ICP research is not a database-filter exercise. It is the discipline of deciding where sales attention deserves to go, what the team actually knows, what it is only assuming, and what evidence would make the next action smarter. A strong ICP explains who could become a good customer.

Strong ICP research goes further:

  • Why this market?
  • Why this account?
  • Why this buyer?
  • Why this problem?
  • Why now?
  • What would prove the hypothesis wrong?
  • How much research is economically justified before outreach begins?

The objective is not to know everything about a company. The objective is to know enough to make a better commercial decision.

1. What ICP Research Is Actually For

ICP research should improve decisions across the revenue system.

It should help answer:

  • Which markets deserve investment?
  • Which accounts should enter active outreach?
  • Which accounts should be monitored instead?
  • Which companies look good on paper but create weak economics?
  • Which buyer roles are connected to the likely problem?
  • Which stakeholders could influence the decision?
  • Which public signals improve timing confidence?
  • Which evidence is verified?
  • Which evidence is inferred?
  • Which assumptions should be tested in conversation?
  • Which accounts should be removed before buyer attention is spent?
  • What language is natural for the buyer and industry?
  • What level of research is justified by the potential opportunity?

Research is valuable when it changes what the sales team does next.

2. Market Hypothesis vs Account Evidence

A market hypothesis is a belief about where a problem should exist. Account evidence is what supports or weakens that belief for a specific company.

Example market hypothesis:

B2B software companies adding AEs may experience pressure on outbound coverage while the team ramps.

Account evidence might include:

  • Eight open AE roles
  • A newly hired CRO
  • Expansion into two new territories
  • No visible SDR hiring
  • A public leadership comment about pipeline creation

The market hypothesis creates the search direction. Account evidence determines whether the company deserves priority. Do not confuse the two. A good market hypothesis can still produce a bad account.

3. Market → ICP → Priority Account

Use three different levels of thinking.

Market

The broader environment where the company can compete.

ICP

The type of company most likely to become a good customer.

Priority Account

An ICP-aligned company with enough evidence, timing, access, and commercial value to deserve attention now. This distinction prevents the entire addressable market from becoming one giant calling list.

4. Use GATE → SCORE → SIGNAL → ACCESS → ACTION

A practical research sequence is:

GATE

Confirm non-negotiable eligibility.

Examples:

  • Supported geography
  • Relevant company type
  • Commercial minimums where applicable
  • Required technical compatibility
  • No legal or competitive conflict
  • Realistic implementation path

SCORE

Evaluate the strength of the overall fit.

SIGNAL

Look for evidence that timing or problem intensity may have changed.

ACCESS

Determine whether the team can identify and reach the relevant buying roles.

ACTION

Decide the level of research, personalization, channel effort, and follow-up the account deserves. A strong trigger should not rescue a hard-gate failure.

5. Firmographic Fit vs Operational Fit

Firmographic Fit asks what the company is.

Typical fields include:

  • Industry
  • Employee count
  • Revenue range
  • Geography
  • Ownership type
  • Funding stage
  • Company age
  • Number of locations
  • Business model
  • Customer type

Firmographics are useful for market boundaries and scalable filtering. They do not prove a problem exists. Operational Fit asks how the company works.

Look for:

  • Team structure
  • Number of sites or providers
  • Workflow complexity
  • Sales motion
  • Delivery model
  • Technology environment
  • Regulatory burden
  • Internal capability
  • Centralized vs distributed operations
  • Customer volume
  • Vendor structure
  • Manual processes
  • Decision speed
  • Implementation environment

Operational fit usually gets closer to the actual reason the solution may matter. Use firmographics to define the search space. Use operating reality to determine relevance.

6. Research the Problem Environment

Do not start with a job title. Start with the business situation that could create the problem.

Research:

  • What conditions normally produce the problem?
  • What symptoms may become visible externally?
  • What root causes are plausible?
  • Who would feel the effect first?
  • What business consequence could follow?
  • What alternatives might already solve it?
  • What would make the problem disappear without the seller?

Then map the likely buyer roles.

The sequence is:

BUSINESS SITUATION → PLAUSIBLE PROBLEM → POSSIBLE IMPACT → BUYER ROLES

7. The Three Truths Rule

Separate what is known from what is believed.

