Expansion Propensity Methodology

NRR Strategy Context

Net Revenue Retention (NRR) has two components:

Component	Model	Goal
Churn prevention	churn_model	Keep existing ARR
Expansion revenue	expansion_model	Grow ARR within existing accounts

A SaaS business with 100% NRR merely replaces churned ARR. A business with 120% NRR grows even if it acquires zero new customers. The expansion propensity module directly attacks the second component.

Why this matters for Dropbox-style product metrics: The classic B2B SaaS growth motion is land-and-expand — acquire at a lower tier, prove value, upgrade to the next. The expansion model identifies the "expand" signal before Sales has to guess which accounts are ready.

---

mrr_tier_ceiling_pct — Tier Pressure

The most novel of the 5 expansion features. It answers the question: "How much of their current tier's capacity has this customer already consumed?"

mrr_tier_ceiling_pct = (mrr − tier_floor) / (tier_ceiling − tier_floor)

Plan tier	Floor	Ceiling
Starter	$500/mo	$2,000/mo
Growth	$2,000/mo	$8,000/mo
Enterprise	$8,000/mo	$50,000/mo

Interpretation: - 0.0 = Customer just entered this tier (plenty of headroom) - 1.0 = Customer is at the top of their tier — the natural trigger for an upgrade conversation - > 0.8 = Tier-pressure zone — combine with premium trial signal for high-confidence expansion

Why it beats raw MRR: A $7,500/mo Growth customer has 92% tier pressure and is nearly ready for Enterprise. A $48,000/mo Enterprise customer at 95% pressure is due for a CUSTOM expansion conversation. Raw MRR conflates these two very different signals.

---

Propensity Quadrant Interpretation

The Propensity Quadrant plots all active customers on two axes:

• X-axis: Churn probability (from churn_model)
• Y-axis: Upgrade propensity (from expansion_model)

Four action quadrants:

High Propensity │ Growth Engines  │ Flight Risks
                │ (Low Churn,     │ (High Churn,
                │  High Expansion)│  High Expansion)
────────────────┼─────────────────┼─────────────────
Low Propensity  │ Stable Base     │ Churn Candidates
                │ (Low Churn,     │ (High Churn,
                │  Low Expansion) │  Low Expansion)
                └─────────────────┴─────────────────
                  Low Churn         High Churn

Growth Engines (top-left)

Low churn risk + high expansion propensity. Primary target for upgrade campaigns. Schedule QBR with upgrade conversation agenda. These accounts have proven the platform's value and are organically asking for more.

Flight Risks (top-right)

High churn risk + high expansion propensity. Counterintuitive danger zone. Do NOT start an upsell motion — upselling a churning customer accelerates churn. Senior Exec intervention required: fix the underlying issue first (usually onboarding friction or integration failure), then revisit expansion in 60–90 days.

Stable Base (bottom-left)

Low churn + low expansion. Healthy but not growing. Nurture via product-led expansion signals: in-app prompts for premium features, peer benchmarking, feature adoption campaigns.

Churn Candidates (bottom-right)

High churn + low expansion. Retention priority. No expansion conversation warranted. Focus CS effort on support quality and compliance gap remediation.

---

ARR Uplift Formula

expected_arr_uplift = (MRR × 12) × (multiplier − 1) × propensity

Why (multiplier − 1): A 3× jump from Starter to Growth is a 200% increase. The net uplift is 2× current ARR, not 3×. Using (multiplier − 1) captures only the additional revenue, making the figure defensible in a VP review where double-counting is scrutinised.

Why probability-weighted: A 100% propensity at $10k expected uplift is worth more Sales time than a 15% propensity at the same amount. The weighted figure integrates both dimensions into a single prioritisation metric.

Tier multipliers

Tier	Multiplier	Rationale
Starter → Growth	3.0×	~$1k → ~$3k MRR median ACV jump
Growth → Enterprise	5.0×	~$5k → ~$25k MRR — largest tier leap
Enterprise → Custom	1.2×	Seat/add-on upsell, not a full tier flip
Custom → (ceiling)	0.0×	No automated propensity above this

---

Conflict Resolution: Churn vs. Expansion

When both models are available (via /customers/{id}/360), the conflict matrix prevents harmful upsell motions:

Churn Risk	Expansion Propensity	Action
Low (< 0.25)	High (≥ 0.50)	Growth Engine — schedule upgrade conversation
High (≥ 0.50)	High (≥ 0.50)	⚠️ Flight Risk — Exec intervention required
High (≥ 0.50)	Low (< 0.25)	Retention priority — defer expansion
Low (< 0.25)	Low (< 0.25)	Stable Base — nurture via product signals

This logic lives in ExpansionResult.recommended_action(churn_probability=...) — it is deterministic domain code, not an LLM output.