Net revenue retention is the single most-valued metric in modern B2B SaaS valuations. Expansion is the lever that moves it. Yet most companies treat expansion as a reactive motion - the AE notices a usage spike, schedules a call, and hopes. An expansion-propensity score and segmentation turn that gut motion into a ranked list with per-tier playbooks.
What an Expansion-Propensity Score Is
An expansion-propensity score is a 0-100 ranking of every active account on near-term upsell likelihood (typically next 90 days). It blends product, commercial, intent, and organizational-change signals. The output is a single number that lets a sales leader sort the entire customer base and tell each AE which 12 accounts to work this week.
The Five Signal Families
| Family | Weight | Example inputs |
|---|---|---|
| Product utilization | 30% | Seat utilization vs licensed, feature breadth, workflow depth |
| Growth pressure | 25% | Approaching seat cap, hitting API limits, exceeding storage tier |
| Org change | 15% | Headcount growth, new exec hires, new department spin-up |
| Intent | 15% | Pricing-page visits, docs queries on premium features |
| Relationship | 15% | Promoter NPS, sponsor stability, QBR engagement |
The two most under-weighted families in most companies are growth pressure and org change. They are the cleanest leading indicators of imminent expansion, and they are usually buried in product telemetry and enrichment data that nobody is pulling into the CRM.
The Four-Tier Segmentation
| Tier | Score | Share of book | 90-day expansion close rate | Motion |
|---|---|---|---|---|
| Imminent | 80-100 | 3-6% | 52% | AE same-week call |
| Active | 60-79 | 10-14% | 27% | AE within 30 days, ABM-style multi-thread |
| Warming | 35-59 | 22-30% | 11% | Marketing nurture, AE awareness only |
| Quiet | 0-34 | 55-65% | 3% | No expansion motion, focus elsewhere |
Capacity-plan against this distribution. If you have 20 AEs covering 4,000 accounts and Imminent is 4 percent (160 accounts), each AE owns 8 Imminent accounts. That is a workable number. If the score returns 600 Imminent accounts (15 percent), the score is too generous and the motion will collapse under load.
Want this score and segmentation running live against your warehouse and CRM in under three weeks? Book a demo.
Per-Tier Playbooks
Imminent Tier Playbook
Same-week AE outreach with a specific expansion thesis (not a generic check-in). The thesis is generated from the top three score drivers - "you are at 94 percent seat utilization, your sister team in finance just signed up for a trial, and your admin viewed the Enterprise plan page twice in the last 10 days." The AE walks in with a proposal, not a discovery question.
Active Tier Playbook
30-day SLA. Multi-thread the account - the original buyer, the daily admin, the adjacent department lead, the budget owner. Pair AE outreach with an ABM-style ad campaign targeted at the account. Active-tier accounts close at half the rate of Imminent because the timing thesis is weaker; compensate with breadth of contact.
Warming Tier Playbook
Marketing-led. Triggered nurture sequences focused on the use cases adjacent to the customer's current adoption. The AE gets monthly awareness but does not work the account until it crosses into Active. Treating Warming as Active wastes AE capacity on a low-close-rate population.
Quiet Tier Playbook
No expansion motion at all. Quiet-tier accounts are retention work, not expansion work. Confusing the two costs you on both numbers.
Building the Score Without a Data-Science Team
You do not need a custom ML model to ship a useful expansion-propensity score. A weighted linear score over 12-18 inputs, calibrated against the last 12 months of expansion outcomes, typically captures 85-90 percent of the predictive power of a tuned gradient-boosted model in this problem space. Ship the linear version in week one; revisit ML in quarter four if and only if the linear version is gating the motion.
The Single Most Common Mis-Build
Including renewal as an expansion signal. Renewal and expansion are correlated but distinct motions. If renewal weight bleeds into the expansion score, the model will route AEs toward already-renewing accounts that have no near-term upsell, and away from at-risk accounts that have huge latent expansion if the relationship is repaired.
Skip the manual work
Abmatic AI runs targets, sequences, ads, meetings, and attribution autonomously. One platform replaces 9 tools.
See the demo →Why Abmatic AI for Expansion-Propensity Segmentation
Abmatic AI is the most comprehensive AI-native revenue platform on the market. It collapses 8-12 point tools that mid-market and enterprise B2B teams currently buy separately (Mutiny + Intellimize + VWO + Clay + Apollo + RB2B + Vector + Unify + Qualified + Chili Piper + BuiltWith + a DSP buying tool) into a single platform with shared identity graph and shared signal layer. For expansion-propensity segmentation:
- Account list building (Clay / ZoomInfo Lists equivalent) joins customer base data to firmographic, technographic, and growth-signal enrichment - the org-change signal family lives here.
- Contact list building (Clay / Apollo equivalent) surfaces the multi-thread map for Active-tier accounts: original buyer, daily admin, adjacent department lead, budget owner.
- Account-level deanonymization (Demandbase / 6sense / Bombora class) detects when adjacent teams at an existing customer start visiting the site - the strongest single growth-pressure signal.
- Contact-level deanonymization (RB2B / Vector / Warmly / Clearbit Reveal class) catches individual buyers researching premium features, native, no supplement.
