How do you separate NRR, GRR, and logo retention when board auditors ask which is 'real' in 2027?

The 2 a.m. email before the audit committee meeting

Picture the version of this that actually happens. It is a Tuesday, the audit committee meets Thursday, and the chair sends a three-line email at 10 p.m.: *"Deck says NRR 118%. Last quarter's CS report said retention was 84%. Which one is real?"* Nobody is accusing anybody of fraud. The chair has two numbers that both claim to describe the same customer base and no way to tell whether they are consistent, contradictory, or measuring entirely different things.

Here is what is usually going on underneath that email. Finance built GRR in a model tied to the billing system, using legal entities as the customer unit and locking FX at the start of the period. RevOps built NRR in the warehouse, pulling from a CRM view that treats every billing account as a separate customer, converting foreign currency at whatever the spot rate was on the day the query ran. Customer Success built logo retention in a spreadsheet that counts active workspaces and quietly drops accounts that went dormant rather than formally cancelled. Three teams, three cohorts, three currency policies, three definitions of "customer." Every one of those numbers is defensible in isolation. Together they are incoherent, and the chair can feel the incoherence without being able to name it.

The instinct in that moment is to litigate the formulas. That instinct is wrong, and it is the single most common way a RevOps leader loses an audit committee. The formulas are trivial — three lines of arithmetic a first-year analyst can reproduce. What the chair is really asking is a governance question dressed up as a definitional one: *if I hand your raw data and your written methodology to a second analyst, do they get your number?* If the answer is no, the metric is not real in any sense an audit committee cares about, no matter how correct the arithmetic was.

This matters more than it used to. Retention numbers are load-bearing in a way they were not fifteen years ago. They anchor venture valuations, they appear in S-1 risk factors and MD&A, they drive covenant conversations with lenders, and in a lot of companies they feed directly into executive compensation. Meanwhile they remain entirely non-GAAP: no FASB standard defines NRR, no PCAOB procedure certifies it, no SEC rule prescribes its formula. That gap between how much weight the number carries and how little discipline is imposed on its construction from outside is exactly the crack auditors are probing. The absence of an external standard does not mean anything goes — it means you have to supply the standard yourself, in writing, and then live inside it.

So the answer to "which is real" is not to pick one. It is to demonstrate that all three come from the same place. That single move — one bridge, one cohort, one policy set — converts an argument about definitions into an exhibit the auditor can verify. Everything below is how to build it, plus the adjacent machinery (segmentation, ASC 606 handling, RPO corroboration, board presentation order) that makes it hold up under questioning.

What each metric actually measures, and where each one lies to you

Before the bridge, get precise about what the three numbers do. They are not competing estimates of one underlying truth. They answer genuinely different questions, and each has a characteristic failure mode.

Gross revenue retention takes a starting cohort of recurring revenue and subtracts only the losses: contraction (customers who downgraded but stayed) and churn (customers who left entirely). Expansion is not allowed in. GRR is therefore capped at 100% by construction — you cannot keep more than you started with if upside is excluded. That cap is precisely why auditors trust it most. There is no offsetting term to hide a problem behind. GRR is the structural health of the base, independent of how good your upsell motion is. It is also the leading indicator: when GRR falls, NRR typically follows within two to three quarters, once expansion conditions normalize and stop covering the gap.

Net revenue retention takes the same cohort and adds expansion — upsells, cross-sells, seat growth, consumption growth, contractual price escalators. It can and routinely does exceed 100%. This is the compounding number investors care about, because a business with strong NRR grows even if new-logo acquisition stops. It is also the most manipulable of the three, and every experienced auditor knows the four levers. Cohort selection bias: choose a starting population weighted toward your best accounts and NRR inflates without any behavior changing. Price-increase masking: a 7% list escalator lands in the expansion line even though no customer chose to buy more of anything. Currency tailwinds: a weakening reporting currency inflates the foreign-denominated slice of the cohort. Survivorship: drop churned logos out of the denominator and the surviving base looks devoted.

