ICP Filter: 3 Levels + a Simple Decision Tree
- Marina Lukyanova
- Dec 17, 2025
- 7 min read
Most teams treat “ICP” as a slide. Then they wonder why sales chases odd deals, product keeps absorbing one-off requests, and marketing can’t hold a consistent message.
An ICP that works is not a description. It’s a filter. It helps your team decide, quickly and consistently, who gets your time.
The tool below is the version I use when a company needs alignment across GTM and delivery. It has two parts:
1) a three-level ICP definition (company, buyer, behaviour)
2) a short decision tree your team can actually use in real conversations
The tool: the 3-level ICP definition
Level 1: Company fit (the container)
This is the easiest place to start, and the easiest place to get sloppy.
Define:
- Industry: where you win, and where you do not.
- Stage: pre-seed, seed, growth, mature. Or “first repeatable motion” vs “scaling an existing motion”.
- Size: revenue, headcount, or both. Pick one as your primary anchor.
- Geography: only include if it materially changes the buying process or delivery.
- Business model: B2B SaaS, services-led, marketplace, etc.
- Trigger events: the change that makes your solution urgent.
Write it as a narrow sentence:
“We help [type of company] that is [in this stage] and is [in this change moment].”
Add exclusions on purpose:
“We do not sell to [segment] because [repeatable reason].”
This is how you protect your roadmap and your team’s attention.
Level 2: Buyer fit (the human decision context)
Same company, different buyer, different deal.
Define:
- Economic buyer role: who can say yes.
- Daily owner: who will live with the outcome.
- The problem they feel: what keeps coming back to the CEO, what is creating rework, what is blocking a milestone.
- The KPI or constraint: what they are measured on (or what is breaking).
- The buying motion: self-serve, founder-led sales, committee, board loop.
Practical prompts that produce usable language:
- “What did this person try before they called it?”
- “What gets escalated to them every week?”
- “What would make them say: we cannot keep running the business like this?”
Level 3: Behaviour fit (how they act, not what they claim)
This level is where the real filtering happens. It’s also the level most teams skip.
Define:
- Urgency: are they in a change moment, or just curious.
- Willingness to commit: can they make choices and keep them made.
- Capability: do they have the minimum team and bandwidth to implement.
- Operating rhythm: do they run weeklies, ship regularly, review metrics, or is everything ad hoc.
- Change appetite: are they willing to adjust process, not just buy a tool.
Behaviour fit is not about judging. It’s about predicting friction.
A simple way to write this level is as “green flags” and “red flags”.
Green flags:
- They can name a specific bottleneck and its cost in time, churn, pipeline, or missed delivery.
- They can point to a recent trigger event: pivot, new offer, acquisition integration, fundraising milestone.
- They have an owner who will run the work weekly, not “when things calm down”.
- They are willing to say no to at least one competing priority.
Red flags:
- They want a fix, but they cannot name a decision they will stop revisiting.
- They describe the work as “a quick alignment session” while expecting deep change.
- They need the CEO to approve every detail, but also want the CEO out of the loop.
- They are collecting advice for a future phase.
The ICP filter tree (a decision tree you can use live)
Print this, put it next to your CRM, and train the team on it. The goal is not perfect scoring. The goal is consistent yes, no, and not yet.
Gate 0: Are we solving the right category of problem?
If your product or service solves a specific bottleneck, start here.
- If they cannot describe the bottleneck in one sentence, pause and diagnose.
Outcome: not yet (send to discovery), or no (if it’s clearly out of scope).
Gate 1: Company fit
1) Do they match our must-have company traits (industry, stage, size)?
- Yes: go to Gate 2
- No: go to 1a
1a) Is the mismatch one we can win without bending the offer?
If you need custom features, unusual delivery, or a new positioning story to close, the answer is usually no.
Outcome: no, or not yet (if they are one quarter away from the right stage).
Gate 2: Buyer fit
2) Are we talking to the role that can fund this, or to the role that owns the pain day to day?
- Yes: go to Gate 3
- No: go to 2a
2a) Can we reach the economic buyer within one step?
Outcome: not yet (qualify a path), or no (if the path is political and slow).
Gate 3: Pain and stakes
3) Can they name a repeatable pain pattern we recognise?
Look for: rework, handoff breakdowns, decisions circling back, priorities colliding, stalled pipeline, inconsistent onboarding.
- Yes: go to Gate 4
- No: outcome: not yet (problem framing), or no (if it’s a one-off ask).
4) Is the pain tied to a near-term event?
Examples: board checkpoint, runway pressure, launch, market entry, integration, hiring freeze.
- Yes: go to Gate 5
- No: outcome: not yet (nurture).
Gate 4: Behaviour fit
5) Will they commit to a working rhythm?
For alignment work, that means a weekly cadence and a named owner.