FACT → HYPOTHESIS → BUYER-CONFIRMED TRUTH

Fact

The company is hiring 15 account executives.

Hypothesis

Rapid AE hiring may create prospecting coverage or ramp pressure.

Buyer-confirmed truth

“Yes, the AEs are still carrying too much outbound while the SDR team ramps.” Facts create credible reasons to engage. Hypotheses create questions. Buyers provide the internal truth. Never present a hypothesis as a diagnosis.

8. Evidence Status Should Be Visible

Important research fields should carry an evidence status. Use the canonical five Evidence Confidence levels below. Contradicted is a separate evidence status:

Verified

Directly supported by a credible source.

Strong Evidence

Multiple credible indicators point in the same direction.

Partial Evidence

Some support exists, but the conclusion remains incomplete.

Assumed

The researcher believes it may be true but lacks meaningful proof.

Unknown

No reliable conclusion is available.

Contradicted

Evidence exists against the current hypothesis. Unknown is not true. Assumed is not verified. Contradicted evidence should not disappear just because the account looks attractive.

9. Source Triangulation

One source can be wrong, stale, incomplete, or interpreted incorrectly. For important decisions, triangulate when practical.

Example:

  • A database says the company has 250 employees.
  • LinkedIn appears closer to 130.
  • The company website shows three business units.
  • A recent job post suggests rapid expansion.

Do not silently choose the number that best supports the target hypothesis. Record the uncertainty and use the most credible combination of sources. The more important the decision, the more useful triangulation becomes.

10. Source Quality Is Not Equal

Different sources answer different questions.

First-party company sources

Useful for official services, locations, leadership, announcements, positioning, and stated priorities.

LinkedIn

Useful for roles, tenure, team structure, hiring, public activity, and buying-group mapping.

Job postings

Useful for capability needs, reporting lines, technology clues, growth, and internal priorities.

Public filings

Useful for strategic priorities, financial context, risk factors, operating trends, and large-account economics.

Press releases and reputable news

Useful for expansion, acquisitions, leadership changes, partnerships, funding, and launches.

Technology databases

Useful for possible installed systems and vendor environment.

Review platforms

Useful for recurring customer or employee themes, not unquestioned internal truth.

Commercial databases

Useful for scale and list building, but often contain inferred or stale fields.

CRM, calls, and customer records

Useful for real buyer language, objections, qualification patterns, implementation outcomes, retention, and win-loss evidence. The source should match the question.

11. Evidence Freshness

Correct information can become wrong over time. Call this Evidence Freshness.

Fields that decay quickly can include:

  • Job title
  • Employment status
  • Headcount
  • Open roles
  • Technology stack
  • Funding status
  • Locations
  • Leadership
  • Vendor relationships
  • Contract timing
  • Strategic priorities
  • Buying committee
  • Contact information

Slower-changing fields may include:

  • Core industry
  • Business model
  • Regulatory environment
  • Typical customer type
  • Long-term service structure

Freshness should influence confidence. A verified fact from 18 months ago may deserve less weight than a current partial signal.

12. Evidence Decay

Evidence Decay is the loss of reliability as information ages or the business changes.

A research record should be reconsidered when:

  • A new executive joins
  • A company is acquired
  • A funding event occurs
  • A new market launches
  • A large hiring or layoff cycle happens
  • A new location opens
  • A major vendor changes
  • A regulatory change occurs
  • A prospect says the information is outdated
  • A previously targeted buyer leaves

Research is not permanent truth.

13. Contradictory Evidence Is Valuable

Research teams often search only for reasons to pursue an account. That creates confirmation bias. Look deliberately for evidence that weakens the case.

Examples:

  • The company recently built the internal capability being offered
  • The buyer publicly says the current model is performing well
  • The account renewed a competitor contract
  • The business is reducing the relevant function
  • The company is moving away from the target market
  • The implementation environment is incompatible
  • The economics appear too small
  • The account has repeatedly failed to convert despite good engagement

Strong research is willing to disconfirm itself.

14. The Disconfirming Evidence Test

Before assigning a high priority, ask:

What evidence would make this account less attractive? Then search for it.