Logo retention counts customers, not dollars. It answers the question neither dollar metric can — are you keeping the *number* of relationships, or just the revenue concentrated in a handful of them? A company can post 130% NRR while losing a fifth of its logos, provided the survivors expanded enough to cover. That is a concentration-driven business wearing a healthy NRR mask, and it is fragile in a way the headline hides. Logo retention strips the mask off. Its own weakness is the mirror image: it treats a $12k account and a $2.2M account as one unit each, so in a business with wide account-size dispersion, logo retention alone can look alarming when the dollar reality is fine.

There is a fourth number boards sometimes request and you should decline: net logo retention, which adds new logos won during the period back into the count. It conflates retention with acquisition and produces something uninterpretable — a company can post 110% net logo retention while churning a third of its existing base, purely on new-logo strength. Retention metrics answer one question, *did the base hold*, and the moment acquisition leaks in they stop answering it. If the board wants an acquisition figure, give them gross new logos on a separate line. The same objection kills any "net new ARR retention" that folds new-logo ARR into the cohort.

Finally, the algebraic guardrails an auditor will check without announcing that they are checking. NRR is always greater than or equal to GRR, because NRR is GRR plus the expansion ratio and expansion cannot be negative — report NRR below GRR and you have a calculation error, full stop. GRR never exceeds 100%; if it does, expansion has leaked into the gross calculation and the bridge needs rebuilding. The difference between the two is itself a metric worth tracking: NRR 110% against GRR 90% is a 20-point expansion gap, and a *widening* gap against a flat or falling GRR means expansion is working harder every quarter to mask a stable leak. And the relationship between GRR and logo retention is a concentration gauge: GRR 92% with logo retention 80% means you are shedding small accounts and holding the large ones; GRR 80% with logo retention 95% means you are keeping customers but they are shrinking, which is a pricing or downgrade problem rather than a churn problem.

One bridge, six frozen inputs, three metrics falling out

The mechanism that makes the three numbers coherent is a single retention bridge: a waterfall that starts at frozen cohort ARR and walks line by line to ending ARR. GRR, NRR, and logo retention are then *read off that bridge* rather than computed independently. They cannot disagree, because they share every input.

A bridge for one anniversary cohort:

Read the three metrics off it. GRR = (48,200 − 3,376 − 5,784) / 48,200 = 81.0%. NRR = (48,200 + 9,640 + 1,880 − 3,376 − 5,784) / 48,200 = 104.9%. Logo retention = (612 − 74) / 612 = 87.9%. One table, one cohort, one currency policy, three provably consistent numbers. Note also that escalators sit on their own line rather than buried inside expansion — that separation is what lets you answer "how much of your NRR is customers choosing to buy more versus your contract raising the price" without rebuilding anything.

The bridge is only defensible if six inputs are locked and documented:

• Cohort population — exactly which customers, keyed on a stable immutable identifier. Legal entity ID, not billing account, not email domain.
• Snapshot date and procedure — starting ARR captured at a timestamp and written to immutable storage, never re-derived from a live system that has since changed.
• Currency policy — either lock FX at cohort start (constant currency) or use period-average rates, applied identically to both ends of the bridge.
• ARR definition — what counts as recurring, including a normalization rule for usage-based revenue.
• Expansion/contraction boundary — a customer who drops one product and buys another: net expansion, or churn plus new? Decide once, in writing.
• Contract-modification treatment — how mid-period upsells, co-terms, and renewals-with-changes flow through the bridge.

The frozen-cohort snapshot table is the single most powerful artifact in this whole apparatus, and it should carry more than ARR. Each row: stable customer key, legal entity name, starting ARR in functional currency, the FX rate applied and its date, contract end/renewal date, segment tags (size band, geography, vintage, product), and the source-system record IDs the ARR was derived from. The extra columns exist for verification speed. When an auditor samples customer X, every fact about that customer's cohort membership should live in one row rather than scattered across five systems.

