mbfinotti/advertising-skills / skills/cac-roas-benchmark
cac-roas-benchmark
VerifiedCompute a business's CAC and ROAS family metrics from real spend and outcome data, then judge whether the spend is healthy against three references - the business's own break-even, its own trailing history, and provenance-labelled external benchmarks - returning healthy, watch, unhealthy, or insufficient evidence. Use whenever the user asks whether their CAC is too high, whether a ROAS is good, what ROAS to aim for, whether ads are profitable, or mentions MER, blended CAC, payback period, or a spend health check - even if they never say 'benchmark'. Covers B2B/SaaS (cost per SQL, cost per closed-won, cohort lag) and B2C/ecommerce. Do NOT use to set those thresholds as policy - use mbfinotti/advertising-skills@ad-spend-guardrails instead.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 29.7 KB |
| evals/evals.json | config | 35.7 KB |
| references/benchmark-sources.md | markdown | 15.6 KB |
| references/worked-examples.md | markdown | 8.5 KB |
references/benchmark-sources.md
View on GitHub# Published CAC/ROAS benchmarks - provenance, figures, folklore
Every CAC/ROAS benchmark in circulation is either a vendor's client sample or a self-reported survey. There is no neutral, audited, industry-wide CAC or ROAS benchmark, and no standards body defines what enters the CAC numerator. Quote provenance with the number or do not quote the number.
## The evidence contract
Before citing any external figure in a report, attach all four fields - publisher, year, sample, and the metric variant it measured - plus the publisher's commercial interest when it has one. A figure missing any of these is quoted as "unverified" or dropped.
## Publishers worth naming
Rows run best-first, ordered by what a citation from each tier is worth:
- value: platform-instrumented vendor data > large self-reported surveys > agency client analytics > consultancy and AI-generated summary pages
- effort: all == (roughly an hour of sourcing per figure, whichever tier) - only value separates them, so never settle for a lower tier to save time
Platform-instrumented data is measured rather than recalled, but every tier is skewed to that publisher's own customer base: more credible is not neutral. The bottom tier is not a weaker source, it is not a source - trace the figure to the named publisher underneath it or drop it (see Circular citation below).
| Source | Type | Sample | Measured or self-reported | Commercial interest |
| ---------------------------------------------------------------- | ----------------------- | --------------------------------------------------------------- | ---------------------------- | --------------------------------------------------------------- |
| Triple Whale 2025 | platform-instrumented | 18,000+ ecommerce brands | measured | analytics vendor |
| Polar Analytics 2026 | platform-instrumented | thousands of Shopify brands, 16 industries | measured | analytics vendor |
| Varos / Billo 2025-2026 | platform-instrumented | thousands of campaigns; Billo: 80,000+ Meta video ads (H2 2025) | measured | benchmarking vendors |
| Benchmarkit (Ray Rike), 2025 B2B SaaS Performance Metrics Survey | survey | 583 participants; per-metric n = 21-149 | self-reported | advisory/data vendor |
| Aleph x Benchmarkit 2026 SaaS & AI Survey | survey | 342 companies (198 reported CAC payback), CY-2025 actuals | self-reported | FP&A vendor |
| Gartner CMO Spend Survey 2025 | survey | 402 CMOs, majority >$1B revenue, fielded Feb-Mar 2025 | self-reported | independent research |
| High Alpha (formerly OpenView) 2024 SaaS Benchmarks | survey | 800+ companies | self-reported | VC/studio |
| KeyBanc/KBCM, SaaS Capital | surveys | large private SaaS panels | self-reported | bank / lender to SaaS |
| First Page Sage, CAC by industry | agency client analytics | 120+ B2B and 103 B2C clients, Jan 2022-Aug 2025 | measured, non-representative | SEO agency; **self-disclosed 75% organic / 25% paid weighting** |
This ordering is a default that a variant match overrides: a tier-1 figure measuring a different variant is worth less than a tier-2 figure measuring the user's own. Check the variant before the tier.
OpenView ceased new investments in December 2023. Its SaaS Benchmarks Report moved to High Alpha (with Paddle). Cite 2024+ editions under the new owner.
