"Sector Refresh" — CJ Gustafson (Mostly Metrics)

Why this is in the vault

CJ is reshuffling his entire public-comp universe — 138 companies across nine indexes — to reflect how tech actually monetizes in 2026 rather than how it bucketed up in the 2020 SaaS bull. This is the methodological underbed for every Mostly Metrics weekly going forward: when CJ talks about "the median Vertical SaaS NTM multiple" next month, it's this taxonomy. Three things make it load-bearing for RDCO:

1. The taxonomy is the operator's comp set. It's explicitly designed for CFOs/operators trying to find their comp set, not for sell-side analysts. That's the same lens MAC ships against and the same lens the per-bet P&L ledger needs.
2. AI-infra is now its own bucket (#2 Data & AI Infrastructure, 13 companies). Cerebras, CoreWeave, Snowflake, Equinix, Confluent now live together. Two of those (Cerebras, CoreWeave) didn't exist as public comps two years ago. That's the corroborating evidence for the "capex-intensive AI-infra layer is now a distinct public-market sector" thesis.
3. "Take-Rate Platforms" got split out of Vertical SaaS (new #7, 19 companies). Toast, Shopify, Olo, ServiceTitan-adjacent commerce — anything monetized on transaction volume rather than seat licenses. This is exactly the unit-economics split the agentic-capital-markets framing needs: outcome/transaction monetization vs seat-license monetization is the cleanest empirical line in his data.

The new nine (one-line each)

1. Security & Identity (17) — CISO budget. CrowdStrike, Palo Alto, Cloudflare, Zscaler, Okta…
2. Data & AI Infrastructure (13) — software-native data/AI/GPU. Snowflake, CoreWeave, Cerebras, Confluent, MongoDB, Equinix… Legacy hardware (HPE, NetApp, Lumen, Rackspace, Cisco) cut.
3. Dev Tools & Observability (10) — engineering budget. Datadog, Atlassian, Figma, Dynatrace, GitLab, UiPath…
4. Horizontal SaaS & Back Office (18) — cross-industry ops/HR/finance/collab. Oracle, ServiceNow, Workday, ADP, Zoom, DocuSign, monday, Asana, Workiva…
5. GTM (MarTech & SalesTech) (18) — revenue org budget. Salesforce, Adobe, HubSpot, Trade Desk, Twilio, Klaviyo, Braze, ZoomInfo…
6. Vertical SaaS (16) — industry-specific, no take-rate. Palantir, Autodesk, Veeva, Aspen, Samsara, ServiceTitan, Guidewire, Tyler, Procore…
7. Take-Rate Platforms (19) — NEW. Marketplaces + commerce platforms earning on transaction volume. Uber, Airbnb, Shopify, MercadoLibre, DoorDash, Toast, Lyft, Opendoor, Etsy, Olo…
8. Payments & Money Movement (11) — the rails. Intuit, Fiserv, Adyen, PayPal, Block, Shift4, BILL, Marqeta…
9. Consumer Fintech, Lending & Crypto (16) — front-end, CAC-heavy. Coinbase, Robinhood, SoFi, Chime, Affirm, Upstart, Circle, Klarna…

Cut from universe: HPE, NetApp, Lumen, Rackspace, Cisco, Newegg, BigCommerce. Total now 138 companies.

Methodology (carry-over, unchanged)

• Revenue multiple basis: EV / NTM Revenue (Enterprise Value = Market Cap + Total Debt - Cash). Refreshed weekly because market cap moves daily; debt/cash come from last quarterly. 10x NTM historically = premium ("best of best") territory. Multiples expand in low-rate environments, compress in high-rate.
• CAC Payback: S&M (TTM, lagged one quarter) / (YoY change in latest-quarter revenue × 4) / GM%. The lag is to smooth seasonality and recognize pipeline lead time.
• Revenue per Employee: TTM Revenue / Current Employees. Public-co rule of thumb: north of $450k/employee at scale.
• Rule of 40: TTM Revenue Growth % + TTM Adjusted EBITDA Margin %. Should clear 40.
• NDR called out as great-but-disappearing — most public cos now disclose annually with a threshold (above/below 100%) instead of an exact number.
• OPEX buckets: COGS / S&M / R&D / G&A. All GAAP, includes SBC.
• Data partner: Koyfin.

Mapping against Ray Data Co

Strong mapping — three concrete uses, one of which is today's work.

