Waterfall ICP · a search that knows when to stop
How to find the right decision maker in every company
The short answer
To find the right decision maker in every company, don't search for one perfect job title. List the titles you would accept, from your ideal buyer down to the founder, and search them in that order inside each company. The search keeps the first match and stops there, so a 15-person startup and a 5,000-person company each give you their real decision maker.
On this page10 sections
- How do teams usually find decision makers?
- Why does one job title miss so many decision makers?
- Why doesn’t a broader search fix it?
- How do you find the right decision maker in every company?
- Try it: one list, six companies
- How do you tell the decision maker from a champion or an influencer?
- What does a complete targeting system look like?
- How do you run Waterfall ICP with the API?
- Sources and method
- Video transcript
Say you sell to marketing teams. In every company on your list, you want one person: the decision maker, the one who owns marketing. It sounds simple, and the usual way of searching keeps getting it wrong. Here is why, and how Waterfall ICP, the BlitzAPI people search built for this, fixes it, with the same examples as the video.
- Definition Decision maker also called economic buyer
- The person who can say yes to a purchase and owns the budget for it. In a marketing deal, that is whoever runs marketing in the company: a CMO in a large one, a Head of Marketing in a mid-size one, often the founder in a startup.
PreciselyIn a B2B buying committee, the decision maker, or economic buyer, holds final approval and the budget. Champions push for the purchase, influencers shape the choice, and gatekeepers control access; none of them can sign alone.
How do teams usually find decision makers?
For one account, you check the company’s team page or search LinkedIn by job title. For a whole list of accounts, you open a people search: company filters on one side, job titles on the other. Then you try to fit your whole targeting strategy into a single query.
Company filters
- IndustrySoftware Development
- Employee range11-50 +4
- HQ regionNORAM, EMEA +2
- KeywordsB2B SaaS
People filters
- Job titleCMO, VP Marketing
- Job levelC-Team, VP
- LocationWorld
- Job functionAdvertising & Marketing
1 query8 filters
That query is a bet. The next two diagrams show how it loses.
Why does one job title miss so many decision makers?
Start clean and search for CMOs only. The list looks great: senior and relevant.
Now look at what’s missing. Northwind Logistics calls the role Head of Marketing. Acme has a Marketing Director. At Initech Cloud, 15 people, the founder still runs marketing. Three companies with the same need, and your search found nobody in any of them.
- VP Operations
- Sales Manager
- Head of Marketing
- VP Sales
- CTO
- Marketing Director
- Engineer
- Engineer
- Designer
- Sales
The person who owns marketing, under another title
Why doesn’t a broader search fix it?
So you widen the search to every title with CMO, Marketing or Growth in it. Now you find someone almost everywhere.
Then you run it on Globex Analytics, 5,000 employees. Twenty of the forty people in its marketing team match. Which one is the buyer? Someone on your team now has to sort that out by hand, company after company.
Your query
- CMO
- Marketing
- Growth
- assistant
- intern
- junior
CoverageSomeone found almost everywhere
Marketing team
- Growth Analyst
- Marketing Manager
- Content Manager
- Product Marketing Manager
Returned by the search Not returned
Go narrow, lose the market. Go broad, lose the signal.
The filters aren’t the problem. One query is being asked to understand thousands of different org charts.
How do you find the right decision maker in every company?
Change the logic. Instead of one query, write a ranked list of the people you would accept, from your first choice (P1) to your last resort (P6). In each company, the search tries P1 first. Nobody? It moves to P2, then P3, and it stops at the first person it finds. That is a waterfall search, and the search that runs it is Waterfall ICP.
- Definition Waterfall ICP also called a prioritized title search
- A people search that tries your list of job titles in order, inside one company, and stops at the first person it finds.
PreciselyThe BlitzAPI endpoint POST /v2/search/waterfall-icp-keyword takes a company's LinkedIn page and 1 to 10 ranked levels of title phrases, and returns up to max_results people, each tagged with the level that found them (icp) and a relevance rank (ranking).
- P1CMO · VP Marketing
- P2Head of Marketing · Marketing Director
- P3Head of Growth · Growth Lead
- P4Marketing Manager
- P5Coordinators · broader roles
- P6Founder
Where the search stops, company by company
- Northwind Logistics320 employeesInes DuarteHead of MarketingStopP2
- Acme140 employeesDaniel ChoMarketing DirectorStopP2
- Brightline40 employeesJordan HayesMarketing ManagerStopP4
- Harbor Labs25 employeesLeo BrandtMarketing CoordinatorStopP5
- Initech Cloud15 employeesAva MorenoFounderStopP6
One list. One best person per company.
A Marketing Manager is noise in a 5,000-person company and the right buyer in a 40-person one. Same playbook, different company, different answer.
Try it: one list, six companies
Pick a company and compare the three searches. Watch where the cascade stops.