## B2C / ecommerce figures
- Median ecommerce ROAS **2.04** - Triple Whale 2025, 18,000+ brands, vendor-instrumented. Far below the folk 4:1. Median paid-social ROAS 1.86-1.93x across ~35,000 brands in the same vendor's earlier dataset.
- By channel: paid search median ~4.5x, Meta 2.2-2.8x, TikTok ~1.4x rising to ~2.25x with value optimisation - Varos / Billo, 2025-2026, platform-instrumented campaign samples.
- Seasonality large enough to invert a verdict: Toys & Games ROAS 1.46 in July vs 2.90 in December - Billo, H2 2025, 80,000+ Meta video ads. Black Friday / Cyber Monday can collapse per-order contribution margin 40-60% (Luca, vendor analysis).
- CAC and conversion by revenue tier (Polar Analytics 2026, thousands of Shopify brands):
- the $5M-$20M revenue tier shows the worst CAC efficiency
- $100M+ brands post the highest conversion rates at 6.29%
- First Page Sage ecommerce CAC $86 (combined organic + paid, 75% organic-weighted agency-client data, Jan 2022-Aug 2025).
## B2B / SaaS figures
- CAC payback median (three panels, presented as a conflict, never averaged - they disagree because panels and definitions differ):
- **18 months in 2024**, up from 14 in 2023 (Benchmarkit 2025, 583 participants)
- **16 months FY2025**, top quartile ≤6, bottom quartile ≥24 (Aleph x Benchmarkit 2026, 198 of 342 reporting)
- **20 months in 2024**, down from 25 in 2022 (KeyBanc)
- CAC payback by ACV segment (Benchmarkit 2025/2026, replacing generic working bands with a measured, segmented one - ACV predicts payback better than any other single attribute): sub-$5K ACV ~11 months; $10K-25K ACV ~12 months; $25K-50K ACV ~14 months; mid-market ~14-18 months; $50K-100K ACV ~22 months; above $100K ACV 18-24 months; above $250K ACV ~24 months. A below-median number at high ACV is not automatically a red flag - weigh it against net revenue retention, since high NRR can justify a longer upfront acquisition period.
- New CAC Ratio: median **$2.00 of S&M spend per $1.00 of new-customer ARR** in 2024, up 14% YoY, 4th quartile $2.82 (Benchmarkit 2025). Uses the SaaS Metrics Standards Board convention: New CAC Ratio = total S&M expense ÷ new-customer ARR - a convention, not an accounting standard.
- LTV:CAC medians: 3.2:1 (Optifai, N=939, Q2 2025-Q1 2026) and 3.6:1 (Benchmarkit 2025) - the "observed median near 3" is descriptive, not evidence for the 3:1 rule.
- CAC by vertical (First Page Sage, organic-weighted client data, Jan 2022-Aug 2025) - organic CAC runs roughly half of paid in most B2B verticals, so remember the sampling bias before comparing a paid-heavy account to any of these:
- B2B SaaS $239 combined ($205 organic / $341 paid)
- construction $281
- cybersecurity $387
- financial services $784
- real estate $791
- education $1,143
- fintech enterprise $14,772
- Cost per SQL $1,357 average vs cost per lead $198 average in the same 2024 dataset (First Page Sage) - the gap between the two is the point.
- Marketing budget (Gartner 2025, n=402, mostly >$1B-revenue firms) - descriptive of what large firms _do_, not what any firm _should_ spend:
- 7.7% of company revenue, down from 9.5% three years prior
- paid media 30.6% of marketing budget
## Attribution inflation (context for platform-reported ROAS)
- Observational/platform attribution overstated true lift by factors of ~7.0, 9.5, and 7.6 for upper/mid/lower-funnel outcomes vs randomized ground truth - Gordon, Zettelmeyer, Bhargava & Chapsky 2019, Marketing Science 38(2), 15 Facebook RCTs, ~500M user-experiment observations.
- Account audits show Meta and Google jointly claiming 150-200% of real revenue via double-counting - AdBeacon, vendor client audits (indicative, not audited).
