SaaS field notes · Revenue metrics
CARR vs. ARR: make the numbers make sense.
Why SaaS is the most fun you can have with a balance sheet—once the acronyms stop arguing.
See the worked exampleStart with the distinction
A recurring relationship deserves better than a confusing spreadsheet.
SaaS—software as a service—makes a sale feel satisfyingly complete right up until onboarding, adoption, support and renewal arrive with their own agendas. That is part of what makes the business interesting. A signed customer creates an opportunity to earn a lasting relationship, plus a collection of promises the company now has to keep.
I like the strategic puzzle. I like it considerably less when three departments bring three versions of “recurring revenue” to the same meeting and spend the first half hour negotiating the alphabet.
ARR and CARR are operating metrics. Neither is a bank balance, and ARR is not GAAP recognized revenue. ARR commonly describes an annualized recurring subscription base. CARR commonly extends that view to documented commitments and changes that have not yet taken effect. The definitions, timing rules and exclusions need to travel with the numbers.
In its SEC-filed correspondence, Alteryx distinguishes its annualized contract measure from GAAP revenue. Annualizing a short subscription can even produce an amount larger than the contract’s total value. That is a property of a run rate, not evidence that the customer owes an extra year of payments.
Define before comparing
What ARR and CARR mean in this article.
The following is an explicit teaching convention for a fixed-fee subscription business.
ARR: the active annualized rate
I use annual recurring revenue (ARR) to mean the annualized recurring subscription rate active on the measurement date. A flat $2,000 monthly subscription contributes $24,000. A one-time implementation fee contributes zero to this recurring metric.
CARR: the rate after documented changes
I use committed annual recurring revenue (CARR) to mean active ARR adjusted for signed future starts, incremental signed expansions, known downgrades and confirmed nonrenewals taking effect within the stated horizon. Unsigned pipeline is excluded. The example uses the next 12 months and no other assumed changes.
The definition is part of the number
CARR also appears as “contracted ARR,” with varying policies. Specify the cutoff date, future horizon, eligible commitments, cancellation treatment, currency policy and pricing rules. Two identically labeled charts can describe different populations.
A real disclosure illustrates the difference. UiPath calls ARR “Annualized Renewal Run-rate” and bases its calculation on annualized invoiced amounts for eligible subscription and support items. Its disclosure includes specific exclusions and says the metric should be viewed independently of revenue and deferred revenue. That is UiPath’s definition, not the policy automatically inherited by every SaaS company. Read UiPath’s key performance metric definition.
A compact glossary
Seven numbers that should not share one label.
I want each number to answer a specific question. This glossary uses the same simplified subscription model as the example below; a company’s reporting policy still controls its actual calculations.
| Measure | Useful meaning here | Keep this distinction |
|---|---|---|
| ARR | Annualized active recurring subscription rate at a date. | A snapshot, not revenue earned over the preceding year. |
| CARR | That rate after qualifying documented future changes. | A policy-defined commitment view, not guaranteed cash. |
| Recurring ACV | Annual contract value for the recurring portion of an individual agreement. | For flat pricing, recurring contract value ÷ term in years. Some ACV definitions treat fees differently. |
| TCV | Total contract value over the full agreed term. | Specify whether it includes one-time fees. A two-year total is not a one-year run rate. |
| Bookings | Qualifying contracts signed during a reporting period. | State whether the report uses total value, annual value or another basis, and how renewals are counted. |
| Recognized revenue | Revenue recorded as performance obligations are satisfied under the applicable accounting framework. | Recognition timing follows the accounting analysis, not a sales-stage change. |
| Invoices and cash receipts | Amounts billed and amounts collected, respectively. | An issued invoice can remain unpaid. A prepayment can arrive before service is delivered. |
The accounting distinction is grounded in the transfer of promised goods or services. The IFRS Foundation’s revenue-recognition overview explains the performance-obligation approach and recognition at a point in time or over time. The contract facts determine the treatment; “it is software” does not finish that analysis.
The calculation, with boundaries
A CARR formula that shows its working.