1. Per-bet P&L ledger (TODAY's work). The ledger needs a comp-set column for each bet. CJ just gave us the canonical 2026 taxonomy:

   • Squarely (mobile puzzle, ad + IAP economics) — none of CJ's nine fit cleanly. Closest is Take-Rate Platforms (#7) for the "monetized on engagement-volume not seat license" shape, but mobile-game economics differ enough that Squarely's comp set lives outside the Mostly Metrics universe (App Annie / Sensor Tower benchmarks instead). Worth noting in the ledger that Squarely is deliberately off the SaaS taxonomy.
   • MAC info-product — Horizontal SaaS & Back Office (#4) is the wrong bucket; this is closer to a creator-economy / info-product unit-economics shape that CJ doesn't index at all. But the CAC Payback + Rev/Employee + Rule of 40 triplet IS the right operator-comparable metric stack to track for MAC. Pick the metrics, skip the bucket.
   • Sanity Check newsletter — same story. Not in CJ's universe, but the framework (CAC payback as the discipline check, Rule of 40 as the growth/efficiency balance) applies.
   • Any future RDCO agent-as-service bet — would map to Horizontal SaaS & Back Office (#4) IF sold seat-license, OR Take-Rate Platforms (#7) IF outcome/transaction-priced. This is the seat-vs-outcome split that the agentic-capital-markets thesis [[2026-05-04-innermost-loop-singularity-measured-by-minds]] hinges on, and CJ's data now lets us empirically compare median multiples across the two monetization shapes.

2. AI-infra IPO comp set (corroborates [[2026-05-12-mostlymetrics-cerebras-ipo-s1-breakdown]]). Cerebras-the-public-company is now in a 13-name peer set led by Snowflake/CoreWeave/Equinix. The S1 breakdown flagged 86% revenue concentration risk; the sector context now shows Cerebras entering a sector where most names are not capex-intensive (Snowflake, MongoDB, Confluent are software-margin). This sharpens the read on Cerebras vs CoreWeave as the only two true capex-heavy AI-infra public comps — they will trade differently from the software-margin peers in the same index, and the median will be misleading. For the singularity-economy index-buy thesis [[2026-05-16-moonshots-ep255-anthropic-spacex-leopold-singularity-economy]], #2 Data & AI Infrastructure is the closest sector-ETF approximation; cap-weighted, it will be dominated by Snowflake/Arista/Equinix not by the actual AI-compute names.

3. Sanity Check voice study (medium). CJ keeps doing the pop-culture handle (Drake's three albums → his nine sectors; the Slipknot footnote). Banked as a voice-anchor data point for the Sanity Check Mostly-Metrics-style cadence study, alongside [[2026-05-11-mostlymetrics-ltv-cac-nickelback]] and [[2026-05-03-mostlymetrics-revenue-hierarchy-is-it-cake]].

No new SC topic. This piece is methodological/taxonomic, not analytical — restating it would be derivative per the no-derivative-Sanity-Check-pieces rule. The taxonomy becomes evidence for future SC pieces (e.g., a piece on the seat-vs-take-rate split as the canonical SaaS-disruption fault line, citing CJ's 19-name Take-Rate index split-out as proof the market is already pricing this difference), not a topic of its own.

Related

• [[2026-05-14-mostly-metrics-q1-2026-startup-benchmarks]] — the private operator benchmark companion to this public-comp taxonomy refresh. Use them together: CJ's private CAC/NDR/Rule-of-40 medians by ARR-band + CJ's public sector medians by index = full operator comp picture.
• [[2026-05-12-mostlymetrics-cerebras-ipo-s1-breakdown]] — Cerebras gets a home in new sector #2 Data & AI Infrastructure; this refresh is what made that placement formal.
• [[2026-05-11-mostlymetrics-ltv-cac-nickelback]] — same CJ voice cadence (banger / pop-culture handle).
• [[2026-05-03-mostlymetrics-revenue-hierarchy-is-it-cake]] — same operator-lens-on-unit-economics frame; the six-level revenue hierarchy and the nine sectors are complementary cuts at "what kind of revenue does this company actually have."
• [[2026-01-23-mostlymetrics-vertical-saas-metrics]] — Vertical SaaS metrics piece; #6 in the new taxonomy. Note the bucket got smaller because Take-Rate Platforms got split out.
• [[2026-04-03-saas-metrics-that-matter]] — David Sacks metrics dictionary; CJ's CAC Payback / Rule of 40 / Rev-per-Employee definitions are the executable versions of Sacks's framework.
• [[2026-04-25-saas-death-thesis-vault-synthesis]] — the seat-vs-outcome split CJ just empirically separated is the same split that thesis depends on; CJ's data is now usable as direct empirical support.
• [[2026-05-16-moonshots-ep255-anthropic-spacex-leopold-singularity-economy]] — singularity-economy "buy the index" thread; sector #2 is the candidate index.
• [[2026-05-04-innermost-loop-singularity-measured-by-minds]] — agentic-capital-markets framing.