How the AUG v3 framework was calibrated

The AUG v3 framework didn't emerge from a whiteboard. It was developed by auditing a 30+ SaaS portfolio over 18 months, identifying recurring failure-modes, and iterating against real fleet data. This page documents the evidence base — what got included, what got rejected, and which industry research informed each of the 7 factors.

The evidence base — 30+ SaaS portfolio

The framework was calibrated against operator Paulo de Vries's SaaS portfolio: a 30+ live product fleet spanning finance, education, fermentation, web-vitals tooling, book content, and developer tools. Every factor in AUG v3 traces back to a real failure-mode somewhere in that fleet.

Real fleet audits available: /case-studies/ shows 5 anonymized real audits. The same audit ran against 10 active fleet sites on 2026-05-16, revealing that 100% share the same blocker (Acquisition + Monetization both at 1) — exactly what the funnel-order-discipline principle predicts.

Industry research per factor

Acquisition (Factor 1)

• SEMrush 2025 SaaS traffic study: sites under 1,000 sessions/mo have <5% probability of crossing 10k UV/mo within 12 months. Anchors the bottom-tier scoring at <500/mo = 1.
• Aleyda Solis LLM-citation research: the 10-characteristic framework documented at /method/llm-citation/. Reframes Acquisition to include GEO/AEO not just SEO.
• Brian Balfour Growth Loops: compound mechanisms over linear funnels. AUG's multiplicative composite operationalizes the loop concept.

Activation (Factor 2)

• Reforge / Casey Winters research: activation rate below 25% rarely improves; above 50% rarely regresses. Anchors the rubric scoring tiers.
• Mixpanel 2024 cohort report: consumer SaaS activation needs to happen within first 7 days or D30 retention drops 70%. Hence the “first 10 seconds” emphasis.
• Drift conversational-commerce research: blocking signup before value-delivered cuts activation 40-60%. Built into the rubric's Activation anti-patterns.

Engagement (Factor 3)

• Google Web Almanac 2024: median bounce rate 55%; healthy SaaS target <45%; below 30% is top-decile. Anchors the 4-component composite scoring.
• Microsoft Clarity heatmap research: rage clicks correlate 0.78 with activation drop in the next 7 days. Why session-replay rotation is in the weekly cadence.
• Fleet measurement: HoldLens.com's 7% bounce rate + 3.88 PV/session is top-1% — confirms the “bounce <30%, 3.5+ PV/session = 10” scoring tier is reachable, not aspirational.

Retention (Factor 4)

• Mixpanel 2024 retention benchmarks: consumer SaaS D7 <12% rarely compounds; ≥25% indicates real product-market fit. Anchors the rubric.
• Sean Ellis PMF survey research: 40% “very disappointed” threshold. AUG v3 v3 vs PMF survey comparison at/compare/aug-vs-pmf-survey/.
• Fleet calibration: 8 of 10 audited fleet sites lack D7 cohort tracking — proves the measurement-stack mandate at/method/measurement/ is necessary infrastructure, not optional polish.

Advocacy (Factor 5)

• NFX Network Effects research: k-factor below 0.15 = subcritical viral; ≥0.4 = supercritical for consumer SaaS. Anchors the rubric.
• 2025 LLM-citation referral data: chatgpt.com / claude.ai / perplexity.ai referrals convert at 8-18% CTR vs 2-5% for organic SERP #3. Why LLM-citation is its own Advocacy compound, not Acquisition.
• Fleet measurement: bookpop.app has TikTok 2,924 followers + 100K-view videos but only 8 UV/30d on site — proves Advocacy and Acquisition are independent dimensions. High follower count does NOT mean high traffic unless the funnel is wired.

Monetization (Factor 6)

• AdSense pRPM benchmarks 2025: Business/Finance $25-45, Education $10-18, Food $12-25. Calibrates the math floor formula.
• ProfitWell SaaS pricing research: free→paid conversion ≥3-5% for consumer; ≥30% for sales-assisted B2B. Anchors the rubric tiers.
• Cloudflare Pay-Per-Crawl beta data (2026): $0.001-0.10 marketplace pricing; ~20% of crawls actually bill. Honest projections for bot-revenue compound at /method/monetization/.
• Funnel-order discipline (the math floor): below ~1,000 sessions/mo, no monetization tweak beats thesessions × PV × RPM ÷ 1000 ceiling. Why this is enforced in/method/funnel-order/.

Performance (Factor 7)

• Google Page Experience research: every 100ms of LCP costs ~1% conversion. Anchors the rubric scoring.
• Cloudflare Web Vitals dataset (2024): mobile p75 LCP <1.5s is achievable with static export + edge deployment. Anchors the “excellent” tier.
• Fleet measurement: 10 of 10 audited fleet sites have Performance ≥6 (all on Cloudflare edge). Performance is the EASIEST factor to max out — and fixing it does NOT compensate for weak Acquisition.

What got REJECTED from the framework

Net Promoter Score (NPS)

Tried in v1. Removed in v2. NPS responses are heavily biased toward extremes (loud promoters + loud detractors); silent middle invisible. Replaced with measurable D7/D30 cohort return rate which captures the same signal without the survey bias.

CAC / LTV ratio

Tried in v2. Removed in v3. For zero-marketing-budget solo founders, CAC is essentially infinite or zero depending on attribution; the ratio is unmeasurable. Replaced with math-floor revenue ceiling which is calculable from observable sessions × PV × RPM.

Stickiness (DAU/MAU)

Considered in v3 design phase. Rejected because the metric is meaningful for daily-habit SaaS (Slack, Notion) but misleading for occasional-use tools (calculators, references). The Engagement + Retention split captures the same signal more honestly per archetype.

What the framework is opinionated about

• Multiplicative composite: a zero in any factor near-zeros the whole. No additive cheating. See scoring transparency.
• Funnel-order discipline: Acquisition must compound before Monetization optimization is worth operator time. See/method/funnel-order/.
• Dark-pattern rejection: 12 retention hacks hard-rejected because they spike short-term but collapse long-term. See/method/dark-patterns/.
• Performance as growth multiplier, not technical concern: Perf compounds into every other factor. Why it's Factor 7 not optional.

What the framework is NOT opinionated about

• Which channel to acquire from (SEO vs paid vs social vs referral — depends on archetype)
• Which monetization stack (subscription vs ads vs affiliate vs commerce — depends on archetype)
• Which tech stack (static export vs SSR vs full SPA — depends on content type)
• Which design language (every archetype has different conversion-optimal patterns)

Where to dig deeper

• Full 7-factor methodology hub
• Scoring transparency — exact composite formula
• Funnel-order discipline (math floor)
• 12 dark patterns AUG hard-rejects
• LLM-citation 10-characteristic framework
• Day-1 analytics stack (6 trackers)
• 5 real anonymized fleet audits
• Public iteration log (RSS) — every framework change documented