Try it · one list, six companies
Globex Analytics5,000 employees
- CMO onlyMaya Chen, CMORight person
- Every marketing title20 people, the buyer among themToo many
- Waterfall ICPMaya Chen, CMORight person
- P1stop
- P2
- P3
- P4
- P5
- P6
Northwind Logistics320 employees
- CMO onlyNo oneMissed
- Every marketing titleInes Duarte, Head of MarketingRight person
- Waterfall ICPInes Duarte, Head of MarketingRight person
- P1
- P2stop
- P3
- P4
- P5
- P6
Acme140 employees
- CMO onlyNo oneMissed
- Every marketing titleDaniel Cho, Marketing DirectorRight person
- Waterfall ICPDaniel Cho, Marketing DirectorRight person
- P1
- P2stop
- P3
- P4
- P5
- P6
Brightline40 employees
- CMO onlyNo oneMissed
- Every marketing titleJordan Hayes, Marketing ManagerRight person
- Waterfall ICPJordan Hayes, Marketing ManagerRight person
- P1
- P2
- P3
- P4stop
- P5
- P6
Harbor Labs25 employees
- CMO onlyNo oneMissed
- Every marketing titleLeo Brandt, Marketing CoordinatorRight person
- Waterfall ICPLeo Brandt, Marketing CoordinatorRight person
- P1
- P2
- P3
- P4
- P5stop
- P6
Initech Cloud15 employees
- CMO onlyNo oneMissed
- Every marketing titleNo oneMissed
- Waterfall ICPAva Moreno, FounderRight person
- P1
- P2
- P3
- P4
- P5
- P6stop
- CMO only
- Finds the buyer in 1 of 6
- Every marketing title
- Finds the buyer in 5 of 6, plus 19 people to sort by hand
- Waterfall ICP
- Finds the buyer in 6 of 6
Fictional companies from the explainer. The matching follows the documented rules of the endpoint.
How do you tell the decision maker from a champion or an influencer?
The last levels of your list can stay broad on purpose. They catch the companies where titles don’t help, like the founder-run startup. If a broad level sometimes brings back the wrong person, that can be fixed afterwards, with a qualification step that reads each profile and gives each person a role in the purchase:
- Buyer: the decision maker, who owns the budget and signs.
- Champion: wants the change and pushes for it inside.
- Influencer: shapes the choice without owning it.
- Noise: matched a title, plays no part in the decision.
In the video, that step is Jev by TypeSafe AI, which reads each person’s full profile (role, responsibilities, career). With the level that found them, the list can route itself: buyers to your sequencer, champions to your CRM.
- Maya ChenCMO · Globex AnalyticsP1BuyerSequencer
- Jordan HayesMarketing Manager · BrightlineP4BuyerSequencer
- Ava MorenoFounder · Initech CloudP6BuyerSequencer
- Ines DuarteHead of Marketing · Northwind LogisticsP2ChampionCRM
- Daniel ChoMarketing Director · AcmeP2InfluencerWarehouse
AI can recover precision. It can’t recover someone you never discovered.
What does a complete targeting system look like?
Finding the person is one block of five. Before it, a company search decides which accounts are worth searching; after it, qualification, routing and email enrichment turn the people you keep into conversations.
- Company TAMThe companies you sell to.
- Waterfall ICPThe best available person in each one.
- QualificationA second opinion on each person.
- RoutingWho reaches out, and how.
- ActivationEmails and phones for the people you keep, then your sequencer.
Don’t build a bigger search.Build a smarter targeting system.
Show the answer key
- Nothing: the CMO is missing, so the search fails.That is what a single CMO-only search does. The cascade does not stop at P1 when P1 is empty: it moves down the list.
- Its Head of Marketing, found at P2.P1 finds nobody, so the search moves to P2, finds the Head of Marketing and stops there.
- Everyone in its marketing team.That is the broad search. The cascade stops at the first level that finds someone, so you get one person, not the whole team.
For builders
Everything above runs on one endpoint. This part is for the people who will wire it up: the call, a real run, how titles match, and what to check before you ship.
How do you run Waterfall ICP with the API?
One company in, its best available people out, in your order of preference, each tagged with the level that found them.
Send
company_linkedin_urlrequired- The company's LinkedIn page, the input that identifies it reliably.
cascade[]required- Your ranked levels, 1 to 10. Each has include_title, and optionally exclude_title, location and include_headline_search.
max_results- How many people to return across all levels. 10 by default, 100 at most.
Read
icp- The level of your list that found the person.
ranking- The person's overall relevance rank inside the company.
person- The full profile. Pass person.linkedin_url to enrichment for an email or a phone.
How the slots fill: a real run
We ran a four-level cascade on October 14, 2026, against a real company page with more than a thousand employees, asking for 3 people. P1 found two of them and took two slots. P2 found nobody, so the search moved on. P3 filled the last slot, and the search stopped.