Work a five-customer cohort through a year to see the mechanics end to end. Starting snapshot, frozen 2024-01-01: Northwind Logistics $220k enterprise, Pinecrest Health $95k mid-market, Vela Robotics $60k mid-market, Tidewater SMB Co $18k, Glasshouse Media $12k. Starting cohort ARR $405k, five logos. Over the year Northwind expands to $290k on seat growth, Pinecrest holds flat, Vela downgrades to $40k, Tidewater cancels outright, Glasshouse renews flat. Expansion $70k, contraction $20k, churn $18k, one churned logo. GRR = (405 − 20 − 18) / 405 = 90.6%. NRR = (405 + 70 − 20 − 18) / 405 = 107.9%. Logo retention = 4/5 = 80.0%. The three tell a story together that none tells alone: healthy structural base, a real expansion engine, and one in five customers gone. At five customers that single churn is statistical noise. At three hundred, an 80% logo retention alongside 108% NRR is a genuine concentration warning. Same arithmetic, different read, and knowing which read applies is the judgment an audit committee is paying for.

One more thing the bridge buys you: the cardinal rule becomes enforceable. New logos acquired *during* the window never appear in the numerator or denominator. Retention measures whether you kept what you had. This gets violated constantly, almost always because an analyst pulled "all customers active in 2024" instead of "customers active on 2024-01-01." The first population includes new logos; the second does not. An auditor catches it in one sampling pass by checking contract start dates against the cohort date, so build the snapshot such that it is structurally impossible.

The numbers that anchor the conversation

Boards and auditors reason by comparison, so know the reference ranges cold — and know which segment you are actually in, because a blended benchmark applied to the wrong business is worse than no benchmark.

Cite the benchmark source when you present these, because methodologies genuinely differ across the standard surveys — Bessemer's State of the Cloud, the KeyBanc SaaS survey, ICONIQ Growth's reports, operator-community data. A range from a survey of $50M+ ARR enterprise companies is not the range for your $8M PLG business, and an auditor who knows the source will ask.

Public comparables anchor even harder, and it is worth noticing what the large disclosers actually choose to publish. Snowflake reports a consumption-style net revenue retention that ran extremely high around IPO and has compressed substantially as the base matured — a normal and expected pattern, since the denominator grows faster than any expansion engine can compound against. Datadog has sustained high NRR on multi-product expansion. MongoDB's Atlas business shows usage-driven expansion. HubSpot, SMB-heavy, sits near the 100% line on NRR with GRR meaningfully lower. ServiceNow discloses a *renewal rate* rather than a classic NRR. Salesforce has historically discussed *attrition* rather than either. That divergence among the largest and most heavily audited companies in the category is the real lesson: even they choose their retention disclosure carefully, and what auditors require is not a specific metric but definitional consistency and disclosed methodology. Pull the current figures from the latest 10-K or investor supplement before you put any of them in a deck — these move, and a stale comparable is an unforced credibility hit.

Now the segmented exhibit, which is where blended numbers go to die. A single company-wide NRR hides everything that matters:

The blended GRR is a *weighted* average — each segment weighted by its share of starting ARR — not a simple mean. The simple mean here would be 86.5%, materially different, and an auditor who recomputes will flag it. Weight dollar metrics by starting ARR and logo retention by starting logo count. The story in this exhibit is the SMB row: 10% of the base, dragging blended GRR by roughly a point, and the segment most likely to deteriorate first in a downturn. A blended-only deck buries that completely.

Segment by vintage too — the quarter or year a customer first started paying. Vintage curves answer whether retention is improving *structurally* rather than cyclically. A typical healthy pattern has each cohort getting stickier with age as weak accounts wash out and survivors expand: a 2021 cohort at 84% year-one GRR climbing to 88% in year two and 91% in year three. What you are looking for is whether newer vintages beat older ones at the same age. If the 2023 cohort's year-one GRR is 88% against the 2021 cohort's 84%, onboarding and product-market fit are genuinely improving. If newer vintages retain *worse*, you are acquiring lower-quality customers as you scale — the classic growth-stage trap, and one that headline NRR will hide for a year or more.

