They filtered until there was nothing left.

Declared

Facet combination that returns zero, or a grid that dropped from many to none this session.

Example eligibility rule

Trigger when the latest filter reduces a result set from at least 10 items to zero. Preserve every earlier filter and identify the single filter that caused the dead end. A query that started at zero belongs to search recovery instead.

Facets, price, size, dates. The combination returns zero. Default is a blank grid and "try removing a filter". That is a stall mid-job, not a search-results problem. They were close. Then the path died.

Avoid this assumption

Zero-results search is a query problem, covered in case 5. A dead filter is a recovery problem. They narrowed successfully, then one more choice removed everything. Help them back one step.

A close match, a "drop this one facet", or a human is a different experience that is easy to hold out. Null filter is as honest a signal as null search.

Reach

How many eligible people can you reach in the chosen channel and time period?

Cost

What will content, delivery, QA and ongoing maintenance cost?

Return

What incremental outcome would cover that cost, and what is it worth?

Planning inputs, not measured results. Estimate reach × incremental outcome rate × value per outcome, then subtract cost. Use a range and a consistent time period.

Example format: Recommendations

Keep the rest of the filters, offer the nearest in-stock or in-range set, with a reason.

Compare lighter and deeper responses

Choose the lightest response that solves the need.

Assist: Offer useful help

Name the dead facet. One-tap remove. Show how many come back.

Adapt: Change a meaningful part

Keep the rest of the filters, offer the nearest in-stock or in-range set, with a reason.

Transform: Tailor the journey

Recover into a close match plus a way to ask, and remember the combination if they return.

What to watch out for

• Auto-clearing every filter they set.
• Showing out-of-budget items as "close".
• Counting the recovery click as success when the item is still wrong.

Control

Empty grid, generic "no results".

Treatment

Dead-end recovery, with a holdout.

Success measure

Recovered product view or add-to-cart from that session.

Guardrail

People who wanted zero (true miss) being pushed a mismatch.

Among sessions where one filter caused the 10-to-zero drop, compare guided recovery with the standard empty state. Measure recovered detail views and purchases, then inspect whether users immediately restore the removed filter.

Win or lose, record what worked, for whom and why. What you learn becomes the next signal.

• Algolia Hosted search infrastructure with rule-based merchandising and an opt-in machine-learning personalization layer.
• Constructor Ecommerce search and recommendations platform that optimizes recs against a chosen business KPI using clickstream data and merchant rules.
• Coveo Enterprise AI search platform using deep-learning 'product affinities' to personalize ranking, suggestions, and recommendations in real time.
• Bloomreach Ecommerce Search AI-driven ecommerce search and discovery platform (Loomi AI) that personalizes ranking, recommendations, and merchandising from first-party shopper data.
• Athos Commerce Athos Commerce is the January 2025 holding brand created when Searchspring and Klevu combined under PSG ownership, alongside Intelligent Reach. Three separate products with separate contracts and pricing, not one merged platform.
• Algonomy Retail personalization suite (search, recs, dynamic listing pages, content/styling, social proof, shopping agent) built on the earlier RichRelevance/Manthan merger. Now operating under the ADA brand after ADA acquired Algonomy in July 2026.

Search came back with nothing they can use. Show a close match for what they typed.

They searched for something specific. You showed bestsellers. Put what they asked for first.