- Slot 1P1VP Growth & Marketingranking 1
- Slot 2P1Chief Brand & Marketing Officerranking 2
- Slot 3P3Growth Marketing Managerranking 3
- P1Chief Marketing Officer · [CMO] · VP Marketing · Vice President Marketing2 found
- P2Head of Marketing · Marketing Director · Head of GrowthNobody, moves on
- P3Marketing Manager · Growth Manager · Demand Generation Manager1 found, stop
- P4[CEO] · Founder · Co-founderNot reached
The request, as it ran:
{
"company_linkedin_url": "https://www.linkedin.com/company/<company>",
"cascade": [
{
"include_title": ["Chief Marketing Officer", "[CMO]",
"VP Marketing", "Vice President Marketing"],
"exclude_title": ["assistant", "intern", "junior"]
},
{
"include_title": ["Head of Marketing", "Marketing Director",
"Head of Growth"],
"exclude_title": ["assistant", "intern", "junior", "product"]
},
{
"include_title": ["Marketing Manager", "Growth Manager",
"Demand Generation Manager"],
"exclude_title": ["assistant", "intern", "junior", "product"]
},
{
"include_title": ["[CEO]", "Founder", "Co-founder"],
"exclude_title": ["assistant", "intern"]
}
],
"max_results": 3
}And what came back, trimmed to the fields that matter (names and company withheld): a VP Growth & Marketing and a Chief Brand & Marketing Officer, both from P1 with ranking 1 and 2, then a Growth Marketing Manager from P3 with ranking 3. Note how “VP Marketing” caught “VP Growth & Marketing”: the words matched, in another order.
To reach the people it returns, pass each person.linkedin_url to Email Enrichment or Phone Enrichment. For coverage and the other APIs it works with, see decision maker identification with BlitzAPI; plans and rate limits are on the BlitzAPI pricing page.
Two numbers on every result
- icp1
The level that found them
Who reaches out
- ranking1
How central they are
How hard you go
These thresholds are the first routing rule the documentation suggests, a starting point to tune on your own pipeline. Store both numbers: they answer different questions, and a person found at level 3 can still have ranking 1.
How a title matches a level
- All the words, any order. “VP Marketing” matches “VP of Marketing” and “Marketing VP”.
- No synonyms, no abbreviations. “Demand Gen” does not match “Demand Generation”: list each form you want.
- Brackets mean exact. “[CEO]” matches only the title CEO. Without brackets, “CEO” also matches “CEO Office”.
- Location is the person’s country, not the company’s headquarters. WORLD means no filter.
- Current roles only. Someone who left the company will not come back.
Before you ship
- Exclude aggressively: assistant, intern and junior on every level, and product in a marketing cascade.
- Run a second list for companies under about 20 people, with the founder at P1, so that P1 means “ideal buyer” everywhere.
- Set max_results on purpose: 1 for the single best buyer, 3 to multi-thread.
- Headline search in the first level: it trades away precision exactly where you need it most.
- Watch where matches come from. If most companies resolve at the last level, your first levels do not use the titles this market uses.
Sources and method
- Written by
- The BlitzAPI team
- Published
- Updated
- API calls run on
- Waterfall logic, BlitzAPI documentation
- Waterfall ICP Search, API reference
- Keyword filters: how title phrases match
The diagrams redraw the Waterfall ICP explainer with the BlitzAPI design system. The companies and people in them, and in the simulator, are fictional; the matching follows the documented rules of the endpoint.
The API call in the builders section ran against a real company page. The company and the people are withheld; their job titles are kept as returned.
Video transcript
Read the full transcript2:36 · 5 chapters
0:00One perfect query
You want to find the right marketing decision-maker across your TAM. So you open the People Search, right? Company filters on one side. People filters on the other. And then you try to compress your entire targeting strategy into one perfect query. Let’s start clean. CMO only. Great list. Very senior. Very relevant.
0:25Same need, different org chart
Except… what happens to every company that doesn’t actually have a CMO? Maybe they call the role Head of Marketing. Marketing Director. Or it’s a 15-person startup where the founder still owns the whole function. Same need. Different org chart. And your search just made them invisible. So naturally, you broaden it. VP Marketing. Head of Marketing. Directors. Growth. Marketing Managers… Coverage goes up. But now put that query inside a 5,000-person company. Suddenly, half the marketing org qualifies. That’s the People Search trade-off: go narrow, lose the market. Go broad, lose the signal. And the problem isn’t really the filters. It’s asking one query to understand thousands of different org charts.
1:11A search that knows when to stop
That’s where Waterfall ICP changes the logic. Instead of putting every possible buyer into the same search, you define an order. First: CMO or VP Marketing. Nobody? Move down. Heads and Directors. Still nobody? Growth leaders. Managers. Broader roles. And eventually, maybe the founder. Think of it as a search that knows when to stop. A Marketing Manager can be noise in an enterprise… and the perfect buyer in a 40-person company. Same playbook. Different company. Different answer.
1:43Waterfall finds, Jev qualifies
Now here’s where it gets really interesting. The last Waterfall priorities can stay broad. Why? Because you can use Jev. Give the full profile context behind every candidate. Role. Responsibilities. Experience. Career history. So instead of trying to make the title logic perfect, you let Waterfall maximize discovery… and Jev recover the precision. Waterfall finds the best candidates. Jev understands who actually matters. Buyer. Champion. Influencer. Noise. Combine that with Waterfall priority, and the list can route itself.
2:19Discovery first
AI can recover precision. It can’t recover someone you never discovered. Company TAM. Waterfall ICP. Jev. Routing. Activation. Don’t build a bigger search. Build a smarter targeting system.