Welcome Back to This Week’s CFO Mailbag
This week we have the following finance leaders answering your reader questions:
Vanna Krantz, CFO @ GLG (former CFO of Disney+, Grindr, and Masterclass)
Mitzi Yue, SVP + Head of Finance @ Boulder Care (former CFO at Quilted Health, and VP of Finance at Brightline)
If your question is picked for the Mailbag, you’ll win a Mostly Metrics Yeti Rambler ($84 economic value, unlimited street cred).
Here’s what we’ll cover:
Adjusting Metrics Post RIF
Classifying LLM Costs on the P&L
Reporting Usage Based Contracts as ARR
Controller or FP&A? Sequencing Hiring of Finance Team
Monthly KPI Packets for Investors. Gross.
Let’s get into it!
Question #1: Adjusting Metrics Post RIF
We did a 15% RIF last quarter to get to default-alive.
Burn multiple looks great now, but the severance and real estate exit costs are distorting our Rule of 40 for the next two quarters.
Do you adjust the metric, footnote it, or just take the hit and let the trend speak for itself? Worried about setting a precedent for "adjusted" numbers we'll regret later.
Vanna from GLG:
I agree, I believe adjusting numbers is a less desirable approach as it gives the impression of less credibility. Hence, I would let the numbers flow and speak to it - especially if it is just 2 quarters.
Mitzi from Boulder Care:
My preference is to keep it simple. Report as-is, and then footnote what the adjusted number would be less the severance and RE costs.
Question #2: Classifying LLM Costs on the P&L
Our engineering team is burning through $180K/month on LLM API costs, and it's growing 20% MoM.
Right now it sits in R&D, but our CRO is arguing the customer-facing AI features should hit COGS since they scale with usage.
How are you thinking about where AI inference costs belong, and does the answer change how you talk about gross margin to the Street?
Vanna from GLG:
This is a great question! I never considered it hitting COGS if it’s for development and not packaged into the product - I can see some logic although I’m not sure they really scale with usage. I could imagine a world where AI usage is more justifiable as a cost of generating revenue, but I don't know if I see the direct correlation yet. I might think of it as a “tool” like Word or Excel or Salesforce.
Mitzi @ Boulder Care:
It depends! We can't apply monolithic accounting policies to AI inference out of context - i.e. "all customer-facing AI belong in COGS". First, what is your product and what is the inference doing?
In general, I'd say if the AI is required to deliver the product/service to a customer and it scales with customer usage, at least a portion of it belongs in COGS. The challenge is how to determine what portion to allocate to COGS vs. R&D. To that, I'd try to separate costs related to the non-customer facing development (ex: internal use, model training, experimentation, etc.) and allocate that portion to R&D.
Question #3: Reporting Usage Based Contracts as ARR
We sell usage-based contracts with minimum commits.
Our sales team reports the full commit as ARR, but actual consumption is running at 70%.
Investors are starting to ask sharper questions about the gap. What's your framework for which number to lead with, and how do you handle it when commit ARR and consumption ARR tell different stories about retention?
Vanna from GLG:
Another great question. I have relevant experience with the same issue and also have booked full committee as ARR, but that doesn’t always flow through and causes tremendous volatility when these contracts are trued up. I would prefer to book 80% to allow for upside to expectations while still providing less volatility as this support is justifiable.
Mitzi from Boulder Care:
This is not my world, so what she said ^
Question #4: Controller or FP&A? Sequencing Hiring of Finance Team
I'm a fractional CFO transitioning to full-time at a $12M ARR company. The founder wants me to hire a controller first, but I'm leaning toward a senior FP&A analyst since our books are clean and the real pain is forecasting. What's your take on sequencing the first two finance hires, and how do you avoid hiring for yesterday's problem?
Vanna from GLG:
I would advise that wherever you “spike” you complement your skillset as you build out since every CFO for a tiny company is hands on.
Mitzi from Boulder Care:
I'd be careful to assume "clean books" = yesterday's problems are all solved. Whether or not you need a controller depends on your current accounting situation - are you using outsourced accountants, or a more junior in-house team? What is your ERP system - are you on QBO / will you need to implement a new system?
Your current accounting situation may not be a bottleneck right now, but I would stress-test if that is true if the company grows 2-3x before determining you don't need a controller first. Also, forecasting is evolving right now in real-time with AI, you may find you don't need the FP&A headcount right now.
Question #5: Monthly KPI Packets for Investors. Gross.
We just closed our Series C and added two new board members who are asking for monthly KPI packets with deeper cohort cuts than we've ever produced. My FP&A lead is already stretched. How much of this should we just agree to vs push back as to what we think is important about the business?
I'm game to slice and dice data if it's helpful for decision making, but I don't want to go on safaris for data at the investor's whim.
Advice so this doesn't turn into a constant book report?
Vanna from GLG:
I agree, unless the board members are highly experienced operators who have had success with the KPI they are suggesting - then I wouldn’t flex to their curiosity given they are meant to provide governance and oversight rather than management.
Mitzi from Boulder Care:
Been there... It's a balance of not blindly taking every request, while also being careful not to dismiss new perspectives. As management, take the lead here and define the operating metrics that best reflect the business performance, and help the new board members understand why those are the right metrics to focus on.
That said, sometimes fresh board members do have a new take, and those cohort cuts may reveal some interesting insights! If your FP&A lead's bandwidth is the issue, you should try to automate as much as you can.
Wishing you a reporting cadence that doesn’t feel like a book report,
CJ