Trend matters as much as level, and there is one four-quarter exhibit that makes masking visible:

Read Q4 carefully. The headline improved every single quarter. The foundation eroded every single quarter. That is the pattern the GRR-first discipline exists to surface.

And the arithmetic of the dangerous profile, because it is worth making concrete. A company at 78% GRR and 125% NRR on $100M of cohort ARR loses $22M to churn and contraction and must generate $47M of gross expansion just to hit the headline. Soften the expansion environment — a macro downturn, a competitor, budget freezes at customers — so that expansion falls from 47% to 30%, and NRR collapses from 125% to 108% in one year *with no change in churn whatsoever*. The 78% was the real number the entire time; the 125% was borrowing against favorable expansion conditions. That is the treadmill, and it is why the presentation order is not cosmetic.

Where ASC 606 bends the metric, and what you trade away

Retention metrics are non-GAAP, but they live next door to numbers that are audited, and the seam between them is where most disputes actually originate. ARR is a forward-looking annualized run-rate snapshot. ASC 606 revenue is recognized as performance obligations are satisfied. The two will not equal each other in any period, and that is fine — as long as you can explain the reconciliation. When you cannot, every retention number in the deck becomes suspect by association.

Two ASC 606 areas cause nearly all the trouble.

Variable consideration. Usage fees, tiered pricing, rebates, and performance bonuses must be estimated and constrained. For a consumption business this collides directly with ARR-based retention: if a customer's "ARR" is an estimate of annualized usage, then part of their expansion in your bridge is a *re-estimate of variable consideration* rather than a customer decision to consume more. Auditors will want those separated. The cleanest practice, and the one the usage-heavy public companies converge on, is to define a consumption-based retention metric from trailing *actual* consumption rather than forward estimates, and to footnote exactly how starting ARR was derived under the usage model.

Contract modifications. Any change to scope or price is a modification, which covers mid-period upsells, co-terminations, early renewals with changed terms, and add-on products — in other words, most of what generates your expansion line. ASC 606 allows three treatments, and each lands differently in the bridge:

A cumulative catch-up is the nasty one: it produces a revenue figure that looks like a retroactive change to the cohort, and if the bridge is not aware of it, GRR and NRR will not tie to revenue and nobody will immediately know why.

There is a quieter adjacent trap in ASC 340-40, the standard governing capitalized costs of obtaining a contract — mostly sales commissions. It is a cost standard, but it intersects retention through the amortization period, which is tied to expected customer relationship length, which is a direct function of your churn rate. A 75% GRR implies roughly a four-year average customer life. If your commissions amortize over seven, an auditor will challenge one of the two, and they should. Your retention metrics and your commission amortization schedule must tell the same story about how long customers stay.

The capstone artifact is the reconciliation from cohort ARR to GAAP revenue and to remaining performance obligations:

RPO deserves its own attention, because it is the one retention-adjacent figure that actually *is* audited and disclosed. It is contracted future revenue not yet delivered, typically split into current RPO (within twelve months) and long-term. RPO and NRR should corroborate each other, and when they diverge the divergence is informative: NRR at 120% alongside flat or shrinking RPO usually means large multi-year contracts are rolling off without renewal — a churn cliff NRR has not caught yet. RPO growing much faster than NRR usually means contract durations are lengthening, which is good for predictability but means current NRR understates committed expansion. Putting current RPO growth next to NRR on the board exhibit is how you answer "is this NRR durable" with evidence rather than assertion.