Every adjustment is annualized, documented as of the measurement date and effective within the declared horizon. Stripe’s CARR explanation describes this general approach of adding signed starts and expansions while accounting for known losses.
For the example below, the active base stays in the calculation unless a documented change says otherwise. That convention does not assert that every baseline customer is legally committed for the next 12 months. Contract coverage is a separate question. If your business uses a stricter signed-coverage measure, name it and explain the difference.
A same-price renewal adds zero incremental recurring value. Moving a signed customer from pending to active also creates no second CARR addition. Spreadsheets are exceptionally obedient; they will double-count a contract without the slightest embarrassment.
Worked example · Hypothetical
Five customers. One measurement date. No mystery dollars.
Assume the measurement date is September 30, 2026, with a commitment-change horizon ending September 30, 2027. All amounts are USD, prices are flat within each stated phase, and all commitments below are documented by the measurement date. There are no usage charges, taxes, foreign-exchange effects or other customers.
Atlas, Beacon, Cedar and Delta started their current 12-month subscriptions on January 1, 2026. Echo has signed a future 24-month subscription. Existing customers remain active through the dates shown; future changes have not yet entered September ARR.
| Customer | Current rate / documented event | Active ARR | CARR adjustment | CARR contribution |
|---|---|---|---|---|
| Atlas | $2,000/month. Signed same-price 12-month renewal starts January 1, 2027. | $24,000 | $0 | $24,000 |
| Beacon | $3,000/month. Signed expansion adds $1,000/month starting November 1, 2026, continuing through its signed 2027 renewal. | $36,000 | +$12,000 | $48,000 |
| Cedar | $1,500/month. Signed 12-month renewal drops to $1,000/month on January 1, 2027. | $18,000 | −$6,000 | $12,000 |
| Delta | $2,500/month through December 31, 2026. Confirmed nonrenewal takes effect January 1, 2027. | $30,000 | −$30,000 | $0 |
| Echo | Signed September 20. $3,000/month for 24 months, starting December 1, 2026. | $0 | +$36,000 | $36,000 |
| Total | $108,000 | +$12,000 net | $120,000 | |
Active ARR on September 30.
CARR after the documented changes.
Echo’s recurring TCV over 24 months. Its annualized contribution is $36,000.
The reconciliation: $108,000 + $36,000 + $12,000 − $6,000 − $30,000 = $120,000. Atlas’s $24,000 renewal is already represented in the base. Echo’s two-year total is divided by two. Neither is added twice.
Follow the effective dates
A commitment becomes active. The total does not get another bonus.
Assuming every event occurs as scheduled and no other changes happen, October’s active ARR stays at $108,000. Beacon’s expansion brings it to $120,000 on November 1. Echo’s start brings it to $156,000 on December 1. Cedar’s downgrade and Delta’s departure bring it back to $120,000 on January 1.
Under this example’s unchanged assumptions, CARR remains $120,000 through those transitions. The pending additions move into active ARR; the pending losses eventually come out of active ARR. The reconciliation changes location, not economic substance.
December makes the point especially clearly: active ARR is $156,000, but the commitment view is $120,000 because $36,000 of annualized losses are already known. CARR can be lower than ARR under a policy that incorporates future losses.
The September $12,000 gap does not mean the next month generates $12,000 more revenue. It is a net annualized difference across changes with different effective dates. I would keep the underlying schedule beside the headline metric.
One contract, three clocks
Echo can have CARR before it has revenue or cash.
Continue the hypothetical Echo agreement: $72,000 of recurring fees over 24 months, starting December 1. Assume it contains one hosted-service performance obligation delivered evenly over that period, with no additional obligations, adjustments or financing effects. Under those assumptions, the illustrative revenue schedule is $3,000 per month.
On September 30, Echo contributes $36,000 to CARR and zero to active ARR. Service has not begun, so it has generated no revenue in this example. The agreement can appear in September bookings as $72,000 of recurring TCV if that is the company’s stated bookings basis. A bookings report using recurring ACV would show $36,000 instead.
Now assume the first annual $36,000 invoice is issued November 15 and paid November 25. Cash arrives in November. December service produces $3,000 of recognized revenue. At December 31, $33,000 of that advance relates to the remaining service in the first prepaid year under these simplified assumptions.
