Most sales teams have an Ideal Customer Profile. Few have a scoring system. The difference is enormous.

A persona tells you what a good account looks like. A scoring system tells you how good any specific account is — in real time, against real data. Here's how to build one that your team will actually use.

What Makes a Good ICP

A good ICP has three qualities:


  1. Specificity — Not just a vertical and company size, but the actual combination of signals that predicts a deal

  2. Measurability — You can score any account against it, not just eyeball whether it fits

  3. Reproducibility — When you win, you can trace back to the ICP signals; when you lose, you can find the gap


The ICP that lives in your head as a set of loosely-defined criteria fails all three. The ICP that lives in a scoring model passes all three.

Build Your Scoring Criteria

Start with five categories. For each, decide what signals matter and how to weight them.

1. Company Firmographics (20-30% of score)

These are the basics:


  • Employee count — Most impactful predictor of deal size and need

  • Revenue — Indicator of budget and organizational complexity

  • Industry — Directly maps to your product category fit

  • Location — Determines deal velocity and travel cost

  • Founded — Older companies move slower; newer ones move faster but have less process


Weight employee count and revenue highest. Industry matters if you're vertical-specific.

2. Technographic Signals (15-25%)

What tools does this company use?


  • Competitive products — Using a competitor = aware of the problem category

  • Complementary tools — Using your integration partners = higher fit

  • Core platform — Salesforce, HubSpot, Workday = enterprise complexity


Technographic data is third-party enrichment territory. Pull from Apollo, BuiltWith, or similar sources.

3. Hiring & Growth Signals (10-20%)

Companies actively growing signal buying energy:


  • Job postings — Hiring in relevant departments = problem-space awareness

  • Headcount growth — Companies growing tend to be evaluating tools

  • Funding rounds — Recently raised = capital to spend


Hiring signals are strong early indicators. A company suddenly hiring 10 SDRs is probably evaluating a sales tool.

4. Intent Data (15-25%)

Behavior-based signals:


  • Content engagement — Downloaded your content, visited your pricing page

  • Search behavior — Searching for terms related to your category

  • Comparison activity — Visited competitor comparison pages


Intent data is the hardest to get right. Review the signals that actually correlated with closed deals in your CRM.

5. Fit Signals (15-20%)

Company-specific alignment:


  • Geography match — Are they in your sales territories?

  • Team structure — Do they have the role your product serves?

  • Use case match — Does their stated problem map to your solution?


Fit signals are often ignored but they predict rep efficiency. A perfectly scored company in the wrong geography will create travel waste and timezone friction.

Scoring Thresholds

Once you have the model, decide what scores trigger what actions:

  • 90-100: Priority accounts — Direct outreach, personalized messaging, executive touchpoints
  • 70-89: Qualified tier — Standard sequences, product-led growth triggers
  • 50-69: Nurture — Long-cycle content, marketing automation, re-engagement
  • Below 50: Passive only — No outbound investment
The exact thresholds depend on your pipeline volume. If you're drowning in 80+ accounts, raise the bar. If you're starved, lower it until you have 100 accounts per rep to work.

Automation vs. Manual Scoring

You can't manual-score 50K accounts. Build the model once, let enrichment tools auto-populate the data, score in bulk, and let reps override individual scores where they're wrong.

The human correction loop is critical. Your reps are your best source of ground-truth data. When a rep down-scores a company because the contact left, or up-scores a company because the CEO is an industry influencer, that's signal that improves the model.

Set up a quarterly review: pull accounts that were manually adjusted and see if the pattern predicts model improvements.

The ICP + Scoring Combo

ICP without scoring gives you a list. Scoring without ICP gives you a number without context. Together, they give you a prioritized, defensible account list that you can hand to any rep and trust they'll work the right targets.

Build your ICP score in 2 minutes →