"Is LTV to CAC the Nickelback of SaaS Metrics?" — CJ Gustafson (Mostly Metrics)

Why this is in the vault

This is the canonical CJ banger. Title carries the entire essay. Voice study target #1 for Sanity Check — if I can decompose how this opener works and what makes the metric-as-pop-culture-villain frame land, I can scaffold a SC piece that uses the same shape. Also operationally relevant: the "compound metric without an owner" critique is exactly the trap I was about to walk into when I proposed an "AI cost efficiency score" for MAC last week.

⚠️ Sponsorship

Sponsored by Brex. Multiple embedded references throughout (not just a top block). Brex has a corporate-card / spend-management product that maps loosely to the unit-economics audience. The sponsor placement is heavier than the Abacum issues — Brex shows up inline. Doesn't appear to bend the editorial (the LTV/CAC critique stands on its own evidence), but flagging the embed density.

Core thesis

LTV to CAC is overweighted because it sounds quantitative while being structurally unaccountable. It's a compound metric with no single owner, built on theoretical-customer-lifetime assumptions, and routinely mis-calculated three different ways. CJ's prescription: replace it with CAC Payback Period (cash-survival lens) and Net Dollar Retention (existing-customer expansion lens) — both have clear owners and decision implications.

The three common LTV/CAC errors

1. Forgetting gross margin adjustment. Using revenue not contribution margin inflates LTV by the COGS percentage.
2. Wrong churn rate. Using all-customer churn instead of cohort-specific. Or using monthly churn annualized linearly when it isn't linear.
3. Refusing segmentation. Blended LTV/CAC across SMB+Enterprise hides the dynamics that actually matter.

Other frameworks

• 3:1 SMB SaaS, 5:1 Enterprise as common-but-arbitrary benchmarks.
• Three-year minimum for LTV calculation to be statistically meaningful.
• CAC Payback Period as the better "are we still alive" metric.
• Net Dollar Retention as the better "is the existing book worth more" metric.
• The compound-metric ownership problem. If a metric requires four functions to move, no function owns it, so no function changes behavior in response to it.

Voice tactics — Sanity Check study notes

Three moves to lift:

1. Pop-culture-villain framing. "Nickelback of SaaS metrics" works because Nickelback is the agreed-upon villain — no one defends them. The frame instantly aligns the reader against the target before any argument lands. SC equivalent: pick something everyone secretly already disagrees with, brand it with the universally-mocked pop-culture parallel.
2. "I woke up and chose violence." Single-sentence stance-statement that announces tone. Telegraph that this is a take, not a survey. The founder uses a similar move on X. We should bake it into SC openers more deliberately.
3. The amorphous-blob image. Compound metrics "becomes an amorphous blob, dancing through decks without a care." Concrete visual for an abstract problem. Pattern: when arguing against an abstraction, anthropomorphize it as something undignified. Don't argue against the abstraction's logic — make it look ridiculous.

Mapping against Ray Data Co

• Direct kill of the "AI cost efficiency score" idea I was drafting. I was about to propose a single composite for MAC that bundled token cost, latency, output quality, and time-saved. CJ's critique applies one-for-one: no owner, four functions to move, theoretical-customer-lifetime equivalent in the "time-saved" assumption. Killing the composite. Replace with CAC-Payback-equivalent ("how many months until this AI workflow has paid back its setup cost") and an NDR-equivalent ("does the AI workflow get more valuable per quarter or less"). Two clean metrics with owners.
• Sanity Check structural lift. SC piece "The Nickelback of [thing]" template is reusable. Candidate targets: "the Nickelback of dashboards," "the Nickelback of OKRs," "the Nickelback of standups." Founder gets to pick the target. This is a format I'd use freely.
• Voice-match skill input. The three lifted tactics (pop-culture villain, violence stance-statement, amorphous-blob anthropomorphism) should be added to the voice-match reference corpus as positive examples.
• Brex disclosure. When/if MAC is ever Brex-adjacent (we recommend a card stack to a portfolio company), I should disclose I read this and that Brex sponsors CJ's work. Not material now, but worth the mental flag.

Related

• [[2026-05-03-mostlymetrics-revenue-hierarchy-is-it-cake]]
• [[../01-projects/sanity-check/voice-study]]
• [[../01-projects/sanity-check/draft-pipeline]]