Merchandising
Date: 2026-08-16
Deciding what customers see first. It’s the highest-frequency commercial lever you have — changed weekly, by hand, usually with no measurement — and it moves margin mix more reliably than pricing does.
What it is
Merchandising is the arrangement and promotion of products: collection ordering, featured placements, homepage modules, search ranking, cross-sells and recommendations.
It’s a commercial decision executed as a design one, which is why it so often sits with nobody who’s measured on margin.
Position determines sales
Attention falls sharply with position. Roughly, on a category listing:
position share of category clicks
1–4 ~40%
5–12 ~30%
13–24 ~18%
25+ ~12%
[CHECK: the distribution varies enormously by category, layout and device — measure your own rather than using these. The shape is reliable; the numbers are illustrative.]
The first row of a collection does a disproportionate share of the business. Which means the ordering rule — whichever it happens to be — is a commercial decision made continuously, and usually by default.
The default is rarely right
Common ordering rules and what they optimise:
| Rule | Optimises | Problem |
|---|---|---|
| Best selling | Revenue, and itself | Self-reinforcing; new lines never surface |
| Newest first | Freshness | Ignores margin and stock entirely |
| Price ascending | Conversion rate | Systematically depresses AOV and mix |
| Manual | Whatever someone decided | Doesn’t scale, drifts, undocumented |
| Margin-weighted relevance | Contribution | Needs per-product margin data |
Price ascending as a default is the most damaging, and it’s common — it pushes every customer towards the cheapest line and permanently shifts Basket Composition down.
Best-selling sorting has a subtler problem: it’s a feedback loop. Products sell because they’re first and are first because they sell, so the ranking measures its own history rather than customer preference.
Merchandising for contribution, not revenue
The move that matters: rank by expected contribution rather than expected revenue.
product price contribution conversion expected contribution
A £30 £6.00 4% £0.24
B £50 £15.00 2% £0.30 ← rank first
Product A converts twice as well and is worth less per visitor. Ranking by conversion rate or by revenue puts A first; ranking by contribution per view puts B first.
That requires per-product margin in the merchandising system, which is the practical blocker — it usually lives in finance and not in the platform.
Where it interacts
- Stock. Promoting an item about to sell out wastes the placement and creates a poor experience — Stockouts and Availability
- Returns. Promoting high-return lines raises volume and lowers net contribution — Return Rate and Reverse Logistics
- Category pages are the highest-traffic template most sites never optimise — Category Page Design
- Search and filtering. Internal search results are merchandising too, and are almost always ordered by relevance alone — Faceted Filtering
Measuring it
Merchandising changes are made constantly and measured almost never, which makes them the largest source of unexplained metric movement in retail.
- Annotate every significant change. Collection reorders, homepage module swaps, ranking rule changes — Annotation and Change Logs
- Test ranking rules, not just individual placements. It’s a site-wide change with enough traffic to power a test properly
- Watch contribution per session, not conversion rate. A merchandising change that lifts conversion by promoting cheap lines is a loss
- Track category share of contribution against share of revenue — the divergence tells you where merchandising is working against margin — Basket Composition