Quick answer
Account intelligence is the practice of merging what you know about a company, such as visits, chats, product events and CRM data, into one account-level view and score. The unit is the organization, not the lead, so you can see whether it is engaged and what to do next.
Formula: Example health (0 to 100) = 40 x active members / members + 30 x active days / 30 + recency (30 if seen within 7 days, 15 within 30 days, else 0)
What is account intelligence?
Account intelligence is the practice of gathering what you know about a company in one place so you can judge whether it fits, whether it is engaged and what to do next. The unit is the account (the organization), not the individual lead. Instead of asking "what did this person click?", you ask "how many people at this company are active, what are they doing and is it getting better or worse?".
One B2B data vendor defines it as the aggregation of first- and third-party data, turned into insights that help marketing and sales understand accounts and leads. The definition is useful but it comes from a company that sells such data, so read the second half of this page as a method you can run with your own data, with or without a vendor.
The term appears in three places, and it helps to keep them apart:
- Sales and marketing teams use it to prioritise target accounts: who to call, who to nurture, who to leave alone.
- Customer success teams use the same idea after the sale, to see whether a customer organization is still using the product.
- Product teams use it to see usage by company rather than by individual, because in B2B products the buyer, the users and the admin are different people.
What they share is a merge: signals from visits, conversations and product events are joined by an account key, and the account gets a status. A person-level view cannot say whether a customer is still using the product. A visitor who never logs in tells you nothing about the organization, and a single enthusiastic user can hide a team that has gone quiet.
What data goes into account intelligence?
Vendors list many data types; they fall into two groups, first-party and third-party, as in the same vendor's explainer. The table adds product usage and conversations, which matter for teams that sell software.
| Group | Data type | Example | Where it comes from |
|---|---|---|---|
| First-party | Website behaviour | Pages viewed by people from the account | Your own analytics |
| First-party | Product usage | Events by members of the account | Your own product tracking |
| First-party | Conversations | Chats and support threads | Your chat or help desk |
| First-party | CRM and sales engagement | Deal stage, emails, calls | Your CRM |
| Third-party | Firmographic | Industry, size, location | A data provider |
| Third-party | Technographic | Tools the company uses | A data provider |
| Third-party | Intent | Topics the company is researching elsewhere | A data provider |
| Third-party | News and hierarchy | Funding, parent and subsidiary links | A data provider |
You can start with first-party data alone. It is the part you have permission to hold, you know where it came from, and it answers the most useful question: is this account using us, and is that changing? Third-party data adds fit and timing, but it also adds cost, accuracy questions and privacy duties.
Account intelligence vs lead intelligence, ABM and a CDP
| Term | Unit | Question it answers | Relationship |
|---|---|---|---|
| Account intelligence | The organization | What do we know about this company and what should we do? | The insight layer |
| Lead intelligence | The individual | How interested is this person? | Feeds account intelligence |
| Account-based marketing | Target accounts | How do we run campaigns for a short list of companies? | Uses account intelligence to choose and measure |
| Customer data platform | Profiles | Where is customer data collected and unified? | Often the plumbing underneath |
A short way to remember it: a CDP stores and unifies the data, account intelligence reads it at company level, and account-based marketing acts on it.
Worked example: merging one account's signals
Sample data: Acme is a customer with five identified members. Over the last 30 days its members produced the following, each tagged with the account key company:acme.
| Signal | Count | Detail |
|---|---|---|
| Website visits | 14 | Pricing page 4, docs 6, blog 4 |
| Chats | 3 | Two product questions, one billing question |
| Product events | 41 | Mostly reports opened and exports |
| Members active | 3 of 5 | Two members have not appeared in 30 days |
| Active days | 9 of 30 | |
| Last seen | 4 days ago |
One merged view. Instead of five people's histories, you read one timeline: three active members, a billing question and two visits to pricing in the last week suggest a renewal conversation is near. None of the individual signals shows this on its own.