Examples:

If the hypothesis is “AEs need more outbound support,” check whether a mature SDR team already exists. If the hypothesis is “multi-location growth increases compliance complexity,” check whether compliance is centrally managed by a specialized internal team. If the hypothesis is “the company may need IT support,” check whether a long-term internal IT organization is already visible. A model that can only confirm itself is not research. It is selection bias.

15. Positive Fit vs Negative Evidence

Positive Fit can include:

  • Right company structure
  • Right operating model
  • Plausible problem
  • Relevant buyer environment
  • Commercial potential
  • Serviceability
  • Strategic value

Negative Evidence can include:

  • Unsupported geography
  • No credible use case
  • No identifiable owner
  • Poor implementation fit
  • Unprofitable service requirements
  • Competitive conflict
  • Technical incompatibility
  • Regulatory or legal risk
  • Weak economics
  • Repeated no-decision pattern
  • Existing capability that removes the need

The fastest way to improve a lead list is often to remove the wrong accounts earlier.

16. Hard Disqualifier vs Soft Friction

Not every weakness means the account should be rejected.

Hard Disqualifier

A condition that makes pursuit irrational or impossible.

Examples:

  • Cannot legally serve the account
  • Unsupported geography
  • No usable solution fit
  • Impossible technical requirement
  • Direct client conflict
  • No economic path to a viable engagement

Soft Friction

A condition that changes effort, timing, or strategy.

Examples:

  • Long procurement
  • Incumbent vendor
  • Weak timing
  • Complex buying committee
  • Higher onboarding effort
  • Limited buyer access
  • Undefined budget

Hard disqualifiers stop pursuit. Soft friction changes the plan.

17. Fit vs Timing

Fit asks:

Should this company ever be a target?

Timing asks:

Is there a credible reason the issue may matter now?

Use the four-box model:

High Fit + High Timing

Priority now.

High Fit + Low Timing

Monitor, nurture, or use lighter outreach.

Low Fit + High Timing

Investigate carefully. Do not let a news event rescue poor economics.

Low Fit + Low Timing

Deprioritize. Fit and timing are related, but they are not the same decision.

18. Not All Signals Are Equal

Useful signal categories include:

Structural Signal

A condition that makes the problem more plausible. Examples: multi-location model, regulated industry, large field team.

Change Signal

Something materially changed. Examples: leadership change, funding, acquisition, new geography, expansion.

Problem-Adjacent Signal

A signal more closely connected to the likely problem. Examples: repeated hiring for an internal capability, public complaints, process overhaul.

Direct Intent or Buyer Evidence

The company or buyer explicitly reveals interest, evaluation, dissatisfaction, or a relevant project. The closer the signal is to the problem and decision, the more it should influence priority.

19. Signal Quality

Use four dimensions:

Relevance

How closely does the signal connect to the problem?

Proximity

How close is it to the actual buying situation?

Specificity

How clearly can the signal be interpreted?

Freshness

How recent is it?

A useful mental model is:

SIGNAL QUALITY = RELEVANCE + PROXIMITY + SPECIFICITY + FRESHNESS

This is a decision aid, not a universal mathematical formula.

20. Signal-to-Problem Distance

A signal can be close to the likely problem or far away from it.

Short Distance

The connection is commercially plausible.

Example:

A company is hiring a new SDR team while expanding sales territories. A conversation about prospecting coverage or ramp may be reasonable.

Long Distance

The signal could mean many unrelated things.

Example:

The CEO posted about leadership culture. That does not prove a sales, staffing, IT, compliance, or operational problem. The longer the distance, the softer the hypothesis should become.

21. The Four Clocks of Timing

Timing should be researched across four possible clocks.

Problem Clock

When does the current problem become painful enough to matter?

Budget Clock

When can money realistically be allocated, moved, or created?

Contract Clock

When does an incumbent agreement renew, expire, or become reviewable?

Political Clock

When do leadership changes, internal initiatives, reorganizations, or stakeholder dynamics make change more or less likely? A good account can be badly timed because one of these clocks is not ready.

