Editorial Standards

Voice

Target: understated, technical, direct. The reader already runs a business, side project, or micro-SaaS. We give them the numbers, the source, and the limits — not pep talks or venture-twitter aphorisms.

• Understated, technical, direct — solo-founder to solo-founder. Numbers first, no hype.
• No corporate jargon (`leverage`, `empower`, `seamless`, `solutions for`, `industry-leading`, `synergy`).
• No AI slop (`delve`, `tapestry`, `robust` unless fault-tolerant, `unlock`, `navigate these topics`, `in today's landscape`).
• No hustle-culture vocabulary (`10x your`, `hustle`, `grind`, `crush it`, `side hustle` — use `side project`, `supercharge`, `effortlessly`, `ninja`, `rockstar`).
• No prescriptive framing — we run the numbers, the reader makes the call.
• No personal launch logs, MRR screenshots, or venture-twitter voice.

Claims — what we can and can't say

• Never publish invented numbers as executed measurements. Every benchmark links to a named source and year.
• Never claim a cadence we cannot commit to. No 'weekly cadence', 'coming soon', 'new tools every month'.
• Cut unsupportable claims rather than hedge them. The urban legend '90% of A/B tests yield inconclusive results' has no peer-reviewed source; we do not publish it with soft language. The misquoted Reichheld '5% retention = 25–95% profit boost' is context-dependent; it is replaced with specific, sourced claims.
• Hedge honestly where warranted: `roughly`, `as of 2026`, `the OpenView 2024 survey of 1,000 SaaS operators suggests`.
• No 'peer-reviewed' labeling on surveys, consulting reports, or vendor telemetry.

Source hierarchy

Enforced by scripts/validate-content.mjs in requirePrimarySource mode. Vendor-blog domains fail CI. Branded reports without verifiable URLs warn.

• 1. Primary academic

  DOI, arXiv, SSRN, PubMed, peer-reviewed journal URLs — preferred for any 'research shows' claim.

• 2. Authoritative primary

  BLS OEWS/ECEC, Fed FRED, SEC EDGAR, ECB, SBA, Eurostat, UK ONS, national statistical offices. Live, verifiable URLs.

• 3. Verifiable branded reports

  First Round State of Startups, OpenView SaaS Benchmarks, SaaS Capital Annual Survey, Bessemer Cloud Index, CB Insights — only where the specific report is publicly linked.

• 4. Textbooks with chapter/section citation

  Named chapter URLs from Wiley, O'Reilly, and equivalent.

• 5. Founder case studies with named authorship

  Indie Hackers profiles, founder blog posts with quoted metrics, Product Hunt launches with verifiable traction.

• Never

  Vendor-marketing blogs (VWO, Optimizely, HubSpot blog, Mailchimp resources, WordStream), paywalled HBR without a summary we can cite, Forbes contributor posts, 'McKinsey 2024 insights' without a URL, unlinked 'studies show'.

Meta descriptions + titles

Meta descriptions target 140–155 characters with the primary keyword in the first 60 characters. No `no signup / no tracking` tails — they burn SERP real estate without keyword value. Article titles are ≤60 characters (aim ≤55). Enforced by scripts/audit-meta-descriptions.mjs.

Compliance

Calculators compute deterministic math from user inputs. Output is general business information, not legal, tax, or financial advice. Consult a licensed professional for your specific situation. No personalised output; nothing profiles the reader. Pricing data in the AI Stack Cost Calculator carries an AS_OF_DATE constant and surfaces a refresh warning when stale beyond 90 days. Benchmarks older than 18 months are flagged as dated.

Corrections

Dated and append-only at /corrections/. Never silently rewritten. When a benchmark re-sources, a pricing number drifts, or a claim is cut, a dated entry is added to the corrections log.

What doesn't ship

• Personal launch diaries, founder-porn anecdotes, vanity-metric celebrations.
• Urban-legend statistics (the 90%-of-A/B-tests claim, the misquoted Reichheld retention-profit claim, 'users form habits in 21 days').
• Tautological claims ('improvement in X correlates with improvement in X') or circular causation.
• 'Peer-reviewed' labeling on surveys, consulting reports, or vendor telemetry.
• Sponsored content without the `Sponsored` / `Affiliate` chip rendered inline.
• Any UI claim we cannot defend on an honest audit. Boring-but-true over punchy-but-risky.
• Tools without a methodology page linked from them.

Conflict of interest

AI Biz Hub is a self-funded, independent publication. There are no active sponsors, affiliate links, or paid placements. The current state and the rules that apply if that ever changes are documented at /sponsor-disclosure/.

Contact

Factual corrections, methodology disputes, and sponsor inquiries: see the contact details at /about/.