SaaS Social Media and Updates

Variable Pricing Requires Cohort‑Based, Smoothed NRR Metrics

Does NRR exist and/or matter where traditional ARR no longer exists? Usage-based and outcome-based pricing didn’t make NRR irrelevant - but they make "NRR Slop" hard to defend... The old SaaS world was simple: Annual contracts → clean ARR → clean NRR The new world of SaaS & AI-Native software pricing and revenue is different: • Base fees + variable usage • Credits, ramp periods, seasonality • Invoices that change every month • “ARR” that’s increasingly just a proxy for revenue that may recur Trying to calculate NRR similar to the method used just a couple of years ago is how companies end up arguing against their own data, and investors start to question every metric Here’s my line in the sand: If pricing and the associated revenue is variable, NRR must be cohort-based and time-smoothed, or it becomes full of noise and often misleading Monthly and quarterly calculations create artificial expansion and contraction I’ve seen perfectly healthy customers look like churn, and wildly volatile usage-based revenue be used to determine product-market fit where it does not exist What actually works in a variable pricing model: ✅ Define a clear cohort (no survivor bias) ✅ Measure trailing spend, not annualized moments (month/quarter) ✅ Use longer periods so variability has time to offset ✅ Accept that “ARR” is now an opinion (or forced calculation) - not a truth NRR should answer one primary question: "Did this same group of customers deliver more durable economic value over a meaningful amount of time? NRR should not primarily tell you: 👉 Did usage spike this quarter? 👉 Did implied ARR look good last quarter? 👉 Did we annualize the right month(s) or quarter? In a variable pricing environment, ARR is no longer a forward-looking metric It used to straddle the future because ARR was contractual... ...Now in many usage-based environment it’s inherently backward-looking If you want to predict what’s coming next: 🔦Watch sequential product utilization trends 🔦 Watch, capture and try to normalize ramp rates 🔦 Watch seasonality-adjusted usage trends In the new world of software pricing - NRR may be the same acronym - but requires different math with a much higher "prove it" standard Dave "CAC" Kellogg and I unpack the tradeoffs, edge cases, and pros/cons of all of this on two full episodes of The Metrics Brothers including: 👉 Calculating ARR 👉 Calculating NRR 👉 Factoring in Ramp Periods 👉 Defining Expansion ARR 👉 Two-year lookback method My takeaway includes two primary points: 1. In a variable pricing world, fast NRR is usually wrong - slow NRR is more accurate 2. Spend is telling - Spend over time is truth 🎧 to the full discussion on your favorite podcast channel (The Metrics Brothers)