22. Account Priority Is More Than Fit

A practical mental model is:

ACCOUNT PRIORITY = FIT × EVIDENCE × TIMING × ACCESS × VALUE

Treat this as a decision model, not literal mathematics.

Fit

Does the company belong in the target market?

Evidence

How much credible support exists for the problem hypothesis?

Timing

Why might the issue matter now?

Access

Can the relevant buying roles be identified and reached?

Value

Is the opportunity commercially and strategically worth the effort? A high-fit account with no access and no timing may deserve less immediate effort than a slightly smaller account with stronger evidence and a reachable problem owner.

23. Research Depth Should Match the Cost of Being Wrong

Do not use the same research depth for every account.

SMB

Confirm basic fit, right buyer, plausible problem, contact validity, commercial viability, and obvious disqualifiers.

Mid-Market

Add operating structure, multiple buyer roles, public priorities, technology environment, stronger signals, relevant proof, and buying-group clues.

Enterprise or Strategic

Add business units, regional structure, executive priorities, buying committee, economic buyer, technical stakeholders, procurement, security, incumbent vendors, strategic initiatives, financial context where relevant, and implementation dependencies.

Rule:

RESEARCH DEPTH SHOULD RISE WITH DEAL VALUE, COMPLEXITY, AND THE COST OF BEING WRONG.

24. Research ROI

Research has a cost. It consumes seller or researcher time before any buyer conversation occurs.

Research ROI improves when the work changes:

  • Account inclusion
  • Priority tier
  • Buyer selection
  • Message angle
  • Channel choice
  • Qualification strategy
  • Proof selection
  • Timing
  • Disqualification
  • Buying-group map

A simple question is:

Did this research change a decision? If not, the field may be informational rather than operational.

25. Diminishing Research Return

The first few minutes of research often create the largest improvement in judgment. After a point, additional research may add detail without changing the decision. Call this Diminishing Research Return.

Example:

Two minutes confirms the account is a relevant multi-location dental group and identifies the operations owner. Another five minutes finds a recent acquisition and a likely compliance trigger. Another 20 minutes produces executive biographies, old interviews, office photos, and unrelated social activity. The first seven minutes changed the sales decision. The next 20 may not.

26. The Research Stopping Rule

Stop researching when additional information is unlikely to change:

  • Whether the account should be targeted
  • Who should be contacted
  • Why the conversation is relevant
  • Which channel should be used
  • What claim can safely be made
  • What needs to be verified with the buyer

If the research cannot improve the next decision, move to the conversation. The buyer is often the fastest source of the missing truth.

27. Research Debt

Research Debt is uncertainty that is ignored instead of recorded.

Examples:

  • An inferred title is treated as verified
  • An old employee count is copied forward
  • A technology signal is treated as current
  • A guessed problem becomes personalization
  • A weak account is kept because the list needs volume
  • A contact is labeled decision-maker without role evidence

The debt gets paid later through:

  • Wrong-person outreach
  • Lower reply quality
  • Poor call openings
  • Bad personalization
  • Weak qualification
  • Wasted AE time
  • CRM inaccuracies
  • Forecast noise

Unknown fields do not need to be eliminated. They need to remain visible.

28. Research Blindness

More information can make researchers feel more certain without making the account more correct.

Common forms include:

Confirmation bias

Searching only for facts supporting the hypothesis.

Prestige bias

Overrating a large brand because the logo is attractive.

Recency bias

Overweighting the newest signal even when it is commercially weak.

Database authority bias

Assuming a paid data source must be correct.

Title bias

Assuming senior title equals problem ownership.

Activity bias

Believing a company with many public signals must be a better prospect. Research quality comes from judgment, not information volume.

29. Customer-Lookalike Research vs White-Space Research

Customer-Lookalike Research studies successful customers and asks what conditions made them successful.

Research:

  • Use case
  • Operating structure
  • Problem severity
  • Buyer roles
  • Deal size
  • Sales cycle
  • Implementation success
  • Retention
  • Expansion
  • Margin
  • Support burden

Do not clone visible attributes only. Clone the conditions that created a healthy customer. White-Space Research explores segments not yet well represented in the customer base.

