The Minimum-Viable Shape for an Anonymized Sanity Check Client Case Study

The question

What's the minimum-viable artifact shape (length, format, citation density) for an anonymized client case study that establishes credibility without burning the relationship? Framed for the Sanity Check editorial cadence — the founder needs 2-3 anonymized case studies by end of Q3 2026 to establish the data-team-vertical positioning, and anonymization discipline is non-negotiable because he works for a data consultancy (employer/client confidentiality boundary).

What we already know (from the vault)

• The case-study version is explicitly preferred over the synthetic version. [[2026-05-11-practitioner-newsletter-acquisition-tactics]] names "Mechanic 5: serialized field study" as the Phase-2 acquisition lever for the senior-operator persona, and its own follow-up flags that "the case-study version of Mechanic 5 is more credible than the synthetic version" — depth signals seriousness in a way a single hot-take cannot.
• The credibility/confidentiality split has a clean seam: methodology vs identifiers. [[2026-05-21-mammoth-client-field-study-feasibility]] establishes that the technical pattern (MAC 3x6 matrix, PRO-303-style dual-path override) is methodology the founder developed and can publish; column names, table names, BRD content, and client identity are confidential. The recommended posture is Path 2 — anonymized field study built from real engagement patterns, with one pre-clearance conversation with MG leadership (Dylan Cruise / Tom Clinton) unlocking real numbers instead of synthetic ones.
• RDCO has zero documented client successes under its own name yet. [[2026-04-15-commoncog-career-moats-confession]] applies Chin's "believability clock" / 3-successes-with-articulation test: the founder has one full career moat (MG tenure), but RDCO-as-practice has no published proof. The case studies ARE the believability instrument. Chin's "skin in the game" posture (publish the weaknesses, show the real work) is the credibility move, not polish.
• A concrete hand-tool already exists to anchor each study. [[2026-05-11-practitioner-newsletter-acquisition-tactics]] and the MAC matrix work mean each case study can end on a reusable 18-cell testing-matrix artifact the persona forks into their own workspace — the share event, and the proof that the methodology is real.
• Sanity Check has a vendor-neutrality / editorial standard worth holding the case studies to: content published primarily to promote a product is excluded except for exceptional depth (Ananth Packkildurai standard, adopted into SC editorial posture). A case study must read as a field study, not a brochure.

What the web says

• Length is "as long as it needs to be, no longer" — there is no 5,000-word floor. Practitioner consensus is a clear Problem → Solution → Outcome arc with paragraphs kept to 2-4 sentences for scanability (beomniscient.com). The widely-cited "5,000 words" figure is an upper-bound for complex enterprise studies, not a target. For a newsletter-embedded field study, 800-1,500 words is the credible-but-tight band.
• Lead with one specific quantified headline metric; support with 3-5 total. "Increased qualified inbound leads by 312% in 90 days" beats "increased leads significantly." Focus on the 3-5 most relevant metrics rather than dumping all data; structure as one headline metric + supporting metrics + business context (prospeo.io playbook via search).
• Decision-makers demand concrete proof: 87% of B2B buyers require concrete evidence of effectiveness, and case studies are a decisive trust factor in 64% of successful B2B purchases. Quantified outcomes drive reply rates more than any other element (search synthesis).
• Anonymization rule set (mask vs keep) is well-codified (equinetmedia.com): MASK = client name, specific location ("European OEM" not "Bristol-based"), niche service descriptors, granular job titles. KEEP = industry/sector, company-size prestige markers ("fast-growing Series B"), measurable metrics/percentages, regulatory/operational context, process-focused quotes. Explicitly stating "anonymized at client request" itself builds trust.
• "Enough context to feel authentic, not enough to identify the client" is the governing principle (1827marketing.com). Credibility comes from the precision of the environment description (regulatory pressure, growth stage, stack constraints), not from the brand name. Process and methodology belong entirely to the consultant and are fully publishable.
• For very strict NDAs, aggregate-language metrics ("across recent projects, adoption rose ~65% on average") and thematic/composite framings preserve proof while masking identity — but composites must be disclosed as composites to stay honest (uxresearchblog, 1827marketing.com).

Convergences and contradictions

• Strong convergence on the seam: both the vault (methodology-publishable, identifiers-confidential) and the web (mask name/location/niche-descriptors, keep industry/metrics/process) agree the publishable artifact splits at methodology-vs-identity. This is the load-bearing finding — anonymization does not require gutting the proof.
• Convergence on metric density: web says 3-5 quantified metrics with one headline; vault's preference for "case study with named numbers" over synthetic points the same way. The constraint is anonymization-safe numbers (deltas and ratios survive masking; raw absolute values that could fingerprint a client do not).
• Mild tension on length: generic B2B advice tolerates long-form (up to 5,000 words for sales collateral), but the Sanity Check surface is a newsletter for a senior persona who values depth-without-padding. The vault's editorial standard (no brochure, exceptional-depth-or-cut) resolves it toward tight: a field study, not a sales PDF. Serialization (a 3-4 part arc) is how to get depth without any single issue bloating.
• One real-world contradiction the founder must respect: per [[2026-05-21-mammoth-client-field-study-feasibility]], a persona-class reader can re-identify a thinly-anonymized client in ~60 seconds (Nutrafol is already publicly named by three vendors). So anonymization protection must be methodology-focused, not identity-focused — the safety comes from never publishing client-specific facts, not from a flimsy "a DTC brand" label.

