Growth Metrics Cheatsheet: 25 Definitions Growth Engineers Should Have Memorised

What are SaaS growth metrics, and why memorise them?

SaaS growth metrics are quantitative measures of revenue, retention, acquisition, and product engagement that determine whether a subscription business is healthy. They cluster into five families: revenue (ARR, MRR, ARPU, ARPA), retention (NRR, GRR, churn), engagement (DAU/MAU, L7, retention curves), acquisition efficiency (CAC, LTV, payback, magic number, Rule of 40), and product-led indicators (activation, time-to-value, k-factor, NPS, CSAT).

Growth engineers memorise these because boardroom conversations move at the speed of fluency. If your CFO says 'NRR slipped to 104, but GRR held at 92,' you should already know that expansion revenue is offsetting logo loss, that GRR is the cleaner satisfaction signal, and that an enterprise NRR of 104% is below the 118% segment median (SaaS Mag, 2026).

Every metric below comes with a 'when it lies' note. That is the part nobody else writes, and it is the part that prevents you from optimising the wrong number for six months.

What are the core SaaS revenue metrics?

Six metrics describe how subscription revenue is shaped: ARR and MRR measure recurring revenue, ARPU and ARPA measure unit economics, and expansion and contraction explain how that revenue moves inside your existing book of business.

1. ARR (Annual Recurring Revenue)

• Definition: Annualised value of all active subscriptions, excluding one-time fees and professional services.
• Formula: ARR = MRR × 12 or sum of all annual contract values.
• When it lies: ARR includes contracts that have not yet renewed. A customer in month 11 of a one-year contract counts at full value even if they have already told you they are leaving.

2. MRR (Monthly Recurring Revenue)

• Definition: Predictable, normalised monthly subscription revenue from all active customers.
• Formula: MRR = Σ (Active Subscriptions × Monthly Price).
• When it lies: Annual prepays converted to MRR can mask a business that has stopped acquiring monthly customers. Always look at New MRR, Expansion MRR, Contraction MRR, and Churned MRR separately.

3. ARPU (Average Revenue Per User)

• Definition: Average monthly or annual revenue generated per individual paying user.
• Formula: ARPU = Total Revenue / Total Active Users (Paddle).
• When it lies: ARPU averages Enterprise whales with Free-tier minnows. A rising ARPU during a price hike can hide that you lost half your low-tier user base.

4. ARPA (Average Revenue Per Account)

• Definition: Average revenue per account, where one account contains many users. Standard in B2B SaaS.
• Formula: ARPA = MRR / Total Active Accounts (ChartMogul).
• When it lies: ARPA looks healthy while seat counts inside accounts collapse. Always pair ARPA with seats-per-account.

5. Expansion Revenue

• Definition: Additional MRR from existing customers via upsells, cross-sells, seat expansion, or usage growth.
• Formula: Expansion MRR = New MRR from existing customers in period.
• When it lies: Pricing-led expansion (you raised prices) and product-led expansion (they used more) look identical in dashboards. Only the second is durable.

6. Contraction Revenue

• Definition: Lost MRR from existing customers downgrading or reducing seat counts without fully churning.
• Formula: Contraction MRR = Σ (MRR reductions from existing customers).
• When it lies: A slow drift to cheaper plans looks like 'good retention' on logo churn dashboards. It is actually slow-motion revenue churn.

What is the difference between NRR and GRR?

NRR (Net Revenue Retention) measures how revenue from existing customers changes including expansion. GRR (Gross Revenue Retention) measures the same cohort but caps at 100% by ignoring expansion. The gap between them is the most diagnostic number in SaaS.

According to SaaS Mag (2026), public SaaS companies post a median NRR between 110-115%, with enterprise products hitting 118% and SMB-focused products at 97%. Top-quartile companies exceed 130%. SaaS Capital's 2026 bootstrapped benchmarks show bootstrapped companies with $3M-$20M ARR run a median 103% NRR and 91% GRR.

7. NRR (Net Revenue Retention)

• Definition: Percentage of recurring revenue retained from a cohort of customers, including expansion, after one year.
• Formula: NRR = (Starting MRR + Expansion - Contraction - Churn) / Starting MRR × 100.
• When it lies: A handful of large expansion deals can mask widespread SMB churn. A 115% NRR with 30% logo churn is a leaky bucket disguised as growth.

