Reading a Zero Results Log: The Weekly Triage That Fixes Search
Every store's search log contains a list of phrases real customers typed, with money in hand, that returned nothing. The zero results report is the most actionable document in ecommerce and the one most often exported, glanced at, and forgotten. Triaged weekly, in the right order, it becomes a work queue where almost every line is either a ten minute fix or a buying signal.
Sort before you fix anything
Raw zero results lists are dominated by noise: gibberish, pasted URLs, a competitor's product names, one customer's repeated typo. Group by normalized phrase, count frequency, and sort descending. The top of the list is where customers agree with each other, and agreement at volume is a catalogue or vocabulary fact, not an accident.
Then classify each phrase into one of four buckets: we sell it under another name, we sell it and search failed anyway, we do not sell it but adjacent demand exists, or noise. Each bucket has a different owner and a different fix, which is why the undifferentiated list never gets worked.
Bucket one and two are this week's work
Another name means synonyms, redirects, or title enrichment: add the customer's word to the product's searchable text so the next searcher lands. Sold it but search failed is worse, because it means indexing, tokenization, or a filter default is hiding live inventory, and those bugs tax every query, not just the logged ones. Reproduce them in the search console yourself before assigning the fix.
Both buckets should be cleared weekly while the list is short. A zero results backlog that ages a quarter becomes archaeology: the demand moved on and took the lesson with it.
Bucket three is a buying meeting
Repeated searches for products you do not carry are the cheapest range research available. Customers are telling your search box what your assortment is missing, in their own words, with frequency attached. That report belongs in front of whoever buys or builds the range, monthly, with counts.
The discipline is not to fake it. Do not redirect a missing product's query to a loosely related item to make the zero results number prettier. The customer can tell, the bounce teaches your ranking the wrong lesson, and the honest gap in the report was worth more than the salvaged click.
Close the loop in the log itself
Mark each phrase with its disposition and date, and recheck the fixed ones a week later to confirm they now return the right product, not merely a product. A triage that does not verify its fixes quietly rots, as catalogue changes rebreak synonyms and redirects nobody owns.
Thirty minutes a week, same time, top of the list down. Stores that run it for a quarter describe the same result: the list gets shorter, stranger, and more valuable, because it stops repeating the basics and starts surfacing real shifts in how customers talk about the category.
Keep reading
Related guides on this site: Search Synonyms: The Quiet List That Decides Whether Your Store Search Understands Anyone Seasonal Boosts in Search Ranking: Turning Them On Is Easy, Expiring Them Is the Job Practical answers about personalization.
Frequently asked questions
What is a zero results search?
A customer search that returns no products. Logged and grouped, these queries show vocabulary mismatches, search bugs hiding real inventory, and demand for products the store does not carry.
How often should zero results be reviewed?
Weekly works for most stores, thirty minutes triaging the highest frequency phrases first. The fixes are usually synonyms, searchable text, or indexing bugs, and the list stays short when it is worked weekly.
Should zero result queries be redirected to similar products?
Only where the match is genuinely what the customer meant. Redirecting a missing product to a loose substitute hides a real assortment gap and usually bounces, teaching the ranking system the wrong lesson.