Useful for:

  • New offers
  • New geographies
  • Moving upmarket
  • Moving downmarket
  • Adjacent industries
  • New buyer roles
  • New business models

Protect proven segments while testing white-space hypotheses separately.

30. First-Party Evidence vs Third-Party Data

First-party evidence can include:

  • Closed-won customers
  • Closed-lost opportunities
  • Discovery calls
  • Call recordings
  • CRM notes
  • Implementation outcomes
  • Support history
  • Renewals
  • Expansion
  • Churn reasons
  • Win-loss analysis
  • Customer interviews

Third-party data can include:

  • Company databases
  • LinkedIn
  • Websites
  • Job boards
  • Technology data
  • News
  • Directories
  • Public filings
  • Regulatory sources
  • Review platforms

First-party evidence shows what actually happened. Third-party data helps locate more companies where similar conditions may exist. Use both.

31. Buyer-Role Mapping

A job title is a clue, not proof of buying responsibility.

Map likely roles such as:

  • Day-to-day user
  • Problem owner
  • Manager
  • Champion candidate
  • Economic buyer
  • Executive sponsor
  • Technical evaluator
  • Security or compliance
  • Procurement
  • Finance
  • Legal
  • Blocker or skeptic

For each role, research:

  • What they own
  • What they may be measured on
  • What consequence they are likely to care about
  • What evidence they may require
  • What risk they may perceive
  • What language they use
  • What other stakeholders they interact with

32. Buying-Group Coverage

For complex deals, research should move beyond one contact. Buying-Group Coverage means understanding enough relevant roles to shape the account strategy.

Weak coverage:

Five random employees have been identified.

Strong coverage:

The team understands the problem owner, likely economic buyer, technical evaluator, procurement path, and potential executive sponsor. Contact count is not stakeholder understanding.

33. Buyer Psychology by Title

Founder / CEO

Research growth stage, founder involvement, market expansion, hiring, management bandwidth, cash efficiency, and revenue concentration. Use language around growth, leverage, focus, speed, coverage, risk, and revenue.

CRO / VP Sales

Research sales headcount, AE and SDR hiring, territory structure, leadership tenure, sales motion, pipeline pressure, and market expansion. Use language around pipeline coverage, AE productivity, conversion, forecast, ramp, and opportunity creation.

Head of SDR / BDR

Research team size, manager structure, open roles, data environment, sequence model, regional coverage, turnover, and enablement. Use language around connect quality, positive replies, rep productivity, data, coaching, and meeting quality.

RevOps

Research CRM environment, routing complexity, sales stack, data sources, team growth, stage definitions, and reporting model. Use language around data quality, attribution, CRM integrity, funnel leakage, routing, and process consistency.

COO

Research locations, operating model, workflow complexity, service structure, ownership, growth, and process dependencies. Use language around capacity, reliability, workflow, management burden, handoffs, and operating cost.

CFO

Research cost structure, growth investment, headcount changes, commercial model, expansion, and publicly available financial context where relevant. Use language around CAC, payback, fixed cost, margin, cost per opportunity, and revenue efficiency.

34. Industry Changes the Research Model

SaaS / Technology

Research sales motion, funding stage, team size, AE/SDR hiring, integrations, product expansion, target market, ACV clues, and security requirements. Useful signals include new sales leadership, funding, hiring, market expansion, product launch, and enterprise movement.

MSP / IT Services / Telecom / UCaaS

Research service area, recurring-revenue model, technology partnerships, supported platforms, customer size, geography, current provider environment, and contract timing. Useful signals include new locations, acquisitions, cloud growth, technology hiring, vendor changes, and renewals.

Consulting / Professional Services

Research partner structure, service lines, industries served, project vs retainer model, senior delivery involvement, utilization clues, and case studies. Useful signals include new practice areas, new partners, expansion, market entry, and hiring.

Staffing / Recruiting

Research vertical specialization, recruiter headcount, requisitions, client type, geography, contingent vs retained model, and hiring-manager access. Useful signals include hiring spikes, new recruiter teams, client expansion, and geographic growth.

Manufacturing / Logistics / Industrial

Research facilities, plants, distribution footprint, procurement, production model, supply chain, channel partners, and site vs corporate authority. Useful signals include plant expansion, acquisition, capacity investment, new geography, and supply-chain change.

