What are Search & Discovery Anomalies?
When a shopper uses your site search, they're telling you exactly what they want to buy. These are your highest-intent visitors. If search fails them — irrelevant results, missing products, confusing navigation — you lose the most motivated buyers on your site.
Search & Discovery Anomalies automatically detect three types of search friction that were previously invisible: Search Abandon (users view search results but never click a product), High-Intent Underperforming (users click a search result and view the product but don't add to cart), and Search Loops (users search repeatedly in rapid succession, unable to find what they need). Each anomaly type generates a full report with per-query breakdowns, behavioral signals, session recordings, and recommended actions.
These anomalies are detected automatically using the same statistical model as existing behavioral anomalies — comparing daily search metrics against a rolling 7-day baseline. When a metric deviates significantly, the system generates a report pinpointing which queries, devices, and segments are affected, along with estimated revenue at risk.
The Three Anomaly Types
Anomaly Type | What It Catches | Example |
Search Abandon | Users visit search results but don't click through to any product — results are irrelevant or missing | A shopper searches "hiking boots," sees results, but never clicks a product |
High-Intent Underperforming | Users click a search result and view the product but don't add to cart — the product doesn't match what they searched for | A shopper searches "vacuum bags," clicks a vacuum cleaner result, bounces without adding to cart |
Search Loops | Users search 4+ times within 30 seconds — frustration, inability to find what they need | A shopper searches repeatedly for "ski goggles" with different refinements in rapid succession |
How to View Search & Discovery Anomaly Reports
Navigate to Behavioral Intelligence in your dashboard
Search & Discovery Anomalies appear alongside existing behavioral anomaly reports. They're generated automatically when search metrics deviate from baseline.Look for reports tagged with the Search & Discovery category
Each report is labeled with the specific anomaly type — Search Abandon, High-Intent Underperforming, or Search Loops — so you can quickly identify what kind of search friction was detected.Review the Anomaly Snapshot
The top section shows which metric changed, when the anomaly started, its current status, severity, and estimated revenue at risk.Drill into per-query breakdowns
The report ranks the specific search queries driving the anomaly by revenue impact. You'll see the search-to-PDP rate, add-to-cart rate, and session volume for each query compared to site averages.Review behavioral signals and session recordings
Each anomaly includes behavioral context — dead clicks, rage clicks, and scroll depth on search results pages — plus linked session recordings showing real user behavior.Act on recommended fixes
Every report includes specific recommended actions based on the anomaly type, such as reviewing search ranking logic, auditing mobile search layout, or checking catalog indexing.
Where It Works
Surface | Supported |
Main Site (via FERMÀT Pixel) | ✅ |
Dynamic Product Pages | ✅ |
Landers | — (search is a main site behavior) |
Product Funnels | — (search is a main site behavior) |
Use Cases
Catch search ranking regressions before they impact revenue
When a site search algorithm update quietly degrades results for high-volume queries, Search Abandon anomalies flag the drop in click-through rates within 24 hours — before the revenue impact compounds over weeks.
Identify product-query mismatches driving PDP bounces
High-Intent Underperforming anomalies reveal when shoppers click search results but leave without adding to cart. The per-query breakdown shows exactly which queries are returning products that don't match shopper expectations.
Diagnose mobile search UX failures
Search Loops concentrated on mobile devices often signal that the search results layout is harder to scan, filters are inaccessible, or results require too much scrolling. The device-level segmentation surfaces these patterns immediately.
Prioritize search merchandising improvements by revenue impact
Every anomaly report ranks affected queries by estimated revenue at risk — calculated from session volume, expected conversion rate, and average order value. This gives merchandising teams a data-backed priority list for search optimization.
FAQs
How are search pages detected?
Search pages are identified through URL patterns — paths containing /search, and URL parameters like ?q=, ?query=, ?keyword=, or ?search=. This works automatically across most ecommerce platforms without additional configuration.
What baseline is used for anomaly detection?
Each search metric is compared against a rolling 7-day average with a 2-standard-deviation threshold, the same statistical model used for all behavioral anomalies. This adapts to your site's normal patterns rather than using fixed thresholds.
Do non-product searches (like "returns" or "size chart") trigger false positives?
The system accounts for this. Non-product searches with naturally low click-through rates are flagged as content gap opportunities rather than search quality issues in the report's recommended actions.
Can I see session recordings for search anomalies?
Yes. Every search anomaly report links to session recordings showing the actual user behavior — a shopper scanning search results and leaving, clicking a result and bouncing from the PDP, or looping through multiple searches in frustration.
Does this work with AJAX-based search that doesn't change the URL?
The current version relies on URL-based search detection. A pixel enhancement for AJAX search handling and zero-result detection is planned for a future release.