Intent Data Playbook
A good account list tells you who could buy. Intent data is the part that tells you who to contact this week. We keep the two apart, and we only act when signals stack.
What it is and when to use it
Intent data is any evidence that an account is moving toward a purchase: research behavior, hiring for a function, adopting or dropping a technology, raising money, a leadership change or a public incident. It can be first party (your own site and product), second party (engagement on a platform you run campaigns on) or third party (data collected elsewhere).
This playbook covers the intent we have run for clients: job postings, technology data, funding and news, and engagement. It does not cover bought third-party topic intent, which we have not delivered. Two kinds of tools are involved: data sources that return raw rows, and signal engines that watch accounts continuously. Most teams need both.
Use this playbook when
- You have a market list but no timing, and need to know which accounts to work now.
- The client's product only matters to companies running a certain technology, which no firmographic field captures.
- A team is about to buy an intent tool and wants to know which signals actually separate buyers.
Tools we use
Only tools we have run this playbook with, and what each one does in it.
- Blitz: Job-posting search, the main source of hiring and technology evidence
- TheirStack: Technographics and jobs; counts are free before revealing records
- Apify: Daily LinkedIn job scrapes and Google searches for news
- OpenFunnel: Signal agents for hiring and technology on a target account list
- Parallel and Crustdata: Funding rounds and product launches
- Claude Code: Reading each posting and article by meaning, and the adversarial review
- Supabase: One table of accounts and signals, plus replies fed back for re-weighting
The steps
The order we work in. Each step feeds the next, and every result is written to the database as it happens.
Score fit apart from timing
Fit (firmographics, technology, buying power) decides the tier. Intent is a separate timing layer on top. Blending them into one number hides both the quiet perfect buyer and the loud account that will never buy.
Weight what the product depends on
For a product that only matters on certain technology, the technology score carries most of the weight. In one model it was 105 of 150 points.
Load the blocklist first
Customers, open deals, partners and competitors are uploaded before any signal runs, so they are never surfaced as new targets.
Act on stacked signals
Every signal lands on one shared account view. Three or more signals earn a manual touch, two earn a sequence and one is held.
Read job postings by meaning
We sweep postings for the relevant tools and roles, then read each posting for what it says about the hiring company, and remove staffing firms in an adversarial pass.
Pilot, review, re-weight
A scoring model runs on 100 companies first. The output is reviewed by hand, weights are adjusted, and only then does it run on everything.
Feed outcomes back
Replies and meetings go back into the database so the weights can be checked against what actually happened.
What good output looks like
What this has produced:
For a security startup, about 3,000 job postings over 90 days became 913 companies, 150 first-pass qualified and 102 on the send list after adversarial review, with no data cost.
In that review, 21% were dropped from the evidence-rich tier and 47% from the thin tail. Review hardest where the evidence is thinnest.
For a data provider, a scoring pilot on 100 companies flagged 15 for outreach, using $40 of research credit.
Common mistakes, and the fix
Each of these happened on a real run. The fix is now part of how we work.
35% of a job-signal universe were staffing firms and consultancies.
- Cause
- They post their clients' requirements as their own jobs.
- Fix
- Read the posting text for the tells, such as "our client", and remove them in an adversarial pass.
Breach keyword alerts were mostly false positives.
- Cause
- Matches on the wrong company or on generic articles.
- Fix
- Have the model classify each article per company and require a company-specific source.
Technology searches returned zero.
- Cause
- Guessed technology identifiers.
- Fix
- Test the identifiers first, using the free counts.
Some searches returned no companies even though jobs existed.
- Cause
- Throttling returned an empty result instead of an error.
- Fix
- Space calls about a second apart, retry with backoff and assert that jobs found means companies found.
A sailing club and an elevator company appeared in a technology list.
- Cause
- A product name that is also an ordinary word.
- Fix
- Treat single-word product names as ambiguous and confirm them in the posting text.
Website technology crawls missed infrastructure tools.
- Cause
- Infrastructure does not show up on a website.
- Fix
- Use job postings for infrastructure technology.
Late-stage funding outranked real buyers.
- Cause
- Funding carried a large multiplier.
- Fix
- Remove the multiplier and weight recency instead.
A high-confidence verdict had no evidence behind it.
- Cause
- A single model pass was trusted as is.
- Fix
- Run an adversarial pass and a live web check before the list is final.
Questions people ask about intent data
What is intent data?
Intent data is evidence that a company or person is moving toward a purchase: research behavior, hiring for a function, adopting a technology, raising money or a leadership change. We use it as a timing layer on top of a fit score, never in place of one.
Can you give an example of intent data?
A company posting jobs that name the tool your product works with. For a security startup we swept about 3,000 such job postings, read each one by meaning, removed staffing firms that post other companies' jobs, and ended with 102 companies to contact.
What is the difference between first-, second- and third-party intent data?
First-party intent comes from your own properties, such as your website and product. Second-party intent is another platform's first-party data shared with you, such as engagement on a platform you advertise or post on. Third-party intent is collected across many outside sources and aggregated or sold.
What are buyer intent signals?
The individual events behind intent data: a relevant job posting, a technology adopted or dropped, a funding round, a new executive, a public incident or engagement with your or a competitor's content. We act when several stack on one account: three or more earn a manual touch, two a sequence, one is held.
Want this playbook running on your market?
Tell us what you sell and who you sell to. We will show you how this playbook would run for you, and what we would check first.