8. GRR (Gross Revenue Retention)

• Definition: Percentage of recurring revenue retained from a cohort, excluding expansion. Caps at 100%.
• Formula: GRR = (Starting MRR - Contraction - Churn) / Starting MRR × 100.
• When it lies: GRR is hard to fake but slow to react. Cohorts churn over months, so today's GRR reflects last year's onboarding quality. Pair with leading indicators (activation rate, time-to-value).

9. Logo Churn (Customer Churn)

• Definition: Percentage of customers (regardless of size) who cancelled in a period.
• Formula: Logo Churn = Customers Lost in Period / Customers at Start of Period × 100.
• When it lies: Logo churn treats a $50/mo customer the same as a $50,000/mo customer. Low logo churn with high revenue churn means you are losing whales while keeping minnows.

10. Revenue Churn (Gross MRR Churn)

• Definition: Percentage of MRR lost from cancellations and downgrades.
• Formula: Gross Revenue Churn = (Churn MRR + Contraction MRR) / Starting MRR × 100 (Wall Street Prep).
• When it lies: Negative net revenue churn (when expansion exceeds losses) feels like immortality. It often reflects pricing power that is one bad release away from evaporating.

Which retention and engagement metrics should you track?

Retention and engagement metrics describe whether users keep returning and using the product. They are leading indicators for revenue retention. The Sequoia retention curve, the Andrew Chen power-user (L7) curve, and DAU/MAU are the canonical four.

11. Retention Rate

• Definition: Percentage of users from a starting cohort who remain active after a defined period (Day 1, Day 7, Day 30, Day 90).
• Formula: Retention Rate = Active Users on Day N / Cohort Size on Day 0 × 100.
• When it lies: Aggregate retention rate hides cohort drift. A product that improved onboarding three months ago looks identical to one that did not, until you split by cohort.

12. Cohort Retention

• Definition: Retention curve measured by grouping users by signup week or behavioural trigger, then tracking each group separately over time. Per Lenny Rachitsky, the curve must flatten for a business to be sustainable.
• Formula: Build a triangle table: rows = cohort start date, columns = weeks since signup, cells = % of cohort still active.
• When it lies: If your y-axis starts at 0% but your curve flattens at 5%, you do not have product-market fit. Do not let smoothed curves convince you otherwise.

13. DAU/MAU (Stickiness)

• Definition: Ratio of daily active users to monthly active users. Indicates how often a typical monthly user shows up in a given month.
• Formula: DAU/MAU = Daily Active Users / Monthly Active Users × 100.
• When it lies: DAU/MAU above 50% is famous (Facebook era). For B2B tools used Mon-Fri, the math caps near 71% (5 days / 7 days). Comparing a B2B tool's DAU/MAU to a social app is meaningless.

14. L7 (Power User Curve)

• Definition: Histogram showing, for users active in the last 28 days, how many days out of 7 they were active. Coined by Andrew Chen.
• Formula: Plot count of users by 'days active in last 7' (0 through 7). Healthy products have a smile shape with a power user spike at 7.
• When it lies: Average DAU/MAU can hide that 90% of usage comes from 5% of users. L7 surfaces the dependency. If your L7=7 bucket disappears, revenue follows within a quarter.

How do you measure customer acquisition efficiency?

Five metrics tell you whether your go-to-market motion creates value or burns it: CAC, LTV, CAC payback, magic number, and Rule of 40. Together they answer 'is the next dollar of sales and marketing spend going to make us more profitable?'

15. CAC (Customer Acquisition Cost)

• Definition: Fully loaded cost to acquire one paying customer, including sales salaries, marketing spend, tools, and overhead allocation.
• Formula: CAC = Total S&M Spend in Period / New Customers Acquired in Period.
• When it lies: Blended CAC averages free organic signups with paid enterprise deals. Always split paid CAC vs organic CAC vs channel CAC.

16. LTV (Customer Lifetime Value)

• Definition: Total gross profit a customer is expected to generate over their entire relationship.
• Formula: LTV = (ARPA × Gross Margin) / Customer Churn Rate (Wall Street Prep).
• When it lies: LTV uses today's churn rate to predict 5-year value. If churn rises 2 points, LTV can halve overnight. Healthy LTV:CAC ratio is 3:1 to 5:1.