One score. To compare accounts, combine the signals into a number. A simple health score (0 to 100) uses three parts: members active in 30 days (40 points), active days out of 30 (30 points), and recency (30 points if last seen within 7 days, 15 within 30 days, 0 after that). For Acme:
- Members: 40 × 3 ÷ 5 = 24
- Active days: 30 × 9 ÷ 30 = 9
- Recency: seen 4 days ago = 30
- Health = 24 + 9 + 30 = 63
Compare Beta, another account with eight members, one active member, 2 active days and last seen 20 days ago: 40 × 1 ÷ 8 = 5, plus 30 × 2 ÷ 30 = 2, plus 15 for recency, so 22. Acme is healthy; Beta is drifting even though somebody there still shows up.
As a decision aid for this example only, not an industry standard: review any account below 40 this week, and look at accounts between 40 and 70 monthly. The thresholds should come from your own history: take accounts that churned and see what their score was 60 days earlier.
Two cautions about the score: it is a summary of behaviour, not a prediction, and it changes with the weights you choose, so keep the formula visible to the people who use it.
How to build account intelligence in five steps
- Pick the account key. Decide what an account is for your business: a company, a team, a store. Use a stable id, not a name.
- Identify people before you group them. An account is made of known people. Anonymous visitors cannot be assigned to an account by behaviour alone.
- Attach the key everywhere. Put the same account id on product events, chats and server events, so they merge.
- Define health with a small number of parts. Three to four signals you can explain are better than twenty you cannot. Write the formula down.
- Review it on a schedule and test it. Weekly for target accounts. Every quarter, check which accounts that churned had a low score beforehand; if none did, change the formula.
Privacy and what account intelligence cannot do
Account-level numbers about a company are not personal data, but the pieces that build them often are. Under the GDPR, personal data is any information relating to an identified or identifiable natural person, including by reference to an online identifier. Member lists, individual event histories and chat transcripts all qualify. Treat the member-level data accordingly: collect what you need, set a retention period, and be able to delete one person's data.
Account intelligence also has limits:
- It cannot tell you who anonymous visitors are. If nobody is signed in, there is no account.
- It is only as good as the account key. If the same company appears under three ids, you get three half-views.
- It does not predict. A low score is a reason to look, not proof of risk.
- Third-party data goes stale. Firmographic and intent data are snapshots; check how old they are.
Common mistakes
- Scoring accounts on page views alone. Reading is cheap; use people active, days active and recency.
- Counting members without checking identity. Shared logins make one person look like a team.
- Mixing accounts and leads in one number. Keep the unit consistent.
- Setting thresholds once and never revisiting them. Test the score against real outcomes.
- Acting on a score without reading the timeline. The merged timeline is the explanation; the number is only the sort order.
Account intelligence in TapCub
TapCub's version is built from first-party data and gives you an organisation view; it does not supply third-party firmographic or intent data, so traits such as company name or plan come from what you pass in.
- Identify the user. After login, call
tapcub.login, so the person exists in All-platform analytics › Users › People with a timeline of behaviour. - Attach the account. Call
tapcub.group('company', 'acme', { name: 'Acme' }). Later events in that session stay on the account; server events can carry the same account field. One person can belong to several accounts; anonymous visitors are never members. - Open the account list. All-platform analytics › Users › Accounts shows the number of accounts, accounts active in 30 days and members. Open a row for 30-day activity, top events, members, traits and hierarchy; you can set a parent account, for example a store under a region.
- Read the health score. Sort by Last active, Members or Health. Health is 0 to 100 and is computed exactly as in the example above (40 for the share of members active in 30 days, 30 for the share of active days, 30 for last seen within 7 days or 15 within 30 days); it is blank if the account has no last-seen time.
- Ask questions at account level. Switch the funnel or retention identity to account when the question is about the organisation. In that view, rows without an account key are not counted, and you cannot open person-level details or add segments to the analysis.
Chats from identified visitors show who they are, so the same person's conversations and behaviour sit side by side. For setup details see analyze accounts with groups.
Sources
- Demandbase, What is account intelligence?, vendor explainer, checked 2026-11-20.
- European Union, Regulation (EU) 2016/679 (GDPR), Article 4(1), checked 2026-11-20.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Product Adoption Consultant
Explores customer onboarding, live conversations, and self-service, turning common usage questions into practical guidance and improvements.