Know the distinction auditors use as a competence test: deferred revenue is not RPO. Deferred revenue is only the billed-but-unrecognized portion. RPO is deferred revenue *plus* contracted-but-unbilled backlog. A customer on a three-year contract billed annually has one year sitting in deferred revenue and two more in RPO, entirely off the balance sheet. Cite deferred revenue when asked about contracted backlog and you have understated the forward book and signaled you do not understand your own disclosure.

Now the trade-offs, because none of this is free.

The cohort-window trade-off is the sharpest. Anniversary cohorts — everyone active on January 1, measured where that exact group stands the following January 1 — give you a frozen, re-verifiable population, which is precisely what an audit committee needs and what belongs in an S-1. Their cost is latency: you learn about a deterioration up to a year after it began. TTM rolling re-cohorts every month, catches trend faster, and is nearly impossible to audit because the population changes every period. The resolution is not to pick one globally. Run anniversary cohorts for the board and the audit committee, run TTM rolling for the CS team's weekly operating rhythm, and never present one in the other's venue.

Currency is a similar fork. Constant currency — locking FX at cohort start — isolates actual customer behavior, and it matters more than people expect: a 10% swing in a major currency pair can move a Europe-heavy cohort's NRR by four to six points with zero change in what any customer did. Reported/as-converted is what ties to GAAP. Present constant currency as the primary number and reported as the secondary disclosure, footnote the conversion method, and apply the same policy to both ends of the bridge. Period-average rates are an acceptable middle path if applied consistently; spot rates at each end maximize noise and should be avoided.

The customer-unit choice is the definitional swamp that sinks more retention metrics than any formula error. Consider: a parent with five subsidiaries on five separate contracts — one customer or five? A customer consolidating three legacy contracts into one master agreement — did two logos churn? A reseller buying for fifty end users — one logo or fifty? A free-tier account converting mid-cohort — when did it become a customer? There is no universally right answer. There is only a documented, consistently applied one. Most enterprise SaaS companies land on the contracting legal entity. Whatever you choose, apply it identically across all three metrics and every period, and treat the definition as a controlled document with version history.

The pitfalls that get a number called "not real"

The failures cluster tightly, and an experienced auditor probes them in roughly this order.

Re-pulling the cohort from a live system. The starting base gets regenerated from a source that has since been updated — customers merged, accounts renamed, contracts amended retroactively — and the denominator silently shifts between periods. You end up comparing two different populations and calling it a trend. The fix is structural, not procedural: write the snapshot once to an immutable, append-only store, partition it by snapshot date, and lock the job so it physically cannot overwrite an existing partition.

Mixing ARR and recognized revenue. Starting figure is ARR, ending figure is GAAP revenue, and the two get treated as comparable because both are dollars. They are not — one is a forward run-rate, the other a backward recognition. Pick one basis for the metric and reconcile to the other explicitly.

Letting expansion leak into GRR. If your GRR ever prints above 100%, the bridge is broken. Rebuild it before anyone else notices.

Blending segments. A company-wide NRR of 110% might be 125% enterprise and 92% SMB. Those are two different businesses with two different futures, and the blend describes neither.

Silent definitional changes. This is the one that costs trust permanently. Methodologies legitimately improve — moving from billing account to legal entity is usually an upgrade — but an undisclosed refinement that makes a number jump reads as manipulation. Disclose it, restate the prior period on the new basis, and show a bridge:

That table turns a potential red flag into evidence of rigor. It takes ten minutes to build and an auditor can verify it in five.

The edge cases deserve pre-decided answers, because deciding them under audit pressure always looks like rationalizing:

Behind all of it sits the artifact that actually answers "which is real": a controlled retention methodology document, three to six pages, stating the customer definition and stable key, the cohort window and snapshot procedure, the ARR definition including the recurring/non-recurring boundary, the currency policy, the expansion/contraction/churn boundary rules with edge cases, the ASC 606 modification handling, the reconciliation procedure, and a version history with dates, authors, and the reason for each change. The CFO signs the current version. A company without this document is *asserting* its retention metrics. A company with it is *proving* them.

