors reached the basket. Among those who did, 30.22% went on to transact. That puts the earlier stage first on the investigation list.
I would combine this pattern with product-page research, then test a specific change. These event records alone cannot tell whether price, product information or another factor explains the drop.
The source contains 2,756,101 events from 3 May to 18 September 2015. The funnel counts visitors: 1,404,179 viewed, 32,934 then added to a basket, and 9,952 then transacted. Of visitors reaching the basket, 30.22% continued to a transaction. Events with tied timestamps have no within-timestamp ordering. No campaign, margin or experiment data was available, so this is a diagnostic prioritisation, not a causal claim.