ARR Disclosure in S-1: Cohorts, NRR and GRR
For a SaaS CFO preparing an IPO, ARR disclosure in S-1 starts with a calculation the reporting team can reproduce. Define the recurring revenue population, calculate retention and explain growth without confusing new customers with a healthier existing customer base.
Three decisions come first:
- Specify what enters ARR and which renewal assumptions annualization introduces.
- Keep the opening customer cohort fixed when calculating retention.
- Identify whether the published rate is a point-in-time ratio or an average, including its weights.
This discussion addresses domestic SaaS issuers using Form S-1. Foreign private issuers preparing Form F-1 need a separate applicability assessment.
What should ARR disclosure in S-1 explain?
Explain the measure's calculation, usefulness and limitations. ARR describes an issuer-defined recurring business measure at a measurement date. It isn't interchangeable with recognized revenue, cash receipts, deferred revenue, bookings or remaining contractual commitments. Define the activity the figure captures before explaining growth.
Item 303 of Regulation S-K requires material information about financial condition and results, including relevant statistical data and explanations of material changes. It doesn't prescribe one SaaS ARR formula or require every issuer to disclose ARR, NRR and GRR.
The SEC's January 30, 2020 KPI guidance calls for a clear definition and calculation, why investors find the metric useful, and how management uses it. Material assumptions deserve explanation too. It supplies a disclosure framework, not a standardized ARR formula.
Resolve these scope choices; the table is a decision aid, not mandatory inclusions or exclusions.
| Input or event | Question the definition must resolve |
|---|---|
| Active subscriptions | Does ARR use current monthly pricing, annualized contract value or another basis? |
| Setup fees and services | Are one-time charges excluded? Which recurring services qualify? |
| Usage charges | Do you count commitments, recent consumption or estimated future usage? |
| Discounts | Which net price applies, including temporary concessions? |
| Signed contracts awaiting activation | Are these included before service begins? |
| Contracts expiring within twelve months | Does annualization assume renewal, and on what terms? |
| Product migrations | Does a customer enter ARR because the commercial model changed? |
| Acquired customers | When do they enter ARR and the retention cohort? |
For an illustrative $10,000 monthly subscription, multiplying by twelve produces $120,000 of annualized run rate. That arithmetic alone doesn't establish an enforceable twelve-month commitment. If you use the phrase contracted ARR, define the contract population and period it represents.
Is ARR an operating KPI or a non-GAAP measure?
Classify the calculation, not the acronym. A dollar-denominated contractual operating measure isn't automatically non-GAAP. A calculation that adjusts a comparable GAAP financial measure requires a different assessment. Calling adjusted revenue “ARR” or a “KPI” doesn't determine its treatment.
Item 10(e) excludes operating and statistical measures, and specified ratios using GAAP or other non-non-GAAP inputs, from the non-GAAP definition. For a measure within that definition, an S-1 must meet applicable requirements, including equal or greater prominence for the comparable GAAP measure, reconciliation, investor usefulness and management-use disclosure.
Document the inputs, adjustments and relationship to a comparable GAAP measure. Test changes to recognition or measurement carefully. The SEC staff's non-GAAP interpretations, Questions 100.04 through 100.06, address individually tailored adjustments, labels and measures that remain misleading despite extensive disclosure.
For the wider presentation framework, see Finrep's non-GAAP financial measures guide. The issuer's actual calculation and primary SEC requirements control the classification.
Which customers belong in the retention cohort?
Start with the customers present at the opening measurement date and follow that same population. Keep churned customers in the opening denominator, even when their closing ARR is zero. Exclude genuinely new customers from the existing-customer retention numerator.
The word “customer” needs a policy. A subscription, billing account, subsidiary and ultimate parent can produce different cohorts even when the commercial relationships haven't changed.
Klaviyo's August 25, 2023 S-1 defined a customer as a distinct paid subscription. One organization could therefore represent multiple customers, subject to its stated consolidation policy. Its NRR calculation also included subscriptions deactivated and reactivated within the twelve-month period.
Set rules for consolidations, acquisitions, reactivations, product boundaries, currency and renewal grace periods. Apply them to both opening and closing values.
In the example below, removing churned customer C would reduce opening ARR from $400,000 to $300,000. Dividing the remaining customers' $340,000 by that smaller denominator produces 113.3%, rather than 85%. That measures performance of survivors instead of retention of the opening cohort.
How can ARR grow while the customer cohort shrinks?
New customers can more than offset losses from existing customers. Net revenue retention, or NRR, includes expansion and contraction within the opening cohort. Gross revenue retention, or GRR, excludes expansion. Logo retention counts customers rather than dollars, so it answers a third question.
The following dataset is invented for illustration, with amounts in $000. Customers A through D form the opening cohort; E is new. There is no FX, acquisition, reactivation, grace period or expansion followed by contraction within the measurement period.
| Customer | Opening ARR | Expansion | Contraction | Churn | Closing ARR | Closing ARR for GRR |
|---|---|---|---|---|---|---|
| A | 100 | 40 | 0 | 0 | 140 | 100 |
| B | 100 | 0 | 20 | 0 | 80 | 80 |
| C | 100 | 0 | 0 | 100 | 0 | 0 |
| D | 100 | 20 | 0 | 0 | 120 | 100 |
| Opening cohort | 400 | 60 | 20 | 100 | 340 | 280 |
| E: new customer | 0 | 0 | 0 | 0 | 100 | Excluded |
| Total company | 400 | 60 | 20 | 100 | 440 |
The cohort mechanics follow the IPO definitions below. Excluding expansion from GRR follows Workday's September 4, 2024 SEC response; none of these numbers is an issuer result.