The recurring/non-recurring boundary inside that document is worth its own care, because a single misclassified six-figure services line can move a small cohort's NRR by a full point. A multi-year implementation fee billed annually is recurring in cash but non-recurring in substance — exclude it from ARR. Premium support sold as an annual add-on renews with the subscription — include it. A usage overage recurring monthly at a variable level — include but normalize. A one-time migration credit — exclude. An auditor sampling ten contracts is checking whether your classification is consistent, not whether it matches their preference.

Two adjacent disciplines close the loop.

The dry run. Before the real audit, have your finance team play auditor. Ask them to reproduce a quarter's NRR from raw data, trace a sampled customer through the bridge, confirm the cohort was never re-pulled, and verify Q1 and Q4 used identical definitions. Every gap the dry run surfaces is a gap you fix on your own calendar instead of in front of the committee. Teams that dry-run walk in confident; teams that do not improvise, and improvisation is exactly what makes a number look unreal.

Ownership. The three-spreadsheet problem re-emerges the instant an input has two owners. Finance owns the ARR definition, the currency policy, the ASC 606 treatment, and the GAAP reconciliation. RevOps owns the bridge model, the cohort snapshots, and the segmentation. Customer Success owns operational interpretation and churn-driver analysis. The CFO owns the methodology document as a controlled document and signs before every board meeting. Single ownership per input, never split.

On presentation order, one last piece of practical advice: lead with GRR, then decompose NRR as GRR plus the expansion contribution, then show logo retention as the concentration check. An audit committee that sees GRR first concludes you are not hiding behind expansion. One that sees a big bold NRR first concludes you might be. And when the chair asks the question directly, the answer is a sentence you should have ready: *all three are real because they come from one bridge, one frozen cohort, and one currency policy — here is the reconciliation to recognized revenue, here is the methodology document, and a second analyst reproduces these exact figures.* That ends the conversation. An argument about industry norms does not.

Match the rigor to the stakes, though. A seed-stage company with forty customers and no audit committee does not need this apparatus — one churned logo swings the number two to four points, so cohort statistics are noise and a formal bridge is premature overhead. A purely transactional business should use repeat-purchase rate instead, since NRR and GRR are recurring-revenue constructs. A company deliberately sunsetting a product will post terrible retention and *should*; annotate the exhibit rather than torturing the cohort definition to hide a strategic decision. The full machinery is for boards with audit committees, companies approaching an S-1, and financings where retention drives the valuation.

Related questions

Can GRR ever legitimately exceed 100%?

No. GRR excludes expansion by construction, so the maximum is a perfect period with zero contraction and zero churn. A GRR above 100% always means expansion has leaked into the calculation — usually a price escalator or a seat add miscoded as a non-loss. Rebuild the bridge.

Should we present NRR on a constant-currency or reported basis?

Constant currency as the headline, reported as a secondary disclosure with a footnote on the conversion method. Constant currency isolates customer behavior; a ten-point FX swing can move a Europe-heavy cohort by four to six NRR points with no behavioral change at all.

How do we handle a customer acquired by another one of our customers?

Map the acquired entity into the surviving entity and net the ARR — do not record a churn plus a new logo. Document the merge in the snapshot notes so the logo count drop is traceable. If the acquirer terminates the contract outright, that is churn.

What if we have no historical cohort snapshots at all?

You cannot retroactively freeze a cohort; regenerating it from live data is exactly the failure auditors look for. Start snapshotting immediately, disclose that clean anniversary cohorts begin twelve months out, and present TTM figures with an explicit methodology caveat in the interim.

Does logo retention matter if our NRR is strong?

Especially then. Strong NRR with weak logo retention means a few large accounts are carrying the base while the tail bleeds out — concentration risk the dollar metric cannot show. It is the exact pattern that surprises boards when one whale finally leaves.

FAQ

Why do auditors trust GRR more than NRR?

