Search data is the largest honest survey ever fielded — if you read it correctly. Aggregate signals, seasonal ghosts, and the difference between interest and intent.
Every day, people describe their unmet needs to a search box, in their own words, without a moderator in the room. Aggregated and read with care, that's the largest and most honest demand survey in existence. Read carelessly, it's a machine for generating plausible nonsense. The difference is method.
First, the boundary: serious search intelligence works with aggregate, anonymized signals — volumes, co-occurrence patterns, trends. Individual query histories are personal data, and nothing in this post requires them. Aggregation isn't just compliance; it's also the only level at which the signal is stable enough to mean anything.
How people search for a category is itself market structure. Query clusters reveal the axes shoppers actually care about — for coffee machines it might be {type}, {feature}, {use case} — and the hierarchy of those clusters tells you which axis dominates the purchase conversation. When your product naming speaks a different language than the demand, you pay for that mismatch in every funnel stage.
Most "emerging trends" are seasonal ghosts wearing a trendy mustache. Before declaring growth, compare against the same weeks in prior years, and score trends on two axes: velocity (rate of change) and persistence (how long the elevated level holds). Durable demand shifts score high on both; ghosts score high on velocity and collapse on persistence.
The step most search analysis skips: reading demand against the market's response. Rising interest with no competitive supply is an opening; rising interest already met by three launched product lines is a signal you're late. This is where search intelligence earns its keep when joined with product intelligence — the gap between what people ask for and what the market offers is where roadmaps live.
Search data tells you what people want to know. Only supply-side data tells you what they can actually buy. Intelligence lives where the two disagree.
— Elena Rossi
Handled this way, search data stops being a marketing dashboard and becomes what it really is: a continuously fielded survey of unmet demand, free of the framing biases that plague every questionnaire.
Merchant taxonomies describe how retailers shelve products. Search language describes how shoppers think. Where the two disagree, you'll find assortment, naming and SEO decisions waiting to be made.
Seasonal ghosts with trendy mustaches — I’ve worked with at least four of them. The deseasonalize-first rule should be law.
2 comments
(Team note) The velocity/persistence scoring mentioned here is what powers the trend panels in our search intelligence reports.