The payment changes neither Echo’s $36,000 annualized rate nor its $72,000 recurring contract total. It changes the collection position. If payment is late, the cash forecast changes even if the subscription metric policy leaves ARR unchanged.
Annualized does not mean collected over the next year. Using the whole five-customer schedule above, assume monthly revenue follows each active service rate. October 2026 through September 2027 would total $122,000: $9,000 + $10,000 + $13,000 + nine months at $10,000. September CARR is $120,000. Different timing, different question.
The balance sheet still matters enormously. The joke in the original title is about the strategic fun of SaaS; it is not a claim that ARR itself is an accounting balance-sheet line. For an operating plan, I would build a dated revenue and cash forecast from the contract schedule, payment terms and explicit risk assumptions.
Where the simple model needs judgment
Real contracts are less polite than the example.
Ramps, discounts and short terms
A contract priced at $24,000 in year one and $48,000 in year two has a $36,000 average recurring ACV. That average does not describe either year’s actual rate. Decide whether ARR follows the current phase and how CARR treats a signed ramp within the horizon. Keep the schedule available; silently swapping between average and exit rates makes growth look better or worse without a customer changing anything.
Likewise, annualizing a three-month $6,000 subscription produces a $24,000 rate under the simple convention, but only $6,000 is committed by that contract. A short term, a temporary discount and a noncancelable annual commitment are different commercial facts.
Usage, minimums and optional seats
Separate a contractual minimum from variable consumption and optional purchases. For this article’s model, uncommitted usage and optional seats stay outside CARR. A business that reports usage-based ARR needs a stated measurement window and annualization method. A strong month of consumption is evidence about demand; it is not automatically a signed annual commitment.
Conditions, cancellation and collection
Inspect implementation dependencies, acceptance conditions, termination rights and contract amendments before classifying a pending start. Keep an exception record when a scheduled launch slips. A signature can improve visibility without eliminating delivery or collection risk.
I would also separate metric-policy decisions from accounting decisions. An operational flag for a disputed invoice is useful, but it does not by itself settle revenue recognition, collectibility or the treatment of a contract modification.
Use the number to change the work
The interesting part starts after the calculation.
A bigger commitment base can be encouraging. It can also reveal an onboarding queue the company cannot deliver. I want the explanation behind the change before I treat the headline as permission to hire, spend or celebrate.
| Signal | What to investigate | A useful next action |
|---|---|---|
| CARR rises; active ARR barely moves | Future start dates, time awaiting activation, delayed migrations and implementation capacity. | Assign owners and dates to pending launches before increasing acquisition spend. |
| New contracts mask large losses | The separate new, expansion, contraction and churn components; losses by customer cohort and reason. | Investigate product fit and renewal risks rather than relying on a positive net total. |
| ARR rises; cash remains tight | Payment terms, overdue invoices, annual versus monthly collection and delivery spending. | Update the collection and cash schedule before making a fixed-cost commitment. |
| Expansion comes from a few accounts | Customer concentration, adoption breadth and the next renewal dates. | Review account exposure and capacity while checking whether growth is repeatable elsewhere. |
| Signed value grows; delivery becomes harder | Custom work, support burden, infrastructure cost and the promises made during sales. | Revisit qualification, packaging and pricing with the people responsible for delivery. |
These are diagnostic questions, not automatic verdicts. A large pending start can be completely healthy when the implementation plan supports it. A smaller, well-served customer can be more worthwhile than a larger contract that requires permanent improvisation.
Acquisition meets retention
Marketing has to care what happens after the signature.
I would connect the contract data to the customer’s reason for buying. Which customer types activate successfully? Which sales promises create difficult implementations? Which users reach the outcome they purchased, and which accounts renew only after a discount negotiation?
New business can hide deterioration in the installed base. For retention analysis, follow the same starting customer cohort through the period and separate expansion, contraction and churn. Do not include newly acquired customers in that cohort’s net revenue retention calculation. Keep a view of gross losses too, so expansion in one account does not conceal another account’s departure.
