What’s Actually Working: Audit Evidence for Usage & ARR When Usage Commits Get Blown
Highlights from the April shareforum: preparing audit evidence for internally generated usage data, and how companies are calculating ARR when usage overages are in the mix
Hey there 👋,
Welcome to the very first issue of the Gaapsavvy Newsletter Field Notes — an industry view of what's actually happening on the ground in SaaS accounting — what's working, what's broken, and what's next.
After years of running shareforums, I’ve long wanted to start a newsletter and just didn’t know where to start. But as they say, "never let perfect be the enemy of good" — it's better to just start.
At our last shareforum from Klarity HQ, we brought together 65+ accounting professionals from leading tech companies to talk about what’s really happening.
In my accounting career, I've often wondered "someone else must have this same question, I wonder what they are doing."
Turns out you’re right — you’re not alone.
Let’s dive in.
This community and newsletter is brought to you for free by Klarity. AI for Finance Transformation, helping accountants bend the G&A curve with AI-powered automation for process documentation and contract review. Thank you for supporting the Gaapsavvy community!
Check out what’s new in Klarity Architect on June 10th : Register here to learn how you can leverage AI in your process documentation and internal controls.
Want to listen instead? Check out this AI generated audio overview in podcast form, generated by Google’s NotebookLM. Not perfect, but mind-blowingly good.
Quick Hits from the April Shareforum:
Who’s at the table this round: 65+ accounting professionals from leading tech companies. Company profiles: 73% private to 27% public companies - 88% audited by Big 4 auditors (indicating Series C+ maturity preparing for IPO).
Here are the questions we discussed in April:
Partner Reseller revenue share model, how do you account for them? The gross vs. net debate is back in focus — possibly sparked by this this new publication. Reality check: even if you theoretically conclude "gross" treatment is appropriate, you can't record revenue for end-user pricing you don't actually know.
Stripe Billing for Usage with NetSuite, what lessons pitfalls can you share? - Engineers love Stripe Billing, but accountants face manual workarounds for JE upload into NetSuite. Key insight: companies need SQL or Stripe Sigma skills just to extract data, and Stripe vs. Salesforce FX rounding decimal differences create persistent variances.
AWS Marketplace tooling, what intermediary software have you found success with? Cloud marketplace selling creates RevOps visibility gaps and look to specialized tools like Clazar, Tackle, and Suger.
Term License SSP allocation split benchmarking, what’s your % split? - Term license allocation ratio acceptability differ by audit firm - drifting from the classic 80% License/20% PCS split, to more heavily weight PCS. Compared to 2021, more companies are using a fixed term vs. variable term to arrive at estimated SSP (see Deloitte’s publication).
Sales tax presentation, do you include tax on sales orders, or just invoices? - Most companies only show tax on invoices due to system limitations as tax engines integrate with billing not CPQ. International requirements sometimes exceptions often handled manually. Tax compliance penalties are generally treated as operating expenses rather than contra-revenue.
Focus Feature: Two Usage Questions from April
As more companies move to usage-based models, countless challenges emerge.
These two specific questions that came up in our April discussions...How do you make internally generated usage data auditable? And how do you calculate ARR when customers blow past minimum commits?
These challenges reflect a broader pattern - the industry keeps trying to force usage-based models into old frameworks that don’t quite fit.
Q: What have others done for audit evidence to support internally generated consumption-based billing data?
How do you get auditors comfortable with homegrown usage systems?
The Traditional vs. Digital Challenge
Traditional audit evidence doesn't work for digital usage data. Historically, auditors relied on external third-party evidence - shipping documents, customer confirmations sent directly by auditors.
In the digital world, "proof of delivery" is simply usage data generated internally by your homegrown product, where the system captures events based on instructions written in code.
"We built custom tracking and walked through the logic, but it still wasn't enough."
What's Actually Working
Based on our community polling, accounting teams need multiple types of audit evidence, not just one:
Real Examples from the Field:
PwC approach: Created a "test audit account" where auditors signed up as customers to verify system accuracy. Similar approaches confirmed across Deloitte and KPMG.
KPMG approach: Classic customer confirmations - auditors reach out to customers to validate usage data. While this creates a circular reference (customers confirming data from the system being audited), multiple firms still use this procedure.
