Sorting by margin: the sort order shoppers never asked for
Shoppers read sort order as advice
Most shoppers never touch the sort control, which makes the default order the strongest merchandising statement on the page, stronger than banners, stronger than badges. They read the first screen as the store's honest answer to their category intent. Fill that screen with high margin lines they have to scroll past and you have taught them, in about four seconds, that this page is arranged for you, not for them. Some leave. More just convert worse, in a way that looks like weak traffic or weak product in every report that does not isolate sort as a variable.
The tell in your own data is a category with healthy traffic, healthy product, and a first screen full of items with low click through relative to their position. Position bias means top items should earn clicks. When they do not, the ranking is wrong for the shopper, whatever it did for margin.
Where margin legitimately works
Placement is not sort. A featured slot, clearly one slot, inside a relevance ordered page is normal retail and shoppers understand it. Margin belongs in tie breaking between near equal products, in which variants get photographed and described properly, in bundle composition, in the products your email and ads push, and in delisting decisions about lines that earn neither margin nor traffic. It also belongs in search boosting with a cap, so a high margin product can rise within its relevance tier but cannot leapfrog products shoppers demonstrably want more.
What margin should never do is operate invisibly and without a ceiling. If you cannot state the rule in a sentence you would be comfortable printing on the page, the rule is the problem.
Test it the way an operator would
Run the honest comparison: relevance led default against your margin influenced default, same traffic split, long enough to cover your normal buying cycle, and judge it on revenue per session and margin per session together, not margin per order alone. Margin per order almost always flatters the margin sort, because it ignores the sessions that bought nothing. Margin per session is where the left orders show up.
Most stores that run this test end in the same place: shopper led sort, margin in the tie breaks and the featured slot, and a short list of hero lines that get placement because they earn it twice, once in margin and once in demand. That is not ideology. It is just where the money was.
Related reading: Should clearance products rank next to full-price products?, How should out-of-stock products rank while they are unavailable?.
Frequently asked questions
Is sorting by margin ever the right default?
In a pure clearance context, where the page's job is to exit stock and shoppers arrived expecting exactly that, margin and exit value can lead. On an ordinary category page serving ordinary intent, the evidence from operator testing consistently favours shopper led order, judged on margin per session rather than per order.
How is a featured slot different from margin sorting?
Scale and honesty. One or two placed products inside an otherwise relevant order reads as an editorial choice, and shoppers discount it accordingly. Rewriting the whole order by margin changes every position, including the ones shoppers trusted most, and the trust cost lands on the page as a whole.
What should the default sort actually be on a small catalogue?
For most small catalogues, a curated best sellers order, refreshed on a cycle, beats both pure algorithmic relevance and margin order. It reflects real demand, it is stable enough for returning shoppers to navigate, and a human can spot and fix its embarrassments, which an opaque formula never confesses.
Can I boost new products without breaking the sort?
Yes, with a time boxed boost inside a relevance tier and a review date attached. New lines need exposure to earn data. The failure mode is the permanent boost nobody removes, quietly fossilising last year's launch priorities into this year's first screen.