A shopper who opens your site already knows what they want. Somewhere between knowing and checking out, a large share of them leave. The interesting question is what happens in between, because it is usually not indecision.
Across 50 documented studies, the Baymard Institute puts average online cart abandonment at 70.22%. That is the whole of e-commerce, and grocery has its own dynamics, but the headline is a useful anchor: most assembled intent does not become an order.
Grocery abandons for a different reason than retail
In most categories, the funnel is a decision funnel. The shopper is choosing between options, and abandonment usually means the choice went the other way, or the shipping cost surprised them at the end.
Grocery is not a decision funnel. It is an assembly funnel. The shopper mostly knows what they want before they arrive, and often wants close to the same things they wanted last week. The hard part is not choosing. The hard part is finding thirty items, one at a time, each requiring a search, a scan of results, a size check, a quantity decision, and a click.
That is the assembly tax, and it compounds. A friction cost that is trivial per item becomes decisive across a full basket. Nobody abandons a grocery cart on item three. They abandon on item nineteen, because the session started to feel like a chore rather than an errand.
What the tax actually looks like
Watch a real weekly shop on a conventional storefront and it decomposes into roughly the same loop, repeated: recall an item, translate it into a search term the catalog will accept, evaluate results that are ranked for a different purpose, pick a size, set a quantity, add, return to the list you are holding in your head.
Two of those steps deserve attention because they are invisible in analytics. The shopper is translating their intent into your catalog’s vocabulary, and they are holding the remaining list in working memory. Neither shows up as a metric. Both are where sessions die.
This is also why conventional conversion work has limited headroom here. Faster search, better autocomplete, and cleaner product pages all reduce the cost per item. None of them reduce the number of times the loop runs.
What changes when the store assembles the cart
The alternative is to let the shopper state the whole need once, and have the storefront do the assembly against the live catalog: match items, handle quantities, respect dietary limits, and produce a cart the shopper edits rather than builds.
We can report what that looked like in our own pilot with one grocer, and we will be careful about how much weight it carries. In that pilot, shoppers reached a finished cart in a median of roughly two and a half minutes, some in under ninety seconds. Add-to-cart ran around 88%. The median journey to checkout took about four prompts, and a large majority of steps added seven or more items at once. Session-to-order conversion ran far above the store’s normal e-commerce baseline.
Those are directional results from a single pilot, not a benchmark, and not a forecast for your traffic. What they do suggest is where the headroom sits: not in making each of thirty loops slightly cheaper, but in removing most of the loops.
How to size this for your own store
You do not need a vendor’s numbers to estimate whether this is worth testing. You need four of your own, and you probably already have three.
Your session-to-order conversion on the existing storefront, as the baseline everything is measured against.
Median time from first session action to completed order, which is the closest proxy you have for the assembly tax.
Average items per basket, because the tax scales with basket size and grocery baskets are large.
What share of sessions are repeat shops rather than first visits, since repeat weekly shops are the ones the assembly tax punishes most.
Then run the comparison on a defined slice of traffic rather than in a spreadsheet. The point of a contained pilot is that the intent-to-cart gap is an empirical question about your shoppers and your catalog, and it is cheaper to measure it than to argue about it.