- Yes: go to Gate 6
- No: outcome: no (or not yet, if timing is the only blocker and you can set a date).
6) Can they keep decisions made?
Ask: “What decision do you want to stop reopening?”
- Yes: go to Gate 7
- No: outcome: not yet (leadership alignment needed), or no.
Gate 5: Commercial reality
7) Is there a credible budget and authority path?
You do not need the exact number, you need a credible “how this gets funded”.
- Yes: outcome: yes (qualified)
- No: outcome: not yet, or no
Optional quick score (if your team wants one number)
If you need a light score, make it binary and boring:
- Company fit: yes/no
- Buyer fit: yes/no
- Behaviour fit: yes/no
Only move forward when you have at least 2 of 3 as yes, and the missing one is solvable within the next two interactions.
Make the filter real in your operating system
A filter fails when it lives in a doc nobody opens. Put it where decisions happen.
Three simple implementation moves:
- Add three fields to your CRM: Company Fit, Buyer Fit, Behaviour Fit. Make them required for stage changes.
- Add one line to your intake form: “What trigger event makes this urgent now?”
- Review one “not yet” lead every week in pipeline review, so your team learns the pattern and refines the gates.
If you run a services business, add one more constraint:
- Define what “standard delivery” means in two sentences.
This prevents the slow drift into custom work that looks like revenue and behaves like chaos.
How to say “no” and “not yet” without burning the relationship
The filter is only useful if your team can use it without feeling awkward.
No:
“Based on what you’ve shared, we’re not the right fit for this stage. We’re strongest when [your ICP trigger]. If that changes, I’m happy to reconnect.”
Not yet:
“This looks promising, but we’re missing two things: [economic buyer involvement] and [a clear trigger event]. If you can bring those into the next conversation, we can tell you fast whether we can help.”
Yes:
“You’re describing a pattern we see often: [pain pattern]. If we confirm [one gate question], we can map the fastest path to a decision and a weekly cadence.”
Example scenario: turning an ICP slide into an operating filter
A founder-led B2B startup is preparing for a seed extension. The team is under runway pressure and wants revenue fast. Inbound leads are coming from everywhere: agencies, solo consultants, larger enterprises, and a few mid-sized teams.
Without a filter, the team says yes to whoever answers emails. Two weeks later, product is building edge cases, sales is rewriting proposals, and the founder is back in the middle of every decision.
They write the three-level ICP:
Company: tech-oriented services firms, 30 to 200 people, in a transition moment (new offer or new market), with a leadership team that meets weekly.
Buyer: CEO or MD as economic buyer, with an ops or functional lead as the day-to-day owner.
Behaviour: urgency tied to a near-term shift, willingness to run a weekly cadence, and openness to saying no to legacy work that no longer fits.
Then they apply the decision tree to 12 recent leads.
Result:
- 5 are a clear no because they require custom delivery or sit outside the stage.
- 4 are not yet because the economic buyer is not involved or there is no urgency event.
- 3 are yes, and the conversations move faster because everyone is talking about the same pain pattern and the same constraints.
Nothing magical happened. They simply stopped paying attention to the wrong work.
Common mistakes to avoid
- Writing an ICP that reads like a persona poster, not a decision tool.
- Defining company traits, but skipping behaviour traits.
- Letting exceptions accumulate without updating the filter.
- Treating budget as the only qualifier. Misfit deals are expensive even when they pay.
- Building messaging for five ICPs at once.
Try this this week
1) Pick 10 recent deals or leads: 3 that closed cleanly, 3 that churned or stalled, 4 that felt messy.
2) Draft the 3-level ICP on one page. One sentence per level plus 3 green flags and 3 red flags.
3) Build the decision tree with 7 gates max. If it’s longer, nobody will use it.
4) Run the 10 deals through it as a team. Where did you say yes when you should have said not yet.
5) Update one thing only: either a must-have company trait, or a behaviour red flag, or a single gate question.
6) Put the filter in the places decisions happen: CRM fields, intake form, weekly pipeline review.
_________________________________________________________________________________________________
Marina Lukyanova, 𝐅𝐫𝐚𝐜𝐭𝐢𝐨𝐧𝐚𝐥 𝐂𝐎𝐎 specialising in 𝐒𝐭𝐫𝐚𝐭𝐞𝐠𝐲 𝐀𝐥𝐢𝐠𝐧𝐦𝐞𝐧𝐭. I help scaling and pivoting companies turn strategy into operational traction - aligning teams, systems, and focus - to unlock faster growth without wasting resources. I’m known for uncovering hidden misalignments and guiding leadership teams through the complexity with clarity and momentum.
👉 Want help with alignment? Learn more at alignwise.co or reach out to me on LinkedIn to turn strategic intent into operational traction that moves the needle.
Tags:
Comments