Hybrid approach: One team described combining event tracing with database snapshots and quarterly code reviews with their engineering team.
Bottom Line: Start with event tracing + screenshots as foundation. Layer in 2-3 additional evidence types. If you're going public, propose the "auditor test account" approach early - three firms confirmed this works. Audit scrutiny increases dramatically after the public company transition.
Q: For usage contracts with minimum commits, do you include usage overage fees in Annual Recurring Revenue (ARR)?
The Core Problem
Traditional ARR calculations fundamentally break down for usage-based models because ARR assumes recurring predictability that doesn't always exist in consumption models.
ARR is unaudited, non-GAAP, and lacks specific rules on what should be included or excluded. ARR is used by investors as a forward looking metric, with an assumption that ARR will is recognized smoothly as GAAP revenue in future periods.
Usage-based companies often find themselves forcing consumption data into subscription metrics that weren't built for this business model.
What's Actually Working
Snowflake is arguably the trend setting consumption company - and one senior team member shared: "We don't use ARR ourselves. We use net recurring revenue and how that's increasing on a customer and cohort basis because we're usage rather than ratable SaaS."
( Check out Snowflake’s Q12-26 Investor Report as an example)
This highlights a critical insight: companies with truly consumption-driven models may need entirely different metrics rather than trying to retrofit ARR calculations.
The community revealed several distinct approaches, each shaped by company maturity and usage predictability:
The Conservative Exclusion Approach: Smaller companies often exclude all non-committed usage. "We keep track of them, but we don't include them as actual ARR due to our size—we don't want to get burned and have to reforecast."
The "Burned Once, Cautious Twice" Method: Companies that experienced volatility adopt waiting periods. "We got burned and decided to give it first annual experience. Only after 4 quarters when we come up to some reasonable rate, we are adding this to ARR."
The Quarterly Annualization Approach: Others take quarterly overage revenue and multiply by four, with caveats: "It helps to have history—if your overages are all over the map, then ARR times 4 just blows up when it goes down and blows up when it goes up, and investors get confused."
Key Insights:
Public companies are more conservative - 62% exclude non-committed usage vs. 51% for private companies
Usage volatility drives methodology choice - stable patterns enable more aggressive inclusion
The "4 quarters of history" rule is emerging as best practice for transitioning from exclusion to inclusion
Alternative metrics gaining traction - companies with heavy usage / consumption revenue exploring net recurring revenue growth
Bottom Line: If usage represents the majority of revenue, consider abandoning ARR entirely. Follow successful consumption companies' lead with net recurring revenue growth or customer cohort consumption trends. If you must use ARR for investor compatibility, exclude volatile overages until you have 4 quarters of predictable history. The goal isn't perfect ARR calculation—it's meaningful measurement of business trajectory.
What’s Next?
AI is eating our world. Coming off the Klarity SF AI Summit, I can't wait to dive deeper into sharing knowledge on how we’re using AI in our daily work. Quick share from me in crafting this newsletter using chatGPT, and Antropic’s Claude. Claude really blew my mind in how well it coded infographics and dug into insights.
More on AI in the next issue.
Upcoming events:
June 10: Klarity Architect demo - AI for process documentation Sign up here
June 13: Hands-on AI Masterclass Collab with Sowmya Ranganathan ex-OpenAI Controller & Founder at Lumera ( paid) Sign up here - and make sure to check out her super practical newsletter here!
June 17: Deloitte Roundtable: Accounting Implications for AI companies (in person) Sign up here
June 27: Next QTC shareforum for industry accountant community members
Thanks again for being part of this community. Hit reply if something in here sparked a thought, a question, or even a “we’re dealing with this too.” I’d love to hear from you.
A quick note: You're getting this because you’ve expressed interest in access to community content or joined a related Gaapsavvy program. While I have limited shareforum participation to practicing industry accountants only for now, I wanted to start sharing these insights with everyone who's shown interest in what we're building.
Not your jam? No hard feelings. Easy unsubscribe below.
Disclaimer
The information shared in this newsletter comes from real lived experiences our community of accounting professionals - and is intended for learning purposes only. Polling data reveal industry “leans” in an “it depends” world. Every company's situation is unique - always check with your auditors and accounting team before implementing new approaches. Use these insights as starting points for your own research and discussions.
Great share, Angela! Happy to see you here.