SaaS metrics

One synthetic company carries the whole panel, so every number reconciles. VyapaarDesk Ltd (synthetic) sells billing-and-GST-compliance software to Indian SMEs and mid-market firms at roughly ₹2 lakh a year per account. All figures are ₹ crore, FY26.

The raw data. Opening ARR is 400. During FY26, new-logo ARR adds +100 and expansion ARR from existing customers adds +60. Contraction takes −12 and churned ARR −36. Customers start the year at 20,000. The company adds 6,000 and loses 3,200, closing at 22,800. Recognized revenue is 480 (subscription 450 + professional services 30). Subscription gross margin is 78%, against 10% on services. Opex is S&M 175 (FY25: 150), R&D 90 and G&A 50, with SBC of 24 included in those lines. D&A is 15, capex 12, cash taxes 12. Deferred revenue rose +20 during the year (FY25: +30) to close at 230. RPO is 620, of which cRPO is 440 (FY25 cRPO: 360). FY25 revenue was 375 and FY25 billings were 405.

4.3.1 Growth: ARR, MRR and the bridge

MRR is monthly recurring revenue, the subscription revenue run-rate of the current book, measured monthly. ARR = MRR × 12, or for annual-contract businesses, the annualized value of active subscriptions. Two things ARR isn't. It isn't recognized revenue, because it's a run-rate at a point in time, while revenue is a flow over a period. And it isn't GAAP or Ind AS anything. ARR is a management-defined metric, so read its definition every quarter. The ARR quality ladder shows why. Always rebuild the ARR bridge:

Closing ARR = Opening ARR + New-logo ARR + Expansion − Contraction − Churned
VyapaarDesk: 400 + 100 + 60 − 12 − 36 = 512    → net new ARR 112, growth 28%

The bridge is diagnostic gold. Picture two companies with identical 28% growth. One adds 112 of new logos with zero expansion, a leaky bucket refilled by brute force. Another adds 52 of new logos plus 60 of expansion, an install base that compounds by itself. Same growth rate, different businesses.

4.3.2 Retention: NRR/NDR, GRR, and the two churns

Net revenue retention (NRR, also called net dollar retention, NDR) takes only the customers you had a year ago and asks what their ARR is now, expansions and shrinkage included:

NRR = (Opening cohort ARR + Expansion − Contraction − Churned) / Opening cohort ARR
VyapaarDesk: (400 + 60 − 12 − 36) / 400 = 412 / 400 = 103%

Gross revenue retention (GRR) is the same computation with expansion excluded (capped at 100%). It answers a narrower question: how much of last year's revenue survived?

GRR = (Opening − Contraction − Churned) / Opening = (400 − 48) / 400 = 88%

What is a good NRR, and why does it matter?

The thresholds here come from the program brief and market convention. NRR ≥120% is great, because the company grows double digits with zero new customers. The best usage-based infrastructure names have printed 130–170% in their hypergrowth years. 110%+ is strong enterprise-grade, and 100–110% is typical of good SMB SaaS. Below 90% is a leaky bucket. The install base is melting, and every point of reported growth is bought new. For GRR, 95%+ is enterprise-grade and 85–90% typical SMB. Small businesses die, so some churn is actuarial, with nothing competitive about it. VyapaarDesk, at NRR 103% / GRR 88%, has a normal SMB profile. Judge it against SMB peers. An enterprise darling is the wrong yardstick.

Now the two churns. Logo churn = customers lost ÷ opening customers = 3,200/20,000 = 16%. Gross revenue churn = (contraction + churned ARR) ÷ opening ARR = 48/400 = 12%. When logo churn exceeds revenue churn, as here (16% vs 12%), the deaths are concentrated in small accounts, the usual and benign pattern. The reverse is far more alarming. Revenue churn above logo churn means large accounts are leaving, so investigate immediately.

Why is NRR the most valuable number on the panel? Because it compounds. An NRR-120% base doubles its revenue from existing customers alone in under four years (1.2⁴ ≈ 2.07). An NRR-90% base halves in about six and a half (0.9⁷ ≈ 0.48). Take two companies both growing 30% today, one at NRR 120 and one at NRR 95. Their five-year destinies are completely different, and they deserve completely different multiples. Before you read anything else about a SaaS company, then, find NRR and its trend. An NRR that slides from 120 to 112 to 105 across eight quarters is the earliest reliable signal that the product's expansion story is exhausted.

