+150 XP

ARR quality: new, expansion, contraction and churn bridges

A board slide says ARR grew 40% year over year. Nobody claps until someone asks: how much of that came from new customers versus existing ones paying more? That question separates a durable business from one quietly leaking customers while sales scrambles to backfill the hole with new logos. The answer lives in an ARR bridge.

ARR (annual recurring revenue) is the annualized value of active subscription contracts. Unlike one-time revenue, ARR is meant to represent a run-rate: what you'd collect over the next 12 months if nothing changed.

Why a single ARR number lies to you

Two companies can both report $10M ARR growing to $13M ARR in a quarter. Company A got there by expanding existing accounts. Company B got there by signing new logos while losing a third of last quarter's customers. Same headline growth, very different risk profiles.

The bridge (also called an ARR waterfall) decomposes the change into components. This is standard vocabulary in SaaS board decks and in due diligence for private equity and venture deals.

The ARR bridge, component by component

Starting ARR (beginning of period) plus:

  • New ARR: revenue from brand-new customers signed in the period.
  • Expansion ARR: additional revenue from existing customers (upsells, seat additions, upgrades to higher tiers). Sometimes split into upsell (more of the same product) and cross-sell (new product lines).

Minus:

  • Contraction ARR: existing customers downgrading, reducing seats, or negotiating a lower price at renewal.
  • Churned ARR: existing customers who cancel entirely (logo churn).

Ending ARR = Starting ARR + New + Expansion − Contraction − Churn.

Worked example: one quarter, one company

Say a SaaS company starts Q1 with $10.0M ARR. During the quarter:

  • New customer contracts signed: $1.2M
  • Expansion from existing accounts: $0.6M
  • Contraction (downgrades): $0.3M
  • Churn (full cancellations): $0.5M

Ending ARR = 10.0 + 1.2 + 0.6 − 0.3 − 0.5 = $11.0M

Headline growth: 10% quarter over quarter. Looks solid. But look at composition:

  • Gross new ARR: $1.2M (60% of gross additions)
  • Expansion: $0.6M (30% of gross additions)
  • Gross churn + contraction: $0.8M

Net Revenue Retention (NRR), the ratio measuring how existing customers' revenue evolves with no new logos included, would be:

NRR = (Starting ARR + Expansion − Contraction − Churn) / Starting ARR

NRR = (10.0 + 0.6 − 0.3 − 0.5) / 10.0 = 9.8 / 10.0 = 98%

That's the flag. Total ARR grew 10%, but the existing customer base actually shrank slightly (98% NRR means you'd lose ground without new sales). This company is entirely dependent on new logo acquisition to grow, a fragile position if the sales pipeline slows or customer acquisition cost (CAC, the fully loaded cost to acquire one paying customer) rises.

Benchmarks: what "good" NRR looks like

As of recent industry surveys (treat as estimates, methodologies vary by source):

  • Best-in-class enterprise SaaS: NRR of 120% or higher (existing customers alone grow revenue 20%+ a year through expansion). Companies like Snowflake and Datadog have historically reported figures in or near this range in strong years, per their public filings.
  • Healthy mid-market SaaS: NRR around 100 to 110% is often cited as a solid benchmark in investor commentary, for example from Bessemer Venture Partners' State of the Cloud reports.
  • Below 90 to 95% NRR: generally read as a warning sign, especially for venture-backed companies raising growth capital.

Regional nuance: European SaaS benchmarks (commonly referenced via reports from firms like Dealroom or local VCs) tend to skew slightly lower on average NRR than top US comparables, partly reflecting smaller average contract values and more price-sensitive SMB (small and medium business) segments in fragmented markets. Treat any specific European NRR average as directional, not precise, since public disclosure is less standardized than in the US.

Gross Revenue Retention (GRR) is the stricter cousin: same formula as NRR but capping expansion at zero, so it only measures what you kept, not what you grew.

GRR = (Starting ARR − Contraction − Churn) / Starting ARR

GRR = (10.0 − 0.3 − 0.5) / 10.0 = 9.2 / 10.0 = 92%

GRR above roughly 90% is commonly cited as healthy for enterprise SaaS; anything meaningfully below that suggests a product or customer-fit problem, not just a pricing opportunity.

Reading the bridge like an investor

When you see an ARR bridge, ask three questions:

  1. What share of gross new ARR came from expansion versus new logos? Rising expansion share usually signals product stickiness and cross-sell success.
  2. Is churn concentrated or broad? A single large enterprise account leaving is a different story than dozens of small SMB accounts churning, the latter often points to onboarding or product-market fit issues.
  3. Is contraction increasing over time? Rising contraction (downgrades) often precedes outright churn by a quarter or two, it's a leading indicator worth tracking separately from cancellations.

Knowledge check

1. Two companies report identical headline ARR growth for the quarter. Why might a board still consider one company's growth far riskier than the other's?

2. A company's ARR bridge shows large New ARR but also large Churned ARR each quarter. What does this pattern most likely indicate?

3. Why do analysts distinguish Contraction ARR from Churned ARR rather than lumping both into a single 'revenue lost' bucket?

MULTIPLE CHOICE

4. Select ALL correct answers about what the ARR bridge is designed to reveal that a single ending ARR figure cannot.

Select all the correct answers.

MULTIPLE CHOICE

5. Select ALL correct answers about Expansion ARR.

Select all the correct answers.

A simple way to model it

Even a basic spreadsheet bridge clarifies the story. The logic in plain terms:

ending_arr = starting_arr + new_arr + expansion_arr - contraction_arr - churned_arr

nrr = (starting_arr + expansion_arr - contraction_arr - churned_arr) / starting_arr
grr = (starting_arr - contraction_arr - churned_arr) / starting_arr
new_logo_dependency = new_arr / (new_arr + expansion_arr)

That last line, new logo dependency, is a useful gut-check metric even though it's not standardized industry vocabulary: the higher it is, the more growth relies on constantly refilling the top of the funnel rather than deepening existing relationships.

What moves each bucket in practice

  • Expansion grows through seat-based pricing (more users), usage-based pricing (more consumption, common in infrastructure and API-driven products), and tiered upsells (moving customers from a "Pro" to "Enterprise" plan).
  • Contraction often follows economic tightening: procurement teams renegotiate at renewal, or customers right-size seat counts after layoffs.
  • Churn spikes around contract renewal dates, so quarterly bridges can be lumpy if a company has concentrated annual renewal cycles (common with enterprise contracts signed at year-end for calendar-year budget reasons).

🎬 [VIDEO: "SaaS Metrics: Net Revenue Retention Explained" - https://www.youtube.com/results?search_query=net+revenue+retention+explained+saas - A walkthrough of how NRR is calculated and why investors weight it heavily in SaaS valuations]

Key Takeaways

  • An ARR bridge decomposes period-over-period ARR change into new, expansion, contraction, and churn; never accept a single growth percentage without asking for this breakdown.
  • NRR measures existing customer revenue evolution (expansion included); GRR is the stricter version excluding expansion, both calculated from the same bridge inputs.
  • Benchmarks (as of recent industry estimates): NRR above 120% is best-in-class, 100 to 110% is healthy, below 90 to 95% is a warning sign; GRR above roughly 90% is generally considered healthy.
  • High reliance on new logo ARR relative to expansion ARR signals fragile growth, since it depends on continuous new customer acquisition rather than deepening existing accounts.
  • Contraction often leads churn by a quarter or two, so tracking it separately gives an early warning that gross retention alone can miss.