17. CAC Payback Period

• Definition: Months required for gross-margin-adjusted revenue from a new customer to repay the cost of acquiring them.
• Formula: CAC Payback = CAC / (ARPA × Gross Margin %) (Drivetrain). The 2026 B2B SaaS median is 15 months; under 12 months is healthy.
• When it lies: Payback ignores churn after the payback point. A 6-month payback with 24-month average tenure is great. A 6-month payback with 9-month tenure is a money pit.

18. Magic Number (Sales Efficiency)

• Definition: How many dollars of new annualised recurring revenue are generated per dollar of S&M spend in the prior quarter. Coined by Scale Venture Partners' Rory O'Driscoll.
• Formula: Magic Number = (Current Quarter Revenue - Previous Quarter Revenue) × 4 / Previous Quarter S&M Spend. Scale's healthy baseline is 0.7x.
• When it lies: Quarter-on-quarter movements can swing wildly with a single big deal. Use trailing four quarters and pair with payback period.

19. Rule of 40

• Definition: A SaaS company's revenue growth rate plus its EBITDA margin should sum to 40% or higher. Popularised by Brad Feld.
• Formula: Rule of 40 = Revenue Growth Rate (%) + EBITDA Margin (%). Per Abacum (2026), the 2026 sector median sits near 12%, well below the historical 38-45%.
• When it lies: A 60% growth + -20% margin company and a 10% growth + 30% margin company both score 40. They are entirely different businesses. Always show the components.

What activation, virality, and sentiment metrics matter most?

These six metrics describe whether new users reach value, whether existing users bring more users, and whether anyone actually likes the product. They are the core PLG (product-led growth) instrument panel.

20. Activation Rate

• Definition: Percentage of new signups who complete the milestone that signals they have experienced your product's core value.
• Formula: Activation Rate = Users Who Hit Activation Event / Total New Signups × 100. Per Userpilot (April 2026), the SaaS average is 36%; below 20% indicates broken onboarding.
• When it lies: Activation events get watered down to flatter the dashboard. If 'activated' just means 'signed up and clicked once,' you are measuring noise.

21. Time-to-Value (TTV)

• Definition: Elapsed time from first interaction to the moment a user experiences your product's core benefit. Time-to-First-Value (TTFV) measures the first 'aha' moment specifically.
• Formula: TTV = Timestamp of Value Event - Timestamp of Signup. Teams measuring TTV see 15-25% higher trial-to-paid conversion.
• When it lies: Median TTV looks fine while the long tail (the 30% who never reach value) churns silently. Use a TTV histogram, not just the median.

22. Aha Moment

• Definition: The specific event that correlates most strongly with long-term retention. Defined by Sean Ellis and made famous by Facebook's '7 friends in 10 days.' It is a behavioural threshold, not a feeling.
• Formula: Use logistic regression or correlation analysis on early-user behaviours vs Day-30 retention. Pick the event with the steepest retention lift.
• When it lies: Correlation is not causation. Users who add 7 friends might be retaining users, not retaining because of friends. Run an A/B test before betting your roadmap on it.

23. K-Factor (Viral Coefficient)

• Definition: Number of new users each existing user generates through invitations or referrals over a defined period.
• Formula: K = i × c, where i = invitations sent per user and c = invitation-to-signup conversion rate (Wall Street Prep). K > 1 means exponential growth.
• When it lies: K-factor without cycle time is meaningless. K=1.2 with a 6-month cycle is a rounding error. K=0.8 with a 1-week cycle is a rocket ship that needs paid acquisition to top up.

24. NPS (Net Promoter Score)

• Definition: Customer loyalty measure. Ask 'how likely are you to recommend us, 0-10?' Then NPS = % Promoters (9-10) - % Detractors (0-6).
• Formula: Range is -100 to +100. SaaS industry average is ~31; B2B SaaS clusters at 36, B2C SaaS at 47-54 (CustomerGauge). Top scorers: GitHub +73, Notion +71, Stripe +67.
• When it lies: NPS is sample-biased. Power users respond, churned users do not. A rising NPS while logos churn means you are surveying survivors.

25. CSAT (Customer Satisfaction)

• Definition: Touchpoint-specific satisfaction score, usually a 1-5 rating after a support ticket, onboarding milestone, or release. Top-performing SaaS targets 90%+ CSAT.
• Formula: CSAT = % of responses scoring 4 or 5 / Total responses × 100.
• When it lies: CSAT measures the moment, not the relationship. A user can rate every support interaction 5/5 and still cancel because the product does not solve their job.

How should growth engineers actually use these 25 metrics?

Pick one north-star metric per business stage, three to five supporting metrics, and revisit the cheatsheet quarterly. Memorisation matters because real-time decisions happen faster than dashboards refresh. Here is the working model:

1. Pre-PMF: Activation rate, cohort retention curve flattening, time-to-value, aha moment definition.
2. Early PMF ($1M-$10M ARR): Add MRR, NRR, GRR, logo churn, CAC payback.
3. Scaling ($10M-$50M ARR): Add ARPA, expansion MRR %, magic number, Rule of 40.
4. Public-ready ($50M+): All 25, with quarterly cohort and channel decomposition.

The trap most growth engineers fall into is dashboard worship. A metric that goes up is not the same as a metric that goes up because of something you did. Every metric on this list has a 'when it lies' note for one reason: directional confidence is more dangerous than honest uncertainty.

The Princeton GEO research, Conductor benchmarks, and Benchmarkit's 2025 SaaS Performance report all show the same pattern: companies that publish their metric definitions (formulas, edge cases, exclusions) outperform those that do not. Internal clarity beats external sophistication. Print this cheatsheet. Tape it next to your monitor.

Metric	Formula	Healthy Benchmark (2026)	When It Lies
ARR	MRR × 12	26% YoY median growth	Includes contracts that haven't renewed
MRR	Σ (Subs × Monthly Price)	Track New/Expansion/Churn separately	Annual prepays mask monthly stagnation
ARPU	Revenue / Users	Varies by segment	Averages whales with minnows
ARPA	MRR / Accounts	Higher than ARPU in B2B	Hides seat-count contraction
Expansion MRR	New MRR from existing customers	20-30% of new MRR	Price hikes look like product growth
Contraction MRR	MRR lost to downgrades	<2% of starting MRR	Slow plan-drift looks like 'good retention'
NRR	(Start + Exp - Cont - Churn) / Start	110-115% public median	Big expansion masks SMB logo churn
GRR	(Start - Cont - Churn) / Start	85-95% B2B SaaS	Slow to react; reflects last year's onboarding
Logo Churn	Customers Lost / Customers Start	<5% annual for B2B	Treats $50 and $50K customers equally
Revenue Churn	(Churn + Contraction MRR) / Start MRR	<10% annual gross	Negative net churn feels invincible
Retention Rate	Active Day N / Cohort Day 0	Curve must flatten	Aggregates hide cohort drift
Cohort Retention	Triangle table by signup week	Flat curve = PMF	Smoothed curves can hide truth
DAU/MAU	DAU / MAU	>50% (B2C), 20-30% (B2B)	B2B caps at ~71% structurally
L7	Active days in last 7 by user	Power-user spike at 7	Hides 5%-of-users dependency
CAC	S&M Spend / New Customers	Channel-dependent	Blends paid and organic
LTV	(ARPA × GM) / Churn	3-5x CAC	Sensitive to small churn changes
CAC Payback	CAC / (ARPA × GM%)	<12 months healthy	Ignores post-payback churn
Magic Number	(ΔRev × 4) / Prior Q S&M	0.7x baseline (Scale VP)	Single big deal swings the number
Rule of 40	Growth % + EBITDA Margin %	40+ (median 12% in 2026)	60+(-20) and 10+30 score the same
Activation Rate	Activated Users / Signups	36% SaaS average	Watered-down 'activation' events
Time-to-Value	Value Event - Signup	<1 day for PLG	Median hides the broken long tail
Aha Moment	Behaviour with steepest retention lift	Defined per product	Correlation isn't causation
K-Factor	Invites × Conversion Rate	>1 = exponential growth	Cycle time matters more than K
NPS	%Promoters - %Detractors	31 SaaS avg, 70+ elite	Power users respond; churned don't
CSAT	% scoring 4-5 / Total	75-90% target	Touchpoint score hides relationship rot