- Global iOS tracking opt-in sat near 38% in Q1 2026 (Adjust benchmark, reported via SignalSeal), so most mobile attribution is modelled rather than deterministic - another reason a platform figure is a claim, not a measurement.
## Why cross-report comparison is unsound
- **Definitional mismatch.** Benchmarkit, First Page Sage, KeyBanc, and Triple Whale each use materially different numerators, customer definitions, and attribution scopes. Comparing your blended CAC to a published paid CAC measures the definitions.
- **Survivorship and self-selection.** Survey respondents opt in, failed companies are absent, and a VC or lender panel is a portfolio, not a market.
- **Disclosed sampling bias.** First Page Sage weights its blend 75% organic because it is an SEO agency - its own methodology note says so. A 90%-paid business cannot compare to it directly.
- **Small n.** Several Benchmarkit metrics rest on n = 21-43 (Blended CAC Ratio n=43, Expansion CAC Ratio n=21). Report medians and quartiles, never a point estimate.
- **Circular citation.** Many recent consultancy and AI-generated benchmark pages cite each other in loops. Trace a figure to a named publisher with a stated sample, or drop it.
## Folklore - full origins list
None of these traces to a study. Name the origin when a user quotes one, and put the business's own break-even next to it.
| Rule | Originator | Evidence status |
| ------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 3:1 LTV:CAC | David Skok (Matrix Partners), ~2011-2013, "SaaS Metrics 2.0" | Self-admitted guess - on record at SaaStr: "I guessed at that number, after visiting many, many SaaS companies." |
| 12-month CAC payback | David Skok, 2011 | Rule of thumb tied to 2011 fundraising conditions; never empirically validated. |
| 4:1 (4x) ROAS | No traceable author; often misattributed to a 2016 Nielsen study | The Nielsen analysis names no universal 4:1 and stresses category variance. 4:1 is break-even at a 25% contribution margin (1 ÷ 0.25), retroactively declared a target. |
| MER > 4, rising to 5-8 at scale | Taylor Holiday (Common Thread Collective) | Stated heuristic, never measured across a sample. |
| CAC ≈ 25% of gross profit | Taylor Holiday (CTC, 2022), "fuel profit" framing | Same status. |
| SaaS Magic Number 0.75 | Rory O'Driscoll (Scale VP) anecdote, ~2005; popularised by Lars Leckie, 2008 | Single-anecdote origin; Benchmarkit 2025 reports a 0.90 median. |
| "Spend 5-10% of revenue on marketing" | US Small Business Administration "general rule", ~2013 | Folklore; contradicted by the SBA's own measured ~1% actual small-business ad spend. The descriptive ~7.7% is Gartner measuring large-company behaviour. |
| 3x pipeline coverage | Unattributed sales folklore | Silently assumes a 33% win rate; needs ~4x at 25%, 4-7x for enterprise. |
| 3-12 month payback working band (18 at enterprise) | Practitioner-reported across paid acquisition, superseded above by Benchmarkit's measured ACV-segmented figures for B2B SaaS | Keep this row only for non-SaaS paid-acquisition businesses the segmented table doesn't cover; recalibrate against the business's own cohorts and cash position. |
| 2-3x ROAS as "good and achievable" (Google Ads) | Jyll Saskin Gales, ex-Google, "What is a Good ROAS in Google Ads?" | Stated heuristic explicitly conditioned on the business's own margin and LTV, not a study - she frames "good" as "profitable, more money out than in," never a fixed number. |
| LTV of at least $10K, and a $5K/month minimum budget, to justify LinkedIn Ads at all | AJ Wilcox (B2Linked), "How Much Do LinkedIn Ads Cost?" | Stated practitioner heuristic, not a measured benchmark - paired with his own observed CPC range ($8-14, up from $6-8 in 2020 and $2 in 2011) and the claim that LinkedIn's cost-per-SQL runs under half of Facebook's despite the higher front-end CPC. |
The circularity to watch for: a blog post quoting a _descriptive_ survey median (Gartner's 7.7%, Benchmarkit's 18-month payback) as a _prescriptive_ target. Descriptive averages describe what a panel did. They carry no claim about what this business should do.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |