Account-Level vs Contact-Level Scoring
Account scoring tells you which company is in market; contact scoring tells you who to email. You need both, fed from one shared signal graph.
- Account score answers 'is this company in market'; contact score answers 'who do I email'.
- Lead with account score for prospecting, contact score for sequencing known people.
- Roll contact signals up with normalization, not a raw sum that favors big orgs.
- Run dedupe and identity resolution before scoring or both numbers lie.
Two Questions, Two Scores
Account-level scoring answers whether a company is showing buying motion: multiple visitors from one domain in Snitcher, a hiring spike in Cognism, repeated pricing views in Koala. Contact-level scoring answers who inside that company is engaged and worth a personal touch: the VP who opened three Smartlead emails, the engineer who downloaded the docs. Collapsing both into a single number hides the structure of a B2B deal, where companies buy but people decide. You end up either spamming a hot account's whole org chart or ignoring an engaged champion at a quiet company.
Keep them as separate but linked numbers on a shared identity graph. The account score sets priority and the contact score routes the outreach. In Salesforce or HubSpot this maps cleanly to the account and contact objects, but only if your enrichment from Clay or Apollo keeps contacts reliably tied to the right account. When the link breaks, both scores lie, and reps lose trust in the queue.
When to Lead With Each
Lead with the account score for prospecting and prioritization. A high account score from anonymous signals like RB2B de-anonymized visits or Leadfeeder firmographic intent tells the SDR which logo to work, even before a single contact is known. From there, enrich the buying committee with Apollo or Cognism and build a multithreaded play rather than betting on one inbox. The account is the unit of revenue, so it should be the unit of prioritization.
Lead with the contact score once people are known and engaging. A champion who keeps clicking deserves a faster, warmer follow-up than a cold title on the same account, and contact scoring drives that sequencing in Instantly or Smartlead. The two scores work together: a hot account with no engaged contacts means do outbound discovery, while a hot contact at a lukewarm account means nurture and watch for the rest of the committee to wake up.
Rolling Up Without Double-Counting
Decide deliberately how contact signals roll up into the account score. Summing every contact's activity rewards big companies with many employees and punishes lean buying teams, so normalize or use a max-plus-decay roll-up instead. The cleanest pattern keeps account-native signals (domain-level visits, firmographic intent) separate from rolled-up contact signals, then blends them with explicit weights you can audit. This avoids the trap where one noisy contact inflates an entire account.
Guard against the identity problems that quietly corrupt roll-ups. Duplicate contacts, role-based inboxes, and merged accounts all double-count or strand signal, so your dedupe and identity resolution must run before scoring, not after. Store both scores back on the records in HubSpot or Salesforce and expose them in the same view so reps see company heat and person heat side by side. When both are visible, the next action is usually obvious.
- Account score answers 'is this company in market'; contact score answers 'who do I email'.
- Lead with account score for prospecting, contact score for sequencing known people.
- Roll contact signals up with normalization, not a raw sum that favors big orgs.
- Run dedupe and identity resolution before scoring or both numbers lie.
Frequently asked questions
Can I just use one combined score?
You can display a blended view, but compute the two scores separately first. Companies buy and people decide, so the inputs and the actions differ. Keeping them distinct lets you answer both 'which logo' and 'which inbox' without one signal type drowning out the other.
How do anonymous visits fit into account scoring?
De-anonymized account-level visits from tools like RB2B, Snitcher, or Leadfeeder are pure account signals, since you know the company before you know the person. They are ideal for raising an account's priority and triggering buying-committee enrichment in Apollo or Cognism. Treat them as account inputs and resolve contacts as a separate step.
How do I stop large companies from always scoring highest?
Normalize contact roll-ups instead of summing raw activity, or use a max-plus-decay approach that rewards intensity over headcount. Keep account-native signals weighted independently so a lean buying team at a focused company can still rank above a sprawling enterprise with scattered low-intent activity.
Operator-built
Built by someone who runs the playbook, not an agency reselling labor.
You own it
Your data, your CRM, your infrastructure. The system is yours.
No lock-in
Start with a free audit. No multi-month retainer to find out it works.
Privacy-first
Your data stays yours. We pen-test our own funnel before we touch yours.
▸ STOP READING. START PLAYING.
Don't just read about it. Drop your site below and see the revenue you're leaving on the table, live.