Acquisition economics also need delivery economics. Subscription fees alone do not reveal hosting expense, AI usage cost, support effort or the cost of winning the customer. An ARR chart cannot tell you whether the marginal contract is attractive without those inputs.
My SaaS marketing consulting work connects positioning, acquisition, activation and retention. When the underlying question is which buyer to pursue and how to package the offer, the deeper starting point is go-to-market strategy. Better definitions make those conversations more useful; they do not replace them.
A practical reporting routine
Build a metric people can explain without opening six tabs.
Agree on the decision and the dictionary.
Start with finance, sales, customer success, operations and the relevant decision makers. Choose the question the report must answer. Document the measurement date, rate convention, horizon, exclusions and ownership before choosing chart colors.
Give every change a source and an effective date.
Keep customer and contract identifiers, signed dates, start and end dates, current and future recurring amounts, billing terms and links to the governing agreement. Track amendments against the original contract so an expansion does not create a duplicate customer.
Reconcile the movement and inspect exceptions.
Explain opening active ARR plus activated new business and expansion minus effective contraction and churn. Separately reconcile pending commitments. Review slips, disputed classifications and missing dates with named owners. Keep prior snapshots so a policy change cannot quietly rewrite history.
Put the explanation beside the result.
Show the as-of date, definition, comparison period, major drivers and unresolved exceptions. Assign the next decision. My KPI and dashboard design approach begins with that operating purpose, then builds the view around it.
I would rather see a modest report that reconciles to actual contracts than a beautiful dashboard whose most important setting is optimism.
Questions worth clearing up
ARR and CARR FAQs.
What is the difference between CARR and ARR?
In this article, ARR annualizes the recurring subscription rate active on the measurement date. CARR adjusts that baseline for signed future starts and expansions, plus known contraction and churn within a stated horizon. Neither figure is recognized revenue or cash. Check the reporting company’s definitions before comparing numbers.
Is ARR GAAP revenue?
No. ARR is an operating metric that annualizes a subscription base under a defined policy. Revenue recognized under U.S. GAAP follows the applicable accounting requirements. An annualized subscription amount, revenue earned during a period and cash collected can all differ.
Does a renewal increase CARR?
A same-price renewal preserves an existing recurring amount; it does not add that amount again. In the example here, only an incremental signed price or scope increase adds to CARR. A decrease reduces it. A renewal can improve contract coverage even when the annualized amount stays unchanged.
Can CARR be lower than ARR?
Yes, under the convention used here. If known future churn and contraction exceed signed new business and expansion, CARR falls below active ARR. Definitions that exclude future losses will behave differently, which is why the metric policy needs to accompany the number.
Do multi-year contracts contribute their entire value to ARR?
No. In the flat-price example here, a $72,000 recurring subscription over 24 months contributes $36,000 of annualized recurring value. Its recurring total contract value is $72,000. One-time fees are excluded from these recurring measures. Ramped pricing requires an explicit policy rather than an automatic average.
Does CARR include unsigned sales pipeline?
The convention here excludes unsigned opportunities, verbal enthusiasm and assumed upsells. A separate forecast can model pipeline with explicit probabilities and timing assumptions. A signed commitment also needs review for effective dates, cancellation rights and unresolved conditions.
What should I track alongside ARR and CARR?
Track the dates and age of signed contracts awaiting activation, customer adoption, renewal exposure, recurring expansion and losses, collection timing and delivery cost. Use consistent customer cohorts for retention measures. These supporting views help explain whether the recurring base is becoming healthier and whether the business can deliver what it sold.
Make the next decision clearer
Your recurring metrics should help you run the business.
If your SaaS team has plenty of reporting but keeps arguing about what growth actually means, I can help connect the commercial strategy, customer journey and operating measures.
I would start with the decision you need to make, the people responsible for it and a simple measurable objective. From there, the work is to understand the current process, identify constraints and build a practical roadmap that the team can carry out.
Tell me where the picture gets blurry: acquisition quality, slow activation, renewal exposure, pricing or a forecast nobody quite trusts.
Talk with me about your SaaS business