Step 3 of the insight workflow · live public data + labeled synthesis

Demand Miner

Search demand is the largest continuous record of how people describe their own needs. Most companies read it only as an ad bill. Type a brand or a category, and this page runs the operator’s analysis on it: live public interest in the topics around your category, the ads competitors are running in it right now, the query patterns demand usually arrives through, and the spend decisions that should terminate at a person. Steps 1 and 2, the evidence and the segments, live on the Patient Psychology Map.

What is live and what is not: topic interest is fetched from Wikipedia’s open pageviews API while you wait. Query patterns and the spend table are labeled synthesis: the frames of the method, not your data. Your input goes to Wikipedia’s public API and nowhere else; nothing is stored. Competitor ads open in the platforms’ own transparency libraries, in a new tab.

Deterministic: the same inputs always produce the same analysis. Live data refreshes; the frames do not reshuffle.

Public interest: nearest topics to “”

live · wikipedia’s closest article matches to your words · pageviews monthly, trailing 12 months

Fetching live interest…

How demand phrases itself

live · google suggest · real queries around “”

fetching live suggestions…

The category voice

live · public feeds · filtered for language and content

fetching recent posts…

The ads competitors are running, right now

live link-outs · the platforms’ own ad-transparency libraries, seeded with your words

Every major platform publishes the ads it runs, and most operators never look. These links open the primary sources directly.

creative teardown · method, on the primary source

Read longevity, not claims

“Top-converting” is not public data: conversion numbers live inside advertisers’ accounts, and the paid tools that claim them sell estimates. The honest public proxy is longevity: nobody pays to keep a losing ad running for months. In the libraries above, scan for the ads that have run longest, and for the five oldest record:

  • Running since: the start date the library shows; the whole signal
  • Hook: the first line, before the reader chooses to care
  • Offer: what is actually promised, in their words
  • Proof: what evidence is shown, if any
  • Call to action: what the click is framed as

Five rows of that table is a real competitive creative read, from primary sources, in fifteen minutes.

How demand likely enters this category

labeled synthesis · the patterns are the method; the queries are illustrative

pattern 1 · symptom-first entry

People search the problem, not the product

Implication: if the category pages speak product language, this demand lands on someone else’s explainer. A symptom-first landing page is usually the cheapest test in the queue.

pattern 2 · comparison and trust

The consideration set forms in public

Implication: comparison queries are where positioning gets decided without you in the room. One honest comparison page tends to beat bidding on every variant.

pattern 3 · post-purchase demand

Existing customers are searching too

Implication: post-purchase queries are retention demand wearing acquisition clothes. Serving them in-product beats paying to re-reach your own customers.

The spend table an operator audits

labeled synthesis · an illustrative allocation for this category · flags mark where audits go

CampaignShareCAC indexStatus
Non-brand: high-intent product termsscale candidate
Non-brand: symptom and problem termstest coverage
Brand defenseaudit: organic cannibalization
Competitor termsusually the first proposed cut
human gate · always

The rule: the machine mines and proposes; budget moves and sensitive-category calls terminate at a person. The same boundary as the care team assistant on this site, drawn where a mistake would land on someone.