GRR has no offsetting term. It captures only contraction and churn, is capped at 100% by construction, and therefore cannot hide a deteriorating base behind a strong upsell motion. NRR is not less valid, but it has four well-known inflation levers — cohort selection, price escalators, currency tailwinds, and survivorship — that a skeptical reviewer has to rule out one by one. GRR requires none of that ruling-out, so it is the number an auditor starts from.

What exactly does "one frozen cohort" mean in practice?

It means the starting population is captured once, at a timestamp, and written to storage that cannot be overwritten. All three metrics then read from that same stored population. It does not mean "we agreed on the definition" — it means the actual rows are physically preserved, so a second analyst twelve months later can query the identical starting base rather than reconstructing it from a system that has since changed underneath them.

How do usage-based businesses compute a defensible starting ARR?

By deriving it from trailing actual consumption over a defined lookback window rather than a forward estimate, and footnoting the derivation. The reason matters: under ASC 606 variable consideration is estimated and constrained, so if starting ARR is itself an estimate, part of your reported expansion is really an estimate revision rather than a customer buying more. Separating those two is what an auditor will ask for, and trailing-actual is the cleanest way to give it to them.

Do we have to restate prior periods when we change a definition?

Yes, and you should want to. Restate the prior period on the new basis and publish a bridge showing each change's effect in points of NRR or GRR. Audit committees do not object to methodological improvements — they object to numbers that move without explanation. The bridge converts what looks like manipulation into demonstrated rigor, and it takes about ten minutes to build.

Is RPO the same thing as deferred revenue?

No, and auditors use the distinction as a competence check. Deferred revenue is only the billed-but-not-yet-recognized portion. RPO is deferred revenue plus contracted-but-unbilled backlog, which for a three-year contract billed annually means two additional years sitting entirely off the balance sheet. Citing deferred revenue when asked about contracted backlog understates the forward book substantially.

Should logo retention include customers who went dormant but never formally cancelled?

Decide the rule in advance and write it down. The common defensible treatment is to count a logo as churned when the contract terminates or expires unrenewed, not when usage drops to zero — usage-based dormancy is a health signal, not a contractual event. Whatever you choose, apply it identically in every period, because a rule that changes based on which answer looks better is the definition of a metric that is not real.

Sources

• https://asc.fasb.org/ — FASB Accounting Standards Codification (Topic 606, Revenue from Contracts with Customers; Subtopic 340-40)
• https://www.sec.gov/edgar/searchedgar/companysearch — SEC EDGAR full-text search for 10-K and S-1 retention disclosures
• https://www.sec.gov/rules/final/2003/33-8176.htm — SEC final rule on the use of non-GAAP financial measures
• https://viewpoint.pwc.com/us/en/pwc/accounting_guides/revenue_from_contrac.html — PwC Viewpoint guide to revenue from contracts with customers
• https://kpmg.com/us/en/frv/reference-library/2023/handbook-revenue-recognition.html — KPMG Financial Reporting View revenue recognition handbook
• https://www.bvp.com/atlas — Bessemer Venture Partners Atlas, State of the Cloud benchmark reports
• https://www.pcaobus.org/oversight/standards — PCAOB auditing standards library
• https://www.ifrs.org/issued-standards/list-of-standards/ifrs-15-revenue-from-contracts-with-customers/ — IFRS 15, the IFRS counterpart to ASC 606

Related on PULSE

• [How do you calculate true CAC payback period when you have multi-quarter sales cycles?](/knowledge/q414)
• [What does the Rule of 40 actually measure, and how do you explain it when your score misses?](/knowledge/q417)
• [How do you model CAC for usage-based pricing when you have no upfront contract value?](/knowledge/q419)
• [How should you forecast financial health with multi-year contracts, holdbacks, and payment delays?](/knowledge/q423)
• [How do we build a tiered partner program that rewards scale without collapsing margin?](/knowledge/q429)