Evidence standards
The Playbook is built on the assumption that you are tired of vendor pitches, LinkedIn-thread statistics, and AI-generated figures that cite no source. The moment a chapter slips a fabricated or unverifiable claim past you, the Playbook has lost you. The standard below is how that does not happen.
Every statistic that supports a chapter claim traces to a named primary source, retrieved within the last twenty-four months where the topic is AI-adoption sensitive. Every named-company case meets the case study standards and is logged in the subscriber-library register with the source link visible.
The three source tiers
Evidence is accepted in three tiers, used in preference order.
Tier 1 — Primary research
Peer-reviewed studies, methodology-published research from reputable firms, government statistics, academic papers.
Examples the Playbook draws on: McKinsey QuantumBlack’s State of AI survey, Stanford HAI’s AI Index Report, Statistics South Africa, the US Bureau of Labor Statistics, OECD data, and peer-reviewed empirical pieces in Journal of Marketing Research and Harvard Business Review.
Tier 2 — Industry benchmarks
Published analyst research from reputable firms, platform-published usage data, industry-wide surveys with disclosed methodology.
Examples: Forrester Research reports, Gartner Hype Cycles and Magic Quadrants, the Salesforce State of Marketing annual survey, the HubSpot State of Marketing annual survey, Adobe digital trends surveys.
When sample size or methodology matters, the chapter notes it in prose: “Adobe’s 2024 study of 348 US marketers” beats “Adobe’s 2024 study”.
Tier 3 — Named case studies
Publicly documented company implementations with a retrievable primary source. Company blog posts announcing a specific implementation. Conference-talk transcripts or recordings. Press releases with specific metrics. Published interviews in trade press.
Every Tier 3 case the Playbook uses meets the case study standards and is logged in the subscriber-library register, with the source URL, the retrieval date, and a vendor-origination flag where the case was published by the tool’s vendor.
What the Playbook never cites
These source types are rejected. A chapter that uses one is sent back at review.
| Rejected | Why |
|---|---|
| Anonymous case studies (“a major retailer”, “a Fortune 500 firm”) | You discount anonymous cases, correctly |
| Vendor case studies with no disclosed methodology | Unverifiable; likely best-case selection |
| Twitter, LinkedIn, Medium, Substack posts without a primary-source link | Author-attested but not publisher-verified |
| ”Some studies suggest”, “research shows”, “experts agree” | Signals the writer does not have a specific source |
| AI-generated statistics | Every figure traces to a human-authored, dated source |
| Blog aggregators summarising other blog aggregators | Source lineage broken |
| Vendor-sponsored benchmarks where the vendor sells the product being benchmarked | Structural conflict of interest |
| Press releases older than five years for AI-adoption figures | Topic moves too fast; figures rot |
How to verify a statistic
Every bolded figure in the Playbook passes the same test, and you should be able to run it.
- Find the claim in prose. Bolded figures, percentages, ratios, time-savings, dollar figures, counts — anything that supports a chapter claim.
- Retrieve the primary source. The named author, organisation, and year in the prose lead you to the original publication. Not a blog summary of the publication; the publication itself.
- Confirm the figure. The exact percentage, year, and context should match what the chapter cites.
If a figure cannot be traced to a primary source within reasonable effort, the Playbook takes one of three routes: replacement source, mechanism-only rewrite without a specific number, or the claim is cut. “Cite a secondary source because the primary is not findable” is not an option.
How to verify a case study
Every case meets all of the rules below. The case study standards page is the public statement of the bar; the verified register sits in the subscriber library.
- Named company. Not “a major retailer”, not “a Fortune 500 SaaS firm”. Named.
- Named product, team, or campaign. Specific enough that you could find the case independently.
- Date window. “Q2 2024”, “mid-2023”, “2016 pilot to 2019 enterprise deal” — bounded.
- Specific metric with unit. “Reduced brief-to-draft cycle by 58%” beats “dramatically faster”. Comparator included where published.
- Retrievable source. Company blog, press release, conference talk, published interview. Not a vendor landing page that might be removed.
- Vendor-origination note. If the case was published by the vendor supplying the tool, the chapter says so in prose. You weigh vendor-originated cases differently and you are entitled to.
The Playbook never anonymises a case. Every named company case is sourced from publicly available information — company blog posts, press releases, conference talks, published interviews, regulatory filings — so the company is named and the source is retrievable, or a different case is used. Vendor case studies are never cited without acknowledging the vendor in prose. The headline figure from a vendor case study is never used as the expected outcome — Persado’s 450% lift on JPMorgan Chase ad copy (Persado press release, 2019 — vendor-published) is a ceiling, not a median, and the prose says so.
Statistic age
AI-tooling statistics age fast. The rules:
- Any figure older than twenty-four months at the time of writing has its year named explicitly in prose.
- Any figure older than thirty-six months is replaced with a fresher source where one exists.
- Any figure predating November 2022 (ChatGPT public launch) is treated as a different era and framed accordingly.
Sentences the Playbook never writes
If you ever read one of these in Prompt & Profit, that paragraph has failed the standard and the errata channel is the right place to flag it:
- “Studies show that…”
- “Research suggests…”
- “It’s widely believed that…”
- “Many marketers report…”
- “Industry experts agree…”
- “A major retailer…”
- “A Fortune 500 SaaS firm…”
The standard is to find the source or cut the claim.
What the reader holds the Playbook to
The standard, in three lines:
- Every bolded statistic in the Playbook has a named primary source, retrievable today.
- Every named case study clears the case study standards and carries a working primary-source URL in the register.
- Vendor-originated cases are flagged in prose. Best-case figures are described as ceilings, not medians.
If you cannot trace a claim to a primary source, the errata channel is open.