4.3.3 Go-to-market efficiency: CAC payback, LTV/CAC, magic number

CAC payback counts how many months of gross profit from a new customer repay the cost of acquiring them. Compute it two ways and triangulate.

Aggregate (the screener's method). Charge the prior period's S&M against the current period's net new ARR. The lag matches spend to the bookings it produced. Always adjust for gross margin as well, because you repay CAC with gross profit, never with revenue:

CAC payback (months) = Prior-period S&M / (Net new ARR × Subscription GM) × 12
VyapaarDesk: 150 / (112 × 0.78) × 12 = 150 / 87.36 × 12 ≈ 20.6 months

Per-logo (the operator's method). Isolate acquisition S&M from management's split or your own estimate. Say 70% of S&M targets new logos and the rest goes on expansion and retention. Then 0.70 × 175 = 122.5 across 6,000 new logos, which is ₹2.04 lakh CAC per logo. New-logo ARPA = 100 cr ÷ 6,000 = ₹1.67 lakh, so monthly gross profit per logo = 1.67 × 0.78 ÷ 12 = ₹0.1083 lakh. Payback = 2.04 ÷ 0.1083 ≈ 18.8 months. The two methods bracket the truth at 19–21 months.

Under 12 months is great, and that's the brief's line. 12–18 is workable, especially in enterprise. Deals there are bigger and cycles longer. Beyond 24 months the model needs either exceptional retention or cheap capital. And the direction of travel matters more than the level, as the red-flag panel shows. With 88% GRR behind it, VyapaarDesk at ~20 months is a watch item, nothing worse.

How is LTV/CAC calculated?

LTV/CAC sets lifetime value per customer against its acquisition cost:

LTV = ARPA × Gross margin % / Gross revenue churn rate
VyapaarDesk: 1.67 × 0.78 / 0.12 = ₹10.83 lakh   →  LTV/CAC = 10.83 / 2.04 ≈ 5.3×

The 1/churn term is the expected customer lifetime (1/0.12 ≈ 8.3 years). Above 3× is the classic pass mark. Below ~2×, the go-to-market machine consumes the value it creates. But treat LTV/CAC as the most gameable number on the panel. Halve the churn assumption and you double LTV. Use blended (organic-included) CAC instead of paid CAC and the ratio flatters you further. Assume today's churn holds for a decade and you're extrapolating a teenager's metabolism. Damodaran's point applies here: LTV is a mini-DCF, so discount it and stress the churn input. A 5.3× that survives churn at 20% (LTV 6.5 lakh, ratio 3.2×) can be trusted. One that collapses can't. When the two disagree, prefer CAC payback, which is nearer-term and leans on fewer assumptions.

What is the SaaS magic number?

Magic number measures sales efficiency at company level. It's annualized net new ARR per rupee of prior-period S&M (the original Scale Venture Partners form uses the quarterly revenue delta × 4 instead):

Magic number = Net new ARR / Prior-period S&M = 112 / 150 ≈ 0.75

Above 0.75 the go-to-market engine is efficient, so press harder. Below 0.5, every rupee of S&M buys less than 50 paise of annuity, and the spend should be cut or fixed. In between, 0.5–0.75 is the ambiguous middle. Magic number, CAC payback and gross margin are the same information in three costumes: payback ≈ 12 ÷ (magic number × GM). Compute all three anyway. Their definitions fail in different ways, and a company gaming one usually trips over another.

4.3.4 Margin structure: gross margin, Rule of 40, spend intensity

Gross margin. Pure subscription software should run 70–80%+, and VyapaarDesk's subscription line does, at 78%. What drags it down? Hosting-heavy or usage-based products, for one, since infrastructure SaaS runs 60s–low-70s. Bundled hardware, for another. Above all, professional services. Implementation and customization revenue carries 0–20% margins, and at a high mix it signals that the product needs consultants to work. At that point you're looking at a software-flavored services business, which the India map sets against IT services. VyapaarDesk blended: (450 × 0.78 + 30 × 0.10) ÷ 480 = 354 ÷ 480 = 73.8%. No problem there, and the services mix (6%) is modest. A falling gross margin in SaaS is always a question to run down. Look for mix shift, discounting that hits the revenue line, or cloud costs outrunning pricing.