Healthcare / Dental / Medical Services

Research practice type, specialty, providers, locations, ownership model, staff size, administration, compliance exposure, and operational complexity. Useful signals include new providers, new locations, acquisitions, ownership transitions, multi-location growth, and administrative hiring.

Financial Services

Research business line, client segment, regulatory environment, office footprint, advisor or sales structure, controls, and technology environment. Useful signals include expansion, acquisition, product launch, compliance change, new leadership, and advisor hiring.

35. Industry Language Should Come From the Market

Do not force generic sales terminology into every vertical.

Learn how the market describes:

  • The customer
  • The user
  • The location
  • The contract
  • The workflow
  • The business outcome
  • The risk
  • The decision-maker
  • The operating unit
  • The revenue model

Examples:

A dental practice may discuss providers, operatories, office managers, locations, patient flow, and compliance burden. An MSP may discuss endpoints, tickets, cloud migrations, contracts, SLAs, and managed services. A SaaS company may discuss ARR, ACV, demos, pipeline coverage, expansion, and adoption. Research should improve the language of the conversation, not decorate it.

36. Personalization vs Relevance

Personalization answers:

What do we know about this person or company?

Relevance answers:

Why does that information change the business reason for contacting them?

Weak personalization:

  • Congratulations on the promotion.
  • Saw you went to the University of Texas.
  • Loved your recent post.

Stronger relevance:

A new CRO joined while the company is opening two territories, which may change pipeline ownership and outbound capacity. Personalization is only valuable when it improves commercial relevance or human context.

37. Turn Research Into a Problem Hypothesis

Research should not end as a collection of facts. Convert it into a testable business hypothesis.

Structure:

OBSERVED FACT → PLAUSIBLE OPERATING EFFECT → POSSIBLE BUSINESS CONSEQUENCE → BUYER QUESTION

Example:

Observed fact:

The company is adding 10 AEs across two markets.

Possible operating effect:

Prospecting coverage may become uneven while the team ramps.

Possible consequence:

AEs may spend more time creating pipeline instead of progressing opportunities.

Buyer question:

“How are you handling outbound coverage while the new AEs ramp?” That is research becoming conversation.

38. Build an Account Evidence Brief

A useful account brief should stay decision-oriented.

Recommended fields:

  • Account
  • Segment
  • ICP tier
  • Industry
  • Business model
  • Size
  • Locations or operating footprint
  • Key operating conditions
  • Observed signals
  • Signal dates where relevant
  • Likely problem hypothesis
  • Evidence status
  • Likely impact
  • Primary buyer role
  • Secondary stakeholders
  • Buying-group coverage
  • Commercial potential
  • Negative evidence
  • Hard disqualifiers
  • Unknowns
  • Recommended channel
  • Recommended message angle
  • Proof most relevant to the account
  • Next research question

Do not turn the brief into a biography of the company.

39. AI Should Increase Research Capacity, Not Certainty

AI can help:

  • Summarize company information
  • Compare accounts
  • Classify industries
  • Extract public signals
  • Organize buying groups
  • Draft hypotheses
  • Highlight missing fields
  • Find contradictions across provided sources
  • Summarize call notes
  • Cluster objections
  • Identify recurring market language

Human judgment should still verify:

  • Account fit
  • Source credibility
  • Signal meaning
  • Claims
  • Commercial implications
  • Buyer-role ownership
  • Sensitive context
  • Negative evidence
  • Final prioritization

AI can make research faster. It should not turn weak evidence into confident claims.

40. Research Before Outreach vs Research During Outreach

Before outreach, research should establish enough context to earn the first interaction.

Confirm:

  • Basic fit
  • Likely buyer
  • Plausible problem
  • Useful evidence
  • Timing where available
  • Commercial viability
  • No obvious hard disqualifier

During outreach, the market provides new evidence.

Examples:

  • Gatekeeper identifies the actual owner
  • Buyer explains the current process
  • Contact refers another stakeholder
  • Objection reveals an incumbent vendor
  • Buyer clarifies timing
  • Email reply changes the account hypothesis
  • Call reveals no actual problem

Research should continue through conversation. The database is not the final source of truth.

