Is there real demand for this?

The idea under test

<the idea, in one or two sentences>

Before you spend a call

Write down the four Ps — Todd Jackson's frame for what you are actually claiming. Doing this first is what makes the data falsifiable rather than decorative, because each P names the evidence that would refute it.

  PERSONA  who specifically has this problem — a role, not a market

  PROBLEM  what they are doing today instead, and what it costs them

  PROMISE  the one outcome you would put on the homepage

  PRODUCT  the smallest thing that could deliver that promise

Then write the number that would make you walk away, before you know it. A threshold chosen after the data arrives is not a threshold.

THE EVIDENCE LADDER. Rank every signal by what it would cost someone to fake. Prefer signals higher up, and never let a lower rung outvote a higher one:

1. Money someone is paying a competitor, or sustaining ad spend
2. Salary a company is paying humans to do this job
3. Search commercial-intent queries with real, repeated volume
4. Time people build workarounds — spreadsheets, templates, macros
5. Attention views, engagement, followers, upvotes
6. Words someone said they would like it. Never sufficient on its own.

Rungs 1 and 2 cost money to fake. Rung 6 costs nothing, which is why so much of it exists.

FINDING URLS. Several tools take a URL and cannot search for one — social_linkedin_posts, social_reddit_posts, social_reddit_comments, company_jobs, company_person, web_fetch. When you need a page, find its URL with YOUR OWN web search first, using a site: filter. That costs nothing and is exact. web_search does the same job for a billed call and is the fallback when you have no search of your own. Never construct a URL from a company name: a wrong URL comes back as a dead page, which reads as an absence of evidence rather than as your mistake.

BATCHING. The keyword tools bill once per call however many terms you send — keyword_volume takes 1,000, keyword_intent 1,000, keyword_overview 700, keyword_ideas 200. Pool your terms and send them in one call rather than looping. A batch whose rows exceed one response comes back trimmed and says how many rows it held versus returned: size the next batch from that figure and send the terms you did not get. A term missing from a trimmed response was withheld, NOT measured at zero. Scoring it as zero is the single most common way one of these runs reaches a confident wrong answer.

The method — cheapest killer first

1. Harvest the language. Do not invent keywords, find them.

• keyword_ideas on 3–5 seed terms taken from the PROBLEM, not from your

  product name. Buyers search their problem; only your existing customers

  search your category.

• keyword_suggestions on whichever seed came back richest, for the long

  specific phrases where intent lives.

• If you named a competitor: domain_rankings on their domain returns the

  terms they already win, which is the vocabulary of people who are already

  buying this.

Collect every candidate term. Do not price them yet, and do not filter them by taste — the term you would never have written is the one worth finding.

2. Price the whole set in as few calls as it takes.

• keyword_volume on everything you harvested.
• keyword_intent on the same list. Commercial and transactional intent with

  volume is demand. Informational volume alone is an audience, which is a

  different and much harder business.

KILL CRITERION. If the commercial-plus-transactional cluster totals under ~5,000 searches/month — or under ~1,000 for a B2B tool, where a buyer is worth 5–10× a consumer — stop here and say so. That is a real finding and it cost you four calls. Report it as a finding, not as a failure to find something.

3. Look for money. This is the top rung, so reach it early.

• ads_advertisers on your two clearest competitor names or on the money

  terms themselves.

• ads_creatives on any advertiser it returns. Note first_shown and

  `last_shown` on every creative and compute the run length yourself.

• ads_meta_advertisers on a competitor's domain if the buyer is a consumer

  or a small business, where the spend goes to social rather than to search.

  It dates their Meta ads and hands back a `page_id`; `ads_meta_library` on

  that id renders the creatives.

A creative running 90+ days means their acquisition arithmetic works at that message and that price. Nobody sustains a losing ad for a quarter. Note: in the advertiser results the title field is the ADVERTISER NAME, not the ad headline.

4. Look for salary. A posted band is what this problem is worth today.

• Find open roles with your own web search (`site:linkedin.com/jobs/view

  "<competitor>"`), or `web_search` if you have none.

• company_jobs on the two or three listings that look most revealing. Each

  call reads ONE listing, so choose rather than sweep.

Roles clustered on one function reveal the roadmap. A posted salary band is the price of solving this problem with humans, which is the ceiling on what software can charge for solving it without them.

5. Listen for the four forces.

Bob Moesta's frame: a switch happens when PUSH (frustration with today) plus PULL (attraction to the new) beats ANXIETY (fear of switching) plus HABIT (what they would have to stop doing). Most ideas die on anxiety and habit, which is why "they said they loved it" and "they did not buy it" coexist so often.

• Find where the buyer gathers with your own search, then

  `social_reddit_posts` on the threads, and `social_reddit_comments` on the

  one thread that is clearly the canonical complaint.

• app_reviews with sort_by=most_recent if a mobile competitor exists.

  Read the 1- and 2-star reviews first: a paying customer taking the time to

  complain is the strongest words-rung evidence there is.

Quote them verbatim. The exact phrasing is the finding — a paraphrase loses the word the buyer would type into a search box.

6. Check the trend before concluding.

• keyword_history on your top 5 commercial terms.

Distinguish a durable shift from a spike. A rising term with no advertisers is either early or worthless. Say which you think it is and why, and name what would tell the two apart.

The bar

Validated only when ALL of these hold. If one fails, say which:

• Reach — a commercial-intent cluster above ~5,000/mo, or ~1,000/mo B2B
• Triangulation — it survived a kill attempt from at least TWO independent

  families (search, ads, hiring, marketplace reviews, social)

• Explanation — the story of why this demand exists survived the data
• Forces — you can name the push and the pull, and you have a plan for the

  anxiety and the habit

• Priced — you have a CAC band from real CPCs, not a guess

"Promising" is not a verdict.

RULES THAT MAKE THE OUTPUT TRUSTWORTHY

• Cite the tool and the argument behind every number you state.
• Distinguish measured (a tool returned it) from inferred (you reasoned to it).
• State the strongest evidence AGAINST your conclusion before your conclusion.
• Absence of a signal is not evidence of absence. Name which instrument was

  blind and why.

• Coverage is public data and varies by query, country, and date. A thin

  result is a thin result, not an empty market.

• If the data does not settle it, say INCONCLUSIVE and name the one call that

  would settle it. An unspent budget and an uncertain answer at the same time

  is a failed run; so is a confident answer the data did not support.

Deliver

The verdict. The one number that decides it. The two strongest pieces of evidence for, and the strongest against, stated before your conclusion. Then the single cheapest call that would change your mind — and make it a call this API can actually serve.