What is the Rule of 40, and how do you read it?

Rule of 40 is the sector's one-line report card, balancing growth against profitability:

Rule of 40 score = Revenue growth % + FCF margin %      (≥ 40 = healthy)
VyapaarDesk FCF = EBIT 39 + D&A 15 − capex 12 + ΔDeferred revenue 20 − cash tax 12 = 50 → margin 10.4%
Score = 28 + 10.4 = 38.4 — just under the bar

Read the score. Don't worship it. 40+ means the trade-off between growth and margin is being run well at either end. 60%-growth at −20% margin passes, and so does 10%-growth at +30% margin. Below 20 is the brief's red flag, because the company is neither growing nor earning, which is the worst quadrant. Some people compute it on EBITDA or operating margin, so state your basis. Experts make two adjustments. They use organic growth, stripping out acquired revenue. And they sanity-check the FCF leg for SBC. A company printing "FCF margin 25%" while issuing 8% of its share count every year is partly paying its wage bill with your dilution. Recompute FCF minus SBC and see if the score survives, a test the red-flag panel repeats.

Spend intensity. For S&M ÷ revenue, VyapaarDesk runs 175/480 = 36%, typical for a ~30%-grower. Hypergrowth names run 45–60%, and mature software runs down toward 20–30%. The slope of this line as growth slows is the whole margin-maturity story you'll meet in the valuation section. R&D ÷ revenue is 90/480 = 19%, which is healthy. The normal band is 15–25%, and below ~10% in a competitive category means the company is harvesting. Watch for R&D capitalized on the quiet to flatter margins. That's the improper-capitalization flag from the shenanigans work, wearing a hoodie. G&A at 10% and falling with scale is the boring line that should stay boring.

4.3.5 The contract-anchored truth: billings, RPO, cRPO

Everything above except revenue is management-defined. Three numbers are anchored in accounting and contracts, and they're your polygraph.

Billings (calculated) = what was actually invoiced:

Billings = Revenue + Δ Deferred revenue = 480 + 20 = 500     (FY25: 375 + 30 = 405 → growth 23.5%)

RPO (remaining performance obligations) is a required disclosure under ASC 606/IFRS 15/Ind AS 115. It's the total contracted revenue not yet recognized: deferred revenue (billed, still to be earned) + backlog (contracted, still to be billed). VyapaarDesk's is 620. cRPO is the portion expected to convert to revenue within 12 months. Here that's 440, up 22.2% on FY25's 360.

How do billings and cRPO check reported ARR growth?

Now run the polygraph. Management says ARR grew 28%. Billings grew 23.5%, and cRPO grew 22%. So the contract-anchored numbers are running five points below the management metric. That gap isn't automatically fraud. There are benign explanations. A mix shift from annual-prepaid to monthly billing depresses billings without touching true ARR, and a move to shorter contract terms depresses RPO. But it's always a question you must close. The malign explanation is that "ARR" is being inflated by definition games, the kind the ARR quality ladder lists, while the invoices tell the truth. One quarter of divergence is noise. Three-plus quarters of cRPO and billings growth running below reported ARR growth is the brief's red flag, and historically it has been the earliest public signal of SaaS slowdowns.

The contracts decelerate before the story does.

The panel verdict for VyapaarDesk, in the form you'll use in the workups: Good SMB-grade SaaS. 28% growth with positive FCF and a clean 74% gross margin. Retention normal for the segment (NRR 103/GRR 88). Go-to-market efficiency right at threshold (magic number 0.75, payback ~20 months). Rule of 40 a near-miss at 38. Watch items: cRPO/billings running ~5 points below ARR growth (close the definitional question with management), and the payback trend. Nothing broken and nothing elite. The multiple must respect that.