"Usage-Based is Best If You Are Already" — CJ Gustafson (Mostly Metrics)
Why this is in the vault
CJ's contrarian re-frame: usage-based pricing isn't intrinsically better than subscription — it AMPLIFIES whatever the product already does. Strong products get stronger with UBP; mediocre products die faster with UBP. Directly relevant to any MAC pricing-structure conversation and to the broader RDCO question of whether we should be pricing surfaces by usage or by subscription. The "self-fulfilling prophecy" framing is also clean — UBP looks like a magic-bullet metric in the literature only because the companies that switched to UBP were already winning.
⚠️ Sponsorship
Sponsored by Pigment (business planning platform). Pigment's product directly benefits from "your pricing model is complicated, you need a planning tool" framing. CJ doesn't lean into that, but the structural alignment is there. Read with light source-bias filter.
Core thesis
Top usage-based-pricing companies show nearly 2x the growth rate and 20-40% higher NDR than top subscription companies — but this is a selection artifact, not a causal effect. UBP doesn't cause winning; it amplifies whatever differentiation already exists. A genuinely differentiated product benefits disproportionately from UBP because customers expand naturally as they extract value. A weak product suffers disproportionately because customers throttle usage when value doesn't materialize.
The four-quadrant breakdown
| UBP | Subscription | |
|---|---|---|
| Pro | Aligns revenue with value; uncapped upside; natural NDR | Predictable revenue; easier to forecast; lower CAC |
| Con | Volatile revenue; customer can throttle; harder to budget | Caps the upside; misaligned with value; renewal cliff risk |
Frameworks
- The performance comparison matrix as the headline data point — top-quartile UBP companies > top-quartile subscription companies on both growth and NDR.
- The self-fulfilling prophecy model — UBP is the result, not the cause, of being a strong product. Implication: don't switch to UBP hoping it fixes your retention problem; fix your retention problem first, then UBP amplifies it.
- The volatility-vs-upside tradeoff as the core operator decision — which side of the tradeoff you can stomach is a forecasting / cash-management question, not a pricing-theory question.
Voice tactics
- The Dillon Panthers vs. art-school metaphor for "strong vs. weak company under UBP." Specific, weird, vivid. The kind of "I'm willing to commit to a strange metaphor and ride it" move SC voice needs more of.
- "I vomited into my Mostly Metrics yeti." Self-deprecating reaction-image to a difficult operating moment (Salesforce license audit). Plants flag in the same self-deprecating-while-authoritative pattern from the Cake piece.
Mapping against Ray Data Co
- MAC pricing reframe. I was considering "percentage of demonstrated AI savings" as the MAC pricing structure (a UBP-shaped model). CJ's frame says: this only makes sense if MAC's underlying differentiation is real. If MAC really does identify cost optimizations the customer wouldn't find alone, UBP-shaped pricing amplifies that. If MAC is just "another consultant with frameworks," subscription/retainer is safer because it caps the customer's downside-of-disappointment. The pricing structure question is downstream of the "is the product real" question. Honest answer right now: MAC is unproven. Start subscription/retainer, transition to UBP after the first 3 engagements demonstrate repeatable savings.
- Squarely subscription test. Squarely's potential subscription layer (puzzle archive, themed packs) is structurally subscription, not usage. The right test would be a low-price ($1.99/mo or $9.99/yr) subscription anchored to the daily-ritual value, not a usage-based "pay per puzzle" model. UBP for a puzzle app would be friction (counting puzzles), not amplification.
- Sanity Check pricing. If SC ever monetizes via paid tier, it's structurally subscription, not UBP. Newsletters don't have a clean "usage" unit. The pricing literature here is about subscription tiering (free / paid / pro) rather than UBP transition.
- General principle to carry forward. Pricing model amplifies differentiation; it doesn't create it. Useful filter for any time I see a "we'll switch to X pricing model and growth will improve" argument. The right question is always "is the underlying product strong" first.
Related
- [[2026-05-11-mostlymetrics-avoid-pricing-mistakes]]
- [[2026-05-11-mostlymetrics-cac-payback-calculation]]
- [[../01-projects/mac/engagement-structure]]