Seven pitfalls that make search merchandising less effective

1. A commercial boost is too broad

A general boost for a brand, high margin, or campaign category can hit many more queries than intended. A product that is commercially attractive is not automatically relevant to every search query. This can cause the exception of exact product names, model codes and specific long-tail queries to deteriorate.

Limit a rule to relevant candidates. Combine the boost with category, query type, brand context, inventory or minimum relevance threshold. Thus, merchandising changes the order within an appropriate set, rather than pushing irrelevant products into the set.

Test wide and specific queries before going live. Check not only the campaign search term, but also strong existing terms that touch the same products or category. A regression set makes side effects visible.

2. Temporary rules do not have an end date

Campaigns, seasonal products and temporary stock promotions have a beginning and end. Without an end date, boosts and banners remain active after the reason has disappeared. The consequences are not always immediately visible: outdated rules can clash with new campaigns months later.

Make start and end moment part of the rule, not a separate calendar. Add owner and target and let expired rules automatically disable. Save them for evaluation when audit information is needed, but prevent inactive configuration from rendering the management environment unreadable.

Schedule a brief evaluation before the end. Has the campaign reached the intended target group? Were there any side effects on relevant queries? That information helps better organize the next campaign rather than copying the same line blindly.

3. Ranking ignores stock, delivery time or store context

A promoted product that is not available can take valuable top positions. Sometimes showing is logical, for example with expected supplementation or when an alternative is available. The problem arises when that choice is not conscious and status information is updated too late.

Take sales, stock and delivery time with you in a way that suits the category. A B2B-reserve part with longer delivery time can still be the right exact match; an impulse product without stock may require a lower position. Do not use a single rule for the entire range when buying behaviour differs.

Multi-store and international environments need additional context. Stock, price, assortment and campaign may vary by store. A rule should only be active where its conditions apply. Therefore, always check store, language, currency and market configuration.

• Is the product marketable in the active store?
• Is stock current enough for ranking decisions?
• Should an exact match remain visible despite longer delivery time?
• Is there a relevant alternative when the product is not available?
• Does the commercial rule apply in every market and language?

4. The team sends everything manually

Manual pin positions and exceptions give direct control, but scales poorly. When a fixed order is built for each important query, search hardly responds to assortment, behavior and product data. Every change requires human attention.

Use the basic ranking for general relevance and merchandising for clear commercial exceptions. Automate rules where logic is stable and explainable, for example, weighing marketability within relevant results. Keep manual pinning limited to situations where an exact position is really needed.

Make sure a recurring manual correction masks a more fundamental problem. If teams keep boosting the same brand because product titles are incomplete, a product data correction is more obvious. A workaround can be helpful, but should have an owner and expiry date.

5. Campaign visibility is confused with search relevance

A banner, promotional tile and product boost have different features. A banner can provide inspiration or campaign context without greatly changing the organic product sequence. A boost affects which products are perceived as most relevant. Choose the instrument that fits the goal.

Do not put a promotion between results when it interrupts the customer task or resembles an advertisement without being clearly recognizable. Make commercial content relevant to the query and transparent in presentation. On mobile, space is limited; test whether the promotion does not completely push the first usable products out of the picture.

Measure campaign visibility separately from organic search quality. So you can optimize a campaign without hiding a lower relevance score behind more banner clicks.

6. Success is measured only by clicks

More clicks do not automatically mean better merchandising. A striking product can attract a lot of attention but then show low add-to-cart or conversion. It can also displace organic products that deliver fewer clicks but more value.

Choose metrics that fit the target and add guardrails for search quality. For a campaign, you can track visibility, click rate, and conversion, while simultaneously monitoring zero-result searches, first organic click, and strong reference queries. With small numbers, you report absolute volumes and uncertainty.

Compare with a relevant basis. A before-and-after period can be usable when other factors are similar; a controlled experiment provides stronger evidence when infrastructure and volume allow. Always record price, stock and campaign differences.

7. Rules have no owner or governance

Without governance, merchandising becomes a historical archive. No one knows why a rule exists, who assesses the result or whether the original campaign is still relevant. Colliding rules make ranking difficult to explain and delay incident investigation.

Set minimum goal, owner, scope, priority, start, end and evaluation moment for each important rule. Use roles and approval where multiple stores or teams are hit. A support or analysis team may need reading permissions without changing configuration.

Make conflict behavior predictable. If pinning, boost, bury, stock and personalization play at the same time, it should be clear which layer is given priority. Show why a product is in a certain position. Explainability is not a luxury; it is necessary to be able to scale commercial control safely.

• Purpose and hypothesis of the rule.
• Owner and authorized changers.
• Query, category, store and market reach.
• Start, end and evaluation moment.
• Priority in conflict and possibility to reverse.
• Result measurement and controlled reference queries.