What the signal means
A public job posting can disclose a tool in a required skill, team description, migration project, integration responsibility, or preferred qualification. Counting those explicit mentions across a bounded company sample creates a reproducible signal of hiring demand for that technology category.
A term match does not prove that a company owns, is buying, or is replacing a product. Keep the matched job URLs and read the surrounding text before using the signal.
Build a useful first input
- Choose one category, such as security tools or database infrastructure.
- Start with three to five companies whose public ATS boards you can identify.
- Use a short dictionary of distinctive product or category terms.
- Cap each company at 100 jobs while checking whether the dictionary is precise.
{
"companies": [
{
"company": "Stripe",
"boards": [{"provider": "greenhouse", "slug": "stripe"}]
},
{
"company": "Cloudflare",
"boards": [{"provider": "greenhouse", "slug": "cloudflare"}]
}
],
"maxJobsPerCompany": 100,
"termDictionary": ["Okta", "Splunk", "Snyk", "Zscaler"]
}Distinctive names are safer than broad words. A term such as tableau may name a product, while security describes thousands of unrelated roles. Review a small output before adding synonyms.
Read the output beside its evidence
| Field | How to use it |
|---|---|
jobsAnalyzed, matchedJobs | The denominator and matched count for this exact run. |
technologyCounts, technologyShares | Explicit term frequencies and their share of analyzed jobs. |
buyerTeams | Rule-based role-family counts for the matched postings. |
demandScore, scoreReasons | A deterministic sorting aid with its calculation reasons, not a probability. |
evidenceJobUrls | The public listings to inspect before drawing a conclusion. |
atsSources, scrapedAt | Source coverage and observation time for reproducibility. |
Compare companies only when the term dictionary, job cap, and observation window are consistent. A company with no returned row may have no match, an unsupported board, or a temporary source failure; absence is not evidence of no demand.
Avoid overclaiming
- Say "public job posts mentioned Okta" rather than "the company is buying Okta."
- Keep product mentions separate from generic category language.
- Open at least one evidence URL for every high-priority company.
- Store the exact input and scrape time with every exported result.
- Use scheduled reruns to observe disclosed changes, not to infer private procurement decisions.
Each workflow is priced at $18 per 1,000 returned company analyses. A four-company result is about $0.072 in Actor charges. Failed fetches, diagnostics, and companies with no qualifying analysis are not billed as result rows.
Choose a focused workflow
Start with the category closest to the question you are answering:
- Security Tool Demand Intelligence for explicitly named security products.
- Revenue Technology Demand Intelligence for sales, marketing, and customer tooling.
- Database Infrastructure Demand for disclosed database and data infrastructure terms.
- Work Management Tool Demand for collaboration and work-management products.
- BI and Analytics Tool Demand for analytics and visualization tools.
Try the security-tool example
Use the public Security Tool Demand workflow with the bounded company sample and security terms above, then inspect the evidence before increasing coverage.
Open Security Tool Demand