41. Live Buyer Evidence Should Override Weak Database Assumptions

When credible buyer information conflicts with an inferred database field, update the model.

Examples:

Database says 500 employees, buyer explains the relevant division has 40. Technology data suggests one vendor, buyer confirms they migrated six months ago. LinkedIn title suggests responsibility, contact explains the function sits with operations. Public hiring suggests growth, buyer confirms hiring is replacement-only. The market should be allowed to correct the research system.

42. The 100-Point ICP Research Score

For consistency across Biznatron resources, use the same 100-point account model:

  • Company Fit: 25
  • Problem and Impact Fit: 25
  • Timing and Buying Signals: 15
  • Buyer Access and Decision Fit: 15
  • Commercial Fit: 10
  • Strategic Fit: 10

Total: 100

Possible tiers:

  • 80 to 100: Tier A
  • 65 to 79: Tier B
  • 50 to 64: Tier C
  • Below 50: Deprioritize unless materially new evidence changes the account

The score should organize judgment, not manufacture scientific precision. Two accounts can both score 82 and still be very different if one is supported by verified evidence while the other depends on assumptions.

43. Add Evidence Confidence to the Score

For important scoring criteria, record confidence separately. Use the canonical five Evidence Confidence levels. Record Contradicted separately when evidence exists against the current hypothesis:

  • Verified
  • Strong Evidence
  • Partial Evidence
  • Assumed
  • Unknown
  • Contradicted

The account score answers:

How attractive does the account appear?

Evidence confidence answers:

How much should the team trust that conclusion? Both matter.

44. Research Metrics That Matter

Track the full research-to-revenue chain where practical:

  • Accounts researched
  • ICP match rate
  • Tier distribution
  • Verified-contact rate
  • Buyer-role coverage
  • Buying-group coverage
  • Wrong-person rate
  • Referral rate
  • Signal prevalence
  • Signal-to-conversation rate
  • Research-to-connect rate
  • Research-to-positive-reply rate
  • Research-to-meaningful-conversation rate
  • Research-to-qualified-meeting rate
  • Meeting quality by ICP tier
  • Meeting-to-opportunity rate by segment
  • Pipeline by ICP tier
  • Revenue by ICP tier
  • Disqualification rate
  • Top disqualification reasons
  • Performance by industry
  • Performance by buyer title
  • Performance by signal type
  • Research time per account
  • Research time by deal tier
  • Evidence-confidence distribution

The purpose is not to prove researchers are busy. The purpose is to learn which research decisions create better downstream outcomes.

45. Research Diagnostics

High ICP match + low contactability

Check data source quality, contact coverage, title mapping, and channel access.

High contactability + low meaningful conversations

Check problem hypothesis, timing, buyer relevance, and message translation.

High conversations + low qualified meetings

Check whether the research is overestimating problem or impact fit.

High meetings + low opportunities

Check buyer authority, qualification, account economics, and handoff accuracy.

Strong early conversion + weak customer retention

Check serviceability, implementation fit, Fit Debt, and customer economics.

High research time + average outcomes

Check Diminishing Research Return and unnecessary fields.

High scores + low conversion

Audit evidence confidence and confirmation bias.

Many wrong-person responses

Audit buyer-role assumptions and buying-group mapping. Find the first research assumption that stops matching reality.

46. Common ICP Research Mistakes

  • Using employee count as the ICP
  • Buying a list before defining the problem
  • Targeting every industry that could theoretically buy
  • Using titles without understanding ownership
  • Treating funding as automatic buying intent
  • Treating technology data as verified truth
  • Over-personalizing irrelevant details
  • Researching deeply before checking basic fit
  • Ignoring negative evidence
  • Ignoring disqualifiers
  • Copying a competitor’s ICP
  • Building the ICP from one or two customers
  • Treating a large logo as automatically strategic
  • Keeping weak accounts to protect list volume
  • Allowing old fields to remain “verified” forever
  • Letting missing fields become assumptions
  • Researching after the decision is already clear
  • Failing to record contradictions
  • Treating the database as more authoritative than the buyer
  • Never revisiting the model after campaign learning

