DemandCompassby automate rev.ops.
From The Demand Compass

What is account scoring, and why score accounts instead of people?

Lead scoring rates people. The buying decision belongs to a company. Here is how account scoring works, from resolving signals to two scores a rep can check.

Account scoring rates a whole company on how ready it is to buy, instead of rating each person who touched your marketing. Every signal, from an anonymous pricing visit to a job post to a webinar seat, is resolved to the account it belongs to and added up there. In The Demand Compass we score each account on two axes, readiness to buy and brand awareness, each from zero to ten. We do it because the person who signals is rarely the person who signs.

What is account scoring?

Account scoring is the practice of scoring the company, not the contact: every signal is resolved to its account, weighted, decayed and summed into one position per account. Lead scoring asks whether a person has done enough to deserve a sales call. Account scoring asks whether something is happening inside a company that makes a conversation worth having now.

The difference matters because B2B purchases are made by buying groups, and the people who leave the clearest traces are often not the ones you would put in your CRM. An engineer reads your docs. Someone in finance opens the pricing page twice from an office network. A new security leader is hired. None of them filled out a form, so none of them exist for a lead score.

Lead scoringAccount scoring
UnitOne contactOne company
Anonymous visitsDiscarded, no known personKept, resolved to the company
Typical outputA point total that crosses an MQL lineA position on two axes, with the signals behind it
What sales sees"This person is hot"Why this account, with dates and sources

Why does lead scoring miss accounts that are ready to buy?

Lead scoring misses ready accounts because their evidence arrives fragmented across people and channels, and a person-level model scores each fragment alone. A single in-market account might produce five signals in a month: two anonymous pricing visits, an engineer reading the docs, a relevant job post and a name at your webinar. Scattered, each one looks weak and nobody acts.

Put all five on one account and the picture changes: several people at the same company are evaluating you across three channels in the same few weeks. The signal was never weak. It was fragmented, and fragmentation is the disguise a strong account wears.

A system that scores people throws away the anonymous docs reader, because that reader is not a known contact. That reader was often the earliest and best evidence that the account was moving.

There is a timing problem too. Only a small share of any market is buying at a given moment, roughly 5% of B2B buyers by the most cited estimate, so the job is to find the few accounts that are moving, not to rank everyone by how many emails they opened. (Source: John Dawes, Ehrenberg-Bass Institute with the LinkedIn B2B Institute, 2021)

What does account scoring need before it can work?

Account scoring needs identity resolution first: a fixed order of rules that collapses fragments into people, and people into companies. In The Demand Compass that order runs from the strongest identifier to the weakest. A LinkedIn profile URL wins when present. Failing that, a verified email. Failing that, the company domain, which ties the fragment to the right account even when you cannot name the individual.

Two habits keep that layer honest. When two fragments resolve to the same record, the merge is written to a log, so every join can be checked and reversed. And every data source reports under a contract, a defined shape for its rows: a row that does not fit is quarantined for inspection, never silently dropped, because a dropped row is an account you do not know you missed.

Coverage matters as much as resolution. Map the whole addressable market as accounts first, then attach signals as they arrive. An account that is new to you then lands on a company you had already identified, instead of floating as an orphan record.

How does the Demand Compass score an account?

The Demand Compass gives every account two scores from zero to ten, one for readiness to buy and one for brand awareness, each a sum of capped sub-scores with decay applied. Readiness adds up signals such as the seniority and function behind a relevant job post, visits to the pages that matter, developer or product activity, and company or industry events. Awareness adds up connections and follows into your team, reactions to what you publish, recent newsletter engagement, event scans and ad engagement.

Decay is what keeps the score current. The decay rule says a signal is evidence with a half-life: readiness fades fast, awareness fades slowly, and a funding round from fourteen months ago is not current intent. Read more in the decay rule.

Next to the two numbers, the engine writes plain-language tags for each signal that fired, such as "senior security role posted" or "pricing page, three visits". The two scores place the account in one of four quadrants.

QuadrantAwarenessReadinessThe move
ColdLowLowGet known, so you are the first choice when they are ready
NurtureHighLowKeep in touch
In-MarketLowHighReach them first
Sales-ReadyHighHighCall today

DemandCompass, the software built on the framework, reads 81 signal types from 12 sources and rescores accounts every 5 minutes. The structure carries over to any business. The weights do not: how much a pricing visit is worth against a job post is a hypothesis you calibrate against your own closed and lost deals.

Should AI decide the account score?

AI should not decide the account score; in The Demand Compass the score is plain arithmetic, so the same signals with the same weights and decay always produce the same number. That is what lets anyone recompute it, and what lets a rep trust it. A language model that answers "this looks like an 8" cannot show its work and may answer differently the next day.

AI does have a place, at the edge, where a signal needs judgment rather than counting. Is this company really building with AI agents, or does the word just appear on its site? Was there a real security incident there recently? An AI agent can read the web and answer that. It produces one input, a tag and a sub-score, never the score itself.

The routing never says "this account is hot." It says why: new VP of Sales twelve days ago, pricing page three times this week, docs read twice, readiness eight, awareness four, In-Market.

What changes when you score accounts instead of leads?

When a team scores accounts, the list gets shorter and sales starts working it. Northwind, a composite of the cybersecurity deployments we have run, had a list of 3,200 "engaged" contacts that sales would not touch. Rescoring the same contacts on account signals flagged 41 accounts as in-market that quarter. Sales worked the 41 by hand, 9 became real opportunities, and three of those would not have been touched under the old list. These are our own deployment numbers.

3,200engaged contacts rescored
41in-market accounts surfaced
9real opportunities

Filtering is part of the result. In one deployment over one quarter, the system captured 829 signals and surfaced 150 of them to a person, which means 82% were filtered out on purpose. A rep who receives fewer, better accounts, each with its reasons, is a rep who acts on them.

Questions readers ask next

Is account scoring the same as account-based marketing?

No. Account-based marketing chooses a target list and runs programs against it. Account scoring tells you which accounts in your market are moving right now, so you can decide where that effort goes first.

Do I need third-party intent data to start scoring accounts?

No. Your own data is enough to start: anonymous website visits resolved to companies, job posts, form fills and event lists. Third-party intent is one more source on the readiness axis, not the foundation.

What happens to contact-level data in account scoring?

It stays. Contacts become evidence attached to the account, and you still need them to know who to call. What changes is that no single contact decides whether the account deserves attention.

How often should account scores be updated?

As often as new signals arrive, with decay applied each time, because readiness fades within weeks. A score refreshed once a quarter keeps flagging accounts whose window has already closed.

Do this Monday

Take the last thirty days of anonymous visits to your pricing and docs pages. Resolve them to companies, collapse the repeat visitors, and rank the accounts by how many times someone from each came back. That ranked list is your first account-level readiness signal. It was sitting in traffic you already paid for, invisible only because nobody had rolled it up to the account.

This note draws on the chapter The scoring engine of The Demand Compass: The Signal-Based GTM Framework for B2B Marketing Leaders in Complex Sales, by David Moreira and Marcos Stubrin.

Get the book →