Source: frameworks/kit-advisory-onboarding/04-advisory-onboarding-quality.md

04 — QUALITY: Advisory Onboarding Kit

When to run: After every Create (Mode 1) before delivering to the practice owner. Re-run after any Mode 2 update.

Format: Pass/fail with blocking failures. Certain errors are disqualifying regardless of overall quality. Warnings should be fixed if time permits but do not block delivery.

Blocking Failures

If ANY of the following is true, the output cannot ship until fixed. Do not deliver.

Input integrity

[ ] Solo recording or written submission used as input — the kit requires a live interview (file 06). Patterns-input is insufficient. Failure mode: paragraph-form output, no step ordering, mixed real-vs-aspirational. Fix: run the interview, then re-run the kit.
[ ] Stale practice reference data — last updated more than 30 days before this run, OR contains [TBD] / [unknown] entries that the kit would have used. Fix: pause kit run, update the reference data file, then resume.
[ ] Trigger not locked — the output describes post-trigger steps without naming the trigger condition (signed agreement + payment received, by default). Fix: confirm the trigger with the practice owner before producing roadmap content.
[ ] Completion gate missing — the roadmap has no defined endpoint. The advisor and client cannot agree when onboarding is done. Fix: define criteria-based gate (e.g., "technical onboarding done + seasonal meetings booked + mutual responsibility signed") before delivery.

Content integrity

[ ] Special-sauce leak — any element of the kit's spine REQUIRES proprietary content from a specific practice to function. Examples: roadmap structure depends on one practice's proprietary tax strategy; package architecture depends on another practice's specific service-table layout. Fix: generalize the spine; move the proprietary content to a variant noted with attribution in cohort-internal documentation only.
[ ] Fabricated client quotes or examples — quotes, scenarios, dollar amounts, or client names that don't trace to the interview output or to the practice's reference data. Fix: replace with actual interview material or remove.
[ ] Transcript-misrendered proper nouns in client-facing output — "Adam" instead of ATOM, "Carbon" instead of Karbon, "Financial Sense" instead of Financial Cents, "Tax Dome" instead of TaxDome, "Smart Vault" instead of SmartVault, "Mike Mylons" or "Mike Michalowicz" instead of Mike Milan, "Kwang" / "Kuang" / "Wong Kwong" instead of Kwong. Cross-check against business-aos/cyp/tpc/tpc-reference-data.md. Fix: replace with the reference-data spelling.
[ ] Audience-boundary violation — advisor-internal content in the client-facing output. Examples: pricing detail not contractually agreed, advisor's success-rate self-assessment, failure modes from prior clients, advisor's emotional friction notes. Fix: move to advisor-internal section or remove.

Voice and language

[ ] AI patterns in client-facing prose — "It's not X — it's Y" constructions, "leverage" as a verb, "synergy," "circle back," "touch base," double-em-dash separators in client-facing language. Fix: rewrite per business-aos/reference/core/voice.md.
[ ] "Profit First" or any other proprietary methodology name used loosely — only use the name if the kit's output is genuinely Profit First. Otherwise, the practice's own named system. Fix: rename or remove.
[ ] Static-PDF artifact — the roadmap is delivered as a fillable PDF or static document the client cannot interact with or update. Violates pattern 2 (living document). Fix: produce HTML or a shared living document; the PDF can be an export, not the primary artifact.

Structural integrity

[ ] No first-win identified — the roadmap does not name a specific, dollar-measurable result within 30 days of the trigger. Fix: confirm the first-win mechanic with the practice owner; if they can't name one, the engagement isn't ready for advisory onboarding (route to a pre-onboarding diagnostic conversation instead).
[ ] No pre-scheduled cadence — the roadmap describes meeting cadence in the abstract ("seasonal," "quarterly") without specific months/dates booked at onboarding. Fix: pre-schedule the year before delivery.
[ ] No mutual-responsibility reference — the roadmap does not point to or include the mutual-responsibility doc. Fix: ensure the deliverable trio (roadmap + action plan + mutual responsibility) cross-references coherently.