47. ICP Research QA Checklist

Before an account enters active outreach, confirm:

  • The account passes hard gates
  • The market is one the company can actually serve
  • Firmographic fit is understood
  • Operating fit is plausible
  • The problem hypothesis is explicit
  • Facts and hypotheses are separated
  • Important sources are credible
  • Critical evidence is reasonably fresh
  • Contradictory evidence has been considered
  • Negative fit has been checked
  • The right buyer role is plausible
  • Relevant stakeholders are mapped where necessary
  • Timing signals are not being confused with fit
  • Signal strength is appropriate to the message
  • The commercial potential justifies the effort
  • Research depth matches account value
  • Unknowns are visible
  • The message angle follows from the evidence
  • The team knows what to verify in conversation
  • The researcher knows why further research would or would not help

48. The Biznatron ICP Research Operating Model

Biznatron treats research as the first layer of research-led outbound execution.

The operating model can combine:

  • Market definition
  • ICP development
  • Segment prioritization
  • Hard gates
  • Account research
  • Operational-fit research
  • Buyer-role mapping
  • Buying-group mapping
  • Data validation
  • Evidence status
  • Signal research
  • Negative evidence
  • Industry language
  • Title-specific psychology
  • Problem hypotheses
  • Research scoring
  • Qualification criteria
  • Disqualification criteria
  • Cold calling
  • Cold email
  • LinkedIn
  • CRM discipline
  • QA
  • Meeting handoff
  • Campaign learning

Research before outreach does not mean maximum research before outreach. It means enough verified context to make a better decision before spending buyer attention.

49. ICP Research Decision Model

Before an account enters active outreach, ask:

  • Does it pass the hard gates?
  • Does the company fit the market we can serve?
  • What operating conditions make the problem plausible?
  • What is verified?
  • What is inferred?
  • What is unknown?
  • What evidence contradicts the current hypothesis?
  • How fresh is the important evidence?
  • What problem is most plausible?
  • What business consequence could follow?
  • Which buyer role is closest to the issue?
  • Which other stakeholders may matter?
  • What signal changes timing?
  • How strong is that signal?
  • Which of the Four Clocks may matter?
  • What commercial value could the account represent?
  • What could disqualify it?
  • How strong is the evidence confidence?
  • How much research is economically justified?
  • What should the first message test?
  • What would move the account up a tier?
  • What would move it down a tier?
  • What would make the team stop pursuing it?
  • Would more research change the next decision?

Use:

GATE → FIT → EVIDENCE → PROBLEM → SIGNAL → BUYER → ACCESS → VALUE → PRIORITY → OUTREACH → LEARNING

50. Final Principle

A strong ICP is not a description of companies the seller hopes will buy. It is a learning model for identifying companies most likely to have the right operating conditions, problem, economics, buying environment, and timing.

Strong research does four things well:

  • It finds evidence.
  • It preserves uncertainty.
  • It looks for reasons the hypothesis may be wrong.
  • It stops when the next conversation can teach more than another search.

The final formula is:

MARKET HYPOTHESIS → GATES → ACCOUNT FIT → EVIDENCE → PROBLEM LIKELIHOOD → BUYER MAP → TIMING → COMMERCIAL VALUE → PRIORITY → CONVERSATION → BUYER TRUTH → LEARNING

Research creates focus. Evidence creates confidence. Negative evidence protects capacity. Buyer mapping creates relevance. Signals improve timing. Conversation replaces assumptions with truth. Learning improves the next ICP decision. The objective is not to know everything about the account. The objective is to know enough to make the next sales decision better.

How Biznatron Helps

Biznatron builds and operates dedicated outbound programs around defined ICPs, account research, buyer-role mapping, validated data, public signals, human BDR execution, cold calling, cold email, LinkedIn, qualification, QA, CRM discipline, meeting handoff, reporting, and campaign learning.

The focus is simple:

Spend sales attention on accounts with the strongest combination of fit, evidence, timing, access, and commercial value. Keep facts separate from hypotheses. Use live buyer feedback to correct the research model. Remove weak accounts early. Make every campaign improve the next targeting decision.