When Credits Meet ARR
Start simple, track them separately, and let your portfolio smooth the bumps
I keep hearing the same question lately.
From operators. From investors with portfolio companies. From private companies rolling out AI features.
They’ve either launched a credit-based pricing model or are thinking about it.
And they’re stuck on the same problem:
How do you roll credits into your revenue reporting without overcomplicating it?
The temptation I see isn’t to design a perfect system up front, it’s to over-engineer the solution… to think through every possible edge case, every way it could go wrong, and then build something more complex than it needs to be.
I got on a call with someone last week who was considering building a system designed around rolling averages to smooth credit usage.
If a customer has history, you average their trailing six months of usage and roll that into ARR. If they don’t, you use whatever average you can find. Sounds reasonable. In fact my first reaction was, that makes total sense.
As we got deeper into the conversation tho, it gets pretty complex. When credit purchases are lumpy, those averages add complexity without adding much clarity. In fact, imagine a customer buys a bunch of credits in month one and never buys credits again… you’ll be explaining contraction for the next 6 months for one event. And that’s just across one customer. Across dozens or hundreds, it feels like it’d get messy, your expansion and contraction would require looking across 6 months of history…
My bias is toward simplicity.
It’s far easier to add complexity later than to remove it once it’s embedded. A simpler system is faster to set up, easier to explain, and far easier to inspect when something doesn’t look right.
That conversation ended us with a simpler recommendation which admittedly works really only for high-volume, self-serve businesses selling one-off credits:
Track credits separately from your subscription ARR. You can give them their own column or line item in your ARR calculation so they remain distinct from the recurring component of your subscription model. Then, count those credits as expansion or contraction in ARR as they happen.
Yes, that means more volatility at the individual customer level. That’s intentional. This is a portfolio strategy, you’re accepting that when you peel back the curtain on any single customer, it’s going to look spiky. That’s okay.
The bet is that over time, the ups and downs of individual customers will smooth out at the portfolio level, giving you a consistent and predictable baseline. It’s the same principle as investing: picking individual stocks is volatile, but an index fund delivers steadier returns.
Two other important points:
Be upfront – Whether you’re reporting to your board, execs, or management team, be clear about the methodology, why you chose simplicity, and what its limitations are. Commit to evolving it as you learn more about customer behavior.
Set expectations – Volatility will happen at the customer level. If a large expansion or contraction hits in a quarter, it’s not a surprise, you’ve already discussed that it’s possible and agreed on how to handle it.
If you’re sales-led with only a handful of large customers, this approach won’t work… a single customer’s swing could shift the entire portfolio. But for high-volume, self-serve companies, starting simple and managing at the portfolio level will save you from unnecessary complexity.