Warnings (fix if time permits)

These do not block delivery but should be addressed before the next kit run.

[ ] Welcome experience format is generic — "welcome video" without specifying length, content, or delivery mechanism. The variant should be concrete enough that the advisor can produce it.
[ ] Package-design layer is implicit rather than named — the engagement includes implicit scope without an explicit "what's in / what's a separate project" boundary. Stronger for institutional practices; lean practices may legitimately have bundled scope.
[ ] First-win amount is unspecified — "reduce next estimated payment" without a dollar range or example. Acceptable if the advisor will fill in the number during the kickoff; flag if it's missing because the interview didn't produce one.
[ ] Cadence shape doesn't match the practice type — seasonal cadence for a coaching practice that needs activation-then-steady, or activation-then-steady for a tax practice that needs seasonal. Fix: confirm the cadence matches the practice's actual rhythm.
[ ] Confidentiality boundary not named — the roadmap or supporting artifacts don't specify what data stays in human-only channels (SSNs, EINs, account numbers) versus what can move through shared/AI-assisted tooling.

Common Failure Modes

Failure	What happens	How to fix
Mid-process correction during first run	Operator catches a gap mid-build, fixes the output by hand instead of letting the run finish. Kit doesn't learn; same gap recurs next run.	Let the run finish. Compare to golden (or to spec in file 03 placeholder). Document the gap. Fix the kit files (this is Mode 2), not just the output.
Treating solo recording as good-enough input	Output is paragraph-form, missing step ordering, mixes real and aspirational. Roadmap reads as wish-list rather than walked-through process.	Reject the solo input as insufficient. Route to file 06 interview protocol. Don't ship until live interview material exists.
Skipping the audience-boundary filter	Advisor-internal observations leak into client-facing artifact. Client reads it and feels managed, not served.	Re-run the audience-boundary check (file 01). Move internal content to advisor-only sections.
Locking the trigger as "signed agreement" without payment	Advisor delivers welcome experience to someone who hasn't actually paid yet. Buyer's remorse + half-engaged client.	Confirm trigger = signed AND paid. Re-check the practice's actual workflow — sometimes the order is signed-then-payment, sometimes reversed. The trigger condition is BOTH.
Building from a member whose practice is in transition	Output reflects what the practice WILL do, not what it DOES. The roadmap doesn't survive contact with the practice's actual operations.	Defer the build until the practice is past the transition (e.g., Meryl's project-based pivot — wait until the model is stable).
Forcing the lean spine onto an institutional practice (or vice versa)	The roadmap addresses scope-creep, team scripting, manager-tier delegation issues an institutional practice has, OR forces an institutional package-design layer onto a lean solo practice.	Identify the practice model BEFORE producing the roadmap. Use the matching variant. The spine elements are universal; the variants differ.
Special-sauce in the spine	The kit's universal elements end up needing one practice's proprietary content to function. The kit can't ship to other practices without leaking.	Run the special-sauce test (file 01). Generalize the spine. Move the proprietary content to a variant noted with attribution.

Pre-Delivery Checklist

Before handing the output to the practice owner:

[ ] All blocking failures cleared (above)
[ ] Warnings reviewed; addressed if possible, documented if deferred
[ ] Client perspective test passes for every section of the roadmap (see file 00 — Critical Rule)
[ ] Proper nouns cross-checked against tpc-reference-data.md and the member's per-practice reference data
[ ] The trio (roadmap + action plan template + mutual responsibility template) cross-references coherently — same trigger, same cadence, same first-win, same completion gate
[ ] Mode 2 candidates noted — any manual fixes made during this run get queued for kit-file updates
[ ] Practice owner confirms understanding of what they're receiving before the run finishes (see file 00 — Confirm Understanding Before Executing)