- Point-in-time NRR: $340 ÷ $400 = 85%.
- GRR: ($400 − $20 − $100) ÷ $400 = 70%.
- Logo retention: three retained customers ÷ four opening customers = 75%.
- Total ARR growth: ($440 − $400) ÷ $400 = 10%.
The company bridge is $400 opening ARR + $60 expansion − $20 contraction − $100 churn + $100 new-customer ARR = $440. New business creates company growth while the original cohort loses 15% of its ARR.
This establishes cohort shrinkage despite total ARR growth. Demonstrating a decline in reported NRR requires another period's comparable retention calculation.
Why aren't two SaaS companies' NRR figures directly comparable?
Compare definitions before percentages. Two issuers can use different customer populations, contract values, observation dates and averaging methods. A headline retention rate may summarize several cohort observations rather than describe the latest observation.
These are historical IPO disclosures, not current performance comparisons:
| Methodology | Rubrik, April 1, 2024 S-1 | Klaviyo, August 25, 2023 S-1 |
|---|---|---|
| ARR basis | Annualized active subscription contracts, assuming expiring contracts renew on existing terms | Measurement-date contractual MRR multiplied by twelve |
| Retention observation | Same subscription customers, comparing corresponding quarter ends one year apart | Same customer cohort, comparing dates twelve months apart |
| Headline aggregation | Average of current quarterly ratio and prior three quarterly ratios | Twelve monthly point-in-time ratios, weighted by total ARR at each month end |
Sources: Rubrik's Key Business Metrics and Klaviyo's Select Defined Terms. Klaviyo explicitly distinguished contractual MRR from historical GAAP revenue; its NRR includes price increases and expansion while excluding new customers.
The averaging step changes interpretation. In this separate invented illustration, four successive quarterly cohort ratios produce:
| Observation | Q1 | Q2 | Q3 | Q4 | Four-quarter arithmetic average |
|---|---|---|---|---|---|
| Cohort ratio | 120% | 110% | 100% | 90% | 105% |
The latest ratio is 90%, although the average exceeds 100%. This isn't Rubrik's disclosed quarterly trend. In a weighted average, larger-ARR observations exert more influence.
How should the disclosure explain changes in ARR and retention?
Separate changes in customer economics from changes in metric scope. Price increases, usage, upsells, contraction and churn affect an existing cohort. Product migrations, acquisitions and methodology changes can alter the reported population or measurement basis.
Rubrik reported Subscription ARR of $532.929 million at January 31, 2023 and $784.029 million at January 31, 2024. Its average subscription dollar-based net retention rate was 150% and 133%, respectively. The S-1 attributed approximately four percentage points of fiscal 2024 Subscription ARR growth to maintenance customers transitioning to subscription editions.
The explanation distinguishes new relationships from existing customers entering the subscription measure. Calling the entire increase “new demand” would erase that distinction.
When methodology changes, assess whether to explain the differences, reasons and effects, and whether prior periods need recasting. Those are materiality-based considerations in the SEC's KPI guidance. Date the changed policy and keep a bridge that distinguishes operating movement from calculation movement.
What has the SEC actually asked about retention?
The staff has asked for business meaning as well as formulas. In its September 7, 2023 letter to Klaviyo, comment 2, the SEC requested an explanation of why NRR was useful for understanding usage growth across the periods presented. A mathematically complete definition didn't answer that interpretive question by itself.
Workday's September 4, 2024 response reproduced staff questions about gross and net retention definitions, calculations, management's KPI assessment, actual rates and significant fluctuations. The proposed definitions excluded add-ons and net expansion from GRR and included them in NRR. Actual-rate data in the public response was redacted.
Workday's correspondence concerned a post-IPO Form 10-K. It doesn't establish a universal GRR requirement. Both letters illustrate questions reporting teams should be able to answer.
What should the CFO approve before filing?
Approve the definition and the numerical bridge together. A clean formula doesn't explain an unexplained jump, and a persuasive growth narrative doesn't repair an unstable denominator.
The metric-specific review package should contain:
- The ARR inclusion policy, measurement date and renewal assumptions.
- The fixed opening cohort, customer identity rules and zero-ending customers.
- The NRR and GRR calculations, including monthly or quarterly observations and weights.
- The bridge for new customers, expansion, contraction, churn and population changes.
- The explanation of material movements, investor usefulness and management use.
- The classification assessment and agreed scope of any auditor procedures.
For peer research, capture the exact definition, cohort, averaging method and filing date beside each source link. Use the evidence to draft; a peer's choice doesn't settle your policy.
Frequently asked questions
Does ARR mean the next twelve months of GAAP revenue?
No. Annualizing current contractual activity doesn't determine revenue recognition or guarantee renewals. State the issuer-specific calculation and avoid presenting ARR as a forecast of recognized revenue.
Should NRR include price increases or reactivated customers?
It depends on the disclosed policy. Klaviyo's historical S-1 included both. Describe their treatment so readers can distinguish pricing, usage and customer recovery in the result.
Can GRR exceed 100% in the worked model?
No. This model removes contraction and churn from opening ARR and excludes expansion. Assess other methodologies before applying this formula.
Does the financial statement audit cover ARR and NRR?
Their presence in the same prospectus doesn't automatically make them audited. PCAOB AS 6101, paragraphs .11 and .12, distinguishes audited information from the limited procedures used in comfort letters. Confirm the actual engagement scope before describing any assurance.
Which NRR should we use for a peer benchmark?
Use a comparison with matched population, observation window, ARR basis and aggregation method. If those don't match, explain the differences alongside the percentages. A number without its denominator is an unfinished disclosure.