Synthesis for RDCO

Recommended minimum-viable template — the "Sanity Check Field Study" shape, executable for all 2-3 Q3 case studies:

Target length 900-1,400 words (one focused newsletter issue, or one installment of a serialized arc). Five sections: (1) The Setup — anonymized client profile using prestige + environment markers only ("a Series B DTC brand running Magento + Snowflake + dbt, mid-2022 acquisition pressure"); 2-3 sentences, no name, no city, no niche product detail. (2) The Problem — the concrete diagnostic, with one anonymization-safe quantified gap as the hook (e.g. "14 existing schema tests vs the 95 the MAC matrix demanded"). (3) The Work — the methodology, written for full transparency because it's yours to publish (the 3x6 matrix, the dual-path override pattern, severity tiers). This is the longest section and the credibility engine. (4) The Outcome — 3-5 quantified results expressed as deltas/ratios/tier-counts, never raw absolutes that fingerprint the client. (5) The Takeaway + Tool — generalize to a Monday-morning lesson and link the downloadable matrix template (the share event from Mechanic 3).

Citation/metric density: exactly one headline metric up top, 3-5 supporting metrics total across the piece. Every metric must pass the re-identification test — if a number could only belong to one company, convert it to a ratio or a relative delta. Internally, footnote the provenance ("pattern observed across multiple analytics-engineering engagements; client identifiers and code references removed") — this is the honesty disclosure both Chin's skin-in-the-game posture and the web's "anonymized at client request" trust-builder call for.

Anonymization rules (the hard checklist, applied to every draft): MASK client name, location, niche product/service descriptors, granular titles, table/column/BRD names, any absolute number that fingerprints. KEEP industry/sector, company-size/stage prestige markers, regulatory/operational context, the full methodology, and process-focused quotes if any are clearable. Run a grep audit before publish (the MAC landing-page spec already blocks "Mammoth", "Progress", "Nutrafol", "gold_opp_pipeline", "silver_fct_*" — reuse that wordlist). The protection is that no published fact is client-specific, not that the label says "anonymous."

How it fits the cadence and the data-team-vertical positioning: gate quality, not quantity. Two genuinely-credible field studies beat three thin ones for the believability clock. Sequence them as a serialized arc (the 5-episode MAC field-study arc in [[2026-05-21-mammoth-client-field-study-feasibility]] is ready-made: 14-to-95 gap → PRO-303 pattern → severity tiers → cross-model reconciliation → harness adoption) so each issue compounds and creates subscribe-now urgency. The single highest-leverage unlock is the one pre-clearance conversation with Dylan Cruise or Tom Clinton in the founder's first 30 days at phData: it moves every study from Path 1 (synthetic, lower credibility floor) to Path 2 (real anonymized numbers), which is exactly the credibility difference the practitioner-newsletter brief identified. Until that conversation lands, the founder can still ship Path 1 synthetic-but-disclosed studies in the same template — the shape is identical; only the provenance footnote changes.

Open follow-ups

• Has the founder had the Dylan/Tom pre-clearance conversation yet, and can it be a single canonical "anonymization protocol" approval covering all future episodes rather than per-episode asks?
• Does the cc-wrapped harness IP-boundary settlement cover authored content ABOUT the harness (episode 5), or only the harness code? This gates the highest-credibility episode.
• Should Sanity Check publish a one-time standing "How we anonymize client work" editorial note, so every field study can reference it and inherit the trust signal without re-explaining?
• What absolute numbers from the MG engagements are safe to publish as ratios/deltas without re-identification — needs a concrete pass on the candidate metrics before drafting episode 1.
• Is there a non-MG engagement (or a phData-era one, once tenure permits) that could anonymize into a second, independent case study so the proof isn't single-client-dependent?

Related

• [[2026-05-21-mammoth-client-field-study-feasibility]]
• [[2026-05-11-practitioner-newsletter-acquisition-tactics]]
• [[2026-04-15-commoncog-career-moats-confession]]

Sources

Vault:

• [[~/rdco-vault/06-reference/research/2026-05-21-mammoth-client-field-study-feasibility.md]]
• [[~/rdco-vault/06-reference/research/2026-05-11-practitioner-newsletter-acquisition-tactics.md]]
• [[~/rdco-vault/06-reference/2026-04-15-commoncog-career-moats-confession.md]]

Web:

• Omniscient Digital — Writing high-converting B2B case studies: https://beomniscient.com/blog/writing-high-converting-b2b-case-studies/
• Equinet Media — How to write a case study without naming the client: https://www.equinetmedia.com/blog/how-to-write-a-case-study-anonymous-client
• 1827 Marketing — Showcasing success without naming names: https://1827marketing.com/smart-thinking/client-confidential-how-to-showcase-success-without-naming-names/
• UX Research Blog — Case studies when work is under NDA: https://www.uxresearchblog.com/post/how-to-write-ux-research-case-studies-when-work-is-under-nda
• Prospeo — How to use social proof in B2B (2026 playbook): https://prospeo.io/s/how-to-use-social-proof-in-b2b