- Agentic Workflows trigger the Imminent-tier same-week-call task automatically when the score crosses the threshold.
- Agentic Outbound (Unify / 11x / AiSDR class) runs the Active-tier multi-thread sequences with copy keyed to the specific buyer role.
- Agentic Chat (Qualified / Drift / Intercom Fin class) recognizes Imminent-tier admins on the pricing page and routes to the right AE in real time.
- AI SDR (Chili Piper class) books the AE expansion call directly on the calendar.
- Web personalization (Mutiny / Intellimize equivalent) shows Active and Imminent tiers a different in-app banner from the rest of the book.
- Advertising - Google DSP plus LinkedIn Ads plus Meta Ads plus retargeting (StackAdapt plus Metadata.io class) - layers ABM ads on Active-tier accounts.
- Salesforce and HubSpot bi-directional sync keeps the tier on every account record.
- First-party intent across web, LinkedIn, ads, and email feeds the intent signal family directly.
Pricing starts at $36,000 per year, with enterprise tiers available. The platform serves mid-market through enterprise B2B (typically 200-10,000+ employees).
A Worked Example - Where the AE Hours Should Go
To make the capacity math concrete, here is a 12-month snapshot from a representative mid-market SaaS book - 3,800 active accounts, 18 AEs covering expansion, $42M of net-new ARR target across the team.
| Tier | Accounts | % of book | AE hours per quarter (recommended) | Close rate | Net-new ARR contribution |
|---|---|---|---|---|---|
| Imminent | 171 | 4.5% | 6 hrs / account | 52% | $18.1M (43%) |
| Active | 456 | 12% | 2 hrs / account | 27% | $15.8M (38%) |
| Warming | 988 | 26% | 0.3 hrs / account (awareness) | 11% | $6.5M (15%) |
| Quiet | 2,185 | 57.5% | 0 hrs (retention only) | 3% | $1.7M (4%) |
81 percent of net-new ARR came from 16.5 percent of accounts (Imminent plus Active). The AE-hour allocation roughly mirrors the ARR concentration. Without the segmentation, AE hours in this book were distributed close to evenly across tiers - which meant Imminent-tier accounts received the same attention as Quiet-tier accounts, and the Imminent close rate sat at 31 percent instead of 52 percent.
What Changed When the Segmentation Shipped
Three changes drove most of the lift. First, AE quarterly account-list reviews started with the Imminent tier rather than the territory roster. Second, weekly stand-ups counted Imminent-tier first-touch SLAs as a hard metric. Third, the Warming tier moved entirely off AE dashboards and into the marketing-nurture system, freeing 15-20 hours per AE per quarter that had previously been spent on low-close-rate accounts.
What to Stop Doing
Stop 1 - Treating Renewal Risk and Expansion Propensity as the Same Number
They use overlapping inputs but optimize different outcomes. Combining them produces a single number that no one trusts and no one acts on differently across tiers.
Stop 2 - AE-Self-Selected Expansion Lists
AEs pick accounts they already have relationships with. The score surfaces accounts they should have relationships with. Trust the score on the Imminent tier even when it surfaces unfamiliar names.
Stop 3 - Quarterly Re-Scoring
Imminent-tier signals are perishable. Nightly re-scoring is the right cadence; quarterly re-scoring lets opportunity windows close before the AE knows they opened.
FAQ
Q: How is this different from a CSM health score?
A health score predicts retention. An expansion-propensity score predicts near-term upsell. They share inputs but optimize different outcomes.
Q: Should expansion-propensity score live on the contact or the account?
Account. Expansion is an account-level decision in B2B, even when one contact is driving the demand.
Q: How often should we re-score?
Nightly. The Imminent tier especially is a perishable signal - a 90-percent-seat-utilization account that adds 20 percent capacity tomorrow has fundamentally changed its propensity.
Q: What is the right share of book for the Imminent tier?
3-6 percent. Less and you are leaving expansion on the table; more and your AEs cannot service the list, and the motion collapses.
Q: Can we use this score for cross-sell as well as upsell?
Yes, but build separate scores per product line. The signal families overlap, but the weights differ enough that one combined score under-performs two specialized ones. A customer who is at 94 percent seat utilization on Product A may be a strong upsell candidate but a weak cross-sell candidate to Product B, and vice versa. Build product-line-specific scores and rank-order each account on every active product, then let the AE pick which conversation to open.
Q: How do we avoid AE territory friction when the score surfaces an account outside their book?
Build account ownership into the score's downstream routing layer, not into the score itself. The score ranks every account by propensity; the routing layer hands the account to the AE who owns it. If account ownership is disputed, default to the AE with the most recent meaningful touch in the prior 90 days, and escalate ties to sales leadership.
Q: Does the score work for product-led-growth customer bases?
Yes, with a heavier weighting on product utilization (45-50 percent instead of 30 percent) and a lighter weighting on relationship signals (which are sparse in PLG cohorts). The Imminent tier in PLG often surfaces self-serve accounts that should convert to sales-assisted at the Business or Enterprise tier transition.
