Account-level scoring answers the question a B2B operator actually asks: not "how healthy is this one contact" but "how healthy is this company." Devotel Orbit ships that view under Insights → Account Scores. Per-contact churn, intent, and propensity scores roll up to the company the contact belongs to, so a retention or expansion play runs on accounts instead of individual people.
This post walks through what an "account" is here, what each column on the ranked list means, how the scoring math treats high-value contacts, and how to go from reading the list to acting on it.
1. What an "account" is on this page
An account on Account Scores is automatic grouping, not an entity you build. Any distinct value on a contact's Company attribute becomes an account — "Acme Corp" on forty contacts produces one "Acme Corp" row, with no group call and no segment to save. If your contacts already carry a company name, the accounts exist; the page's definition line says this explicitly and links to Audience → Accounts for the alternative case, where account entities come from CDP group calls instead.
The distinction matters because it decides what gets scored. Two forms:
- Company attribute grouping (this page) — the free, zero-setup path. Any contact with a Company value folds into that account the moment its scores compute.
- Audience account entities (Audience → Accounts) — deliberately built entities with declared identifiers, used when the company string alone is ambiguous (franchisees, subsidiaries, aliases).
If two spellings of one real company produce two rows ("Acme" and "Acme Corp"), the fix is data hygiene on the Company attribute — the page cannot reckon with ambiguity it never sees.
2. The ranked list, column by column
Open the page from Insights → Account Scores. The ranked table shows one row per account:
- Health — the account's ABM segment badge: At risk, Expansion, Engaged, or Stable. The segment collapses the numeric columns into an operator verdict; At risk is the one an operator triages first, and it is the only segment colored as an error.
- Contacts — the number of scored contacts on the account.
- Churn risk — the account's rolled-up churn risk, shown as a percentage.
- Expansion — the account's expansion propensity: how much the signals point at this account broadening its usage.
- Intent — the account's intent score.
- Value — the account's rolled-up value (LTV-derived), formatted in your org's billing currency, not a USD hardcode.
Each numeric column sorts the list. The default sort is churn risk, highest first — the natural triage view for retention. Switch the sort to "Expansion propensity" for the growth view, or flip the order to "Lowest first" when the question is "which healthy accounts are under-attended" rather than "which are burning."
A minimum-contacts filter prunes the list to accounts of a real size: "All accounts" includes one-person companies; "2+ contacts", "5+ contacts", or "10+ contacts" focuses on accounts with enough members for the rollup to mean something. Rows load 50 at a time behind a "Load More" button that reports an honest "Showing X of Y" against the true total.
3. How the rollup math works: value-weighted churn
The headline scoring rule: account churn risk is value-weighted. A high-value contact moves the account's churn risk further than a free user does. A 90% churn risk on the contact holding 70% of the account's value is a worse signal than the same risk spread across three low-value members, and the account row reflects that.
This is a deliberate departure from a plain average. Averaging members treats a dormant free seat and an active enterprise admin identically; value-weighting does not. The result is that an account dominated by one expiring high-value relationship surfaces near the top of the churn-risk sort instead of being diluted by its quieter members.
The scores feeding the rollup come from the same predictive layer as the per-contact views on Audience → Propensity segments and Audience → Churn risk — this page re-aggregates them per company, it does not recompute them.
4. The account detail dialog
Click any account row to open its detail. The dialog answers the follow-up question the ranked list cannot: which contacts are driving this?
- Headline stats — churn risk, expansion, intent, and account value as four numbers, with the health-segment badge and the firmographic band under them.
- Members — a per-contact table listing each scored member with their own churn, intent, propensity, and LTV. This is where a bad account number gets decomposed: an account at high churn because its three biggest contacts are drifting reads completely differently from one where the drift is uniform.
If the account has many members, the member list shows the scored subset and labels the truncation honestly.
5. The firmographic band — and how enrichment upgrades it
Every account carries a firmographic band (Solo, Small, Mid-market, Large, Enterprise). Today that band is estimated from the only firmographic signal available without an external provider: the number of contacts on the account. The page is explicit about this — the "Improve account scoring with enrichment" card lists, per available enrichment provider, exactly which real firmographic signals (industry, employee count, revenue range) would replace the contact-count proxy once the provider connects, and links to Integrations → CDP to start one.
The card catalogues the signals a connected provider would add; it does not pretend scoring has already improved. Until an enrichment backend is wired, the band is the contact-count estimate and is labelled as such.
6. Acting on the ranked list
Two patterns recur:
Retention triage. Sort by churn risk, set the minimum to 5+ contacts, and read the top of the list as the account queue for the week. Open the worst account, find which members are drifting in the dialog, and run the retention play on them — a proactive call, an automated journey, or an outreach campaign — at the member level the dialog names.
Expansion capture. Sort by expansion propensity, flip to highest first, and the same controls surface accounts most likely to broaden usage. The contacts those scores came from already carry the segment the Journey builder branches on, so the account view and the campaign that answers it read the same numbers.
Either way the page is a readout, not a destination — its job is to point an operator at the right accounts weekly with honest, value-weighted numbers, and to make "which contacts are driving it" a one-click answer.