Richard Smith - Virginia

Self-directed playbook

Institutional Knowledge Structuring

Which decisions become fragile when someone leaves?

Institutional knowledge represents a major source of trapped value.

Here, trapped value means decision-changing rationale, exceptions, authority, confidence, ownership, and context that exist but cannot yet be reliably found, interpreted, trusted, or reused when a recurring decision must be made.

This is not a quantified financial claim. The goal is not more documentation, and not all expert judgment can or should be codified.

  1. 01Held contextWhat a person or source knows
  2. 02Decision recordThe minimum context that can travel
  3. 03Retrieval testWhether another person can use it
  4. 04Accountable reuseOwnership, review, and a responsible stop

Where value gets trapped

A document can exist while the decision basis remains out of reach.

Investigate the path from a known source to a responsible choice. Do not treat a symptom as proof that a person, team, or repository has failed.

  1. Context is held

    A rule, rationale, exception, or limit lives with an expert, in a source, or across both.

  2. A decision reoccurs

    Someone needs the context at a particular moment, under a particular constraint.

  3. Retrieval breaks

    The source may be findable while its authority, meaning, currency, or exception path is not.

  4. Work depends on rescue

    A knowledgeable person must reconstruct what the record did not carry.

Signals worth testing

  • People repeatedly ask who knows the real rule.
  • A source is easy to find but difficult to interpret.
  • Two experienced people describe different exception paths.
  • A workaround is trusted more than the official source.
  • A decision pauses when one person is unavailable.

Use test

Start with one decision, not a whole knowledge system.

Pick a decision that comes up regularly and matters. Could a capable colleague find what they need and make that decision without asking someone to talk them through it?

Use this when

The same decision keeps coming up, getting it right matters, and the know-how sits with a few people or is scattered across different sources.

Skip or stop when

The decision is rare or low stakes, people can already find what they need, or asking an expert directly is deliberately the safest approach.

Protect

Keep sight of who the decision affects, the outcome it should protect, experts' time, and the signs that a case needs judgment rather than a rule.

Constrain

Be clear about the scope, who can see the information, how long to keep it, who has authority, whose consent is needed, and what should never become a reusable rule.

Smallest useful move

Run one uncoached retrieval test.

Test the existing path before creating a repository, conducting a broad knowledge inventory, or writing a new playbook.

  1. 1

    Select one decision

    Name a recurring choice with a real trigger, consequence, and decision owner.

  2. 2

    Ask a capable colleague

    Have someone who understands the work retrieve and apply its basis without coaching.

  3. 3

    Record the first break

    Capture what they tried, what they found, and the first point where responsible use became uncertain.

  4. 4

    Choose the smallest response

    Repair only the observed gap, then run the same retrieval attempt again.

Failure taxonomy

Where did retrieval first stop being dependable?

Finding
The relevant source cannot be located.
Authority
No one can tell which source governs.
Rationale
The rule is visible, but its protected goal is not.
Exception
The colleague cannot recognize when the rule does not fit.
Judgment
The point where an expert must decide is hidden.
Confidence
The reliability or limitation of the source is unclear.
Ownership
No one is accountable for the decision basis.
Currency
There is no credible review date or reopening trigger.
Escalation
The colleague cannot identify where a disputed case goes.

Stop boundary: if the colleague can find, interpret, and responsibly apply the basis, including its limits and owner, do not create another artifact merely to complete the method.

Observe → Structure → Reinforce

Six moves from held context to accountable reuse.

Move in order, but keep the intervention as small as the observed failure allows. Every move leaves evidence another person can inspect.

  1. Observe01

    Select the decision

    Question
    Which reocurring choice becomes fragile without private help?
    Evidence
    A real trigger, consequence, and accountable decider.
    Leave-behind
    One bounded decision statement.
  2. Observe02

    Watch retrieval

    Question
    What does a capable colleague do without coaching?
    Evidence
    The path tried, sources opened, questions asked, and first break.
    Leave-behind
    A retrieval observation, not an assumed problem.
  3. Structure03

    Trace the decision basis

    Question
    Which source, rationale, authority, and context actually change the choice?
    Evidence
    References to governing material and the people able to explain it.
    Leave-behind
    A traceable basis with provenance.
  4. Structure04

    Separate rule from judgment

    Question
    What can travel, and where must expert interpretation remain?
    Evidence
    Known exceptions, limits, uncertainty, and escalation conditions.
    Leave-behind
    An explicit judgment boundary.
  5. Reinforce05

    Validate and assign ownership

    Question
    Who can validate this basis and remain accountable for it?
    Evidence
    Expert review, confidence, limitations, owner, and escalation route.
    Leave-behind
    A bounded Decision Retrieval Record.
  6. Reinforce06

    Retest and set review

    Question
    Can another person now act responsibly, and what should reopen the record?
    Evidence
    A second uncoached attempt plus a named event, contradiction, or review point.
    Leave-behind
    A readiness decision and review trigger.

Worked example

Decision Retrieval Record

This fictional regional workflow exception shows the minimum context another person needs. It is an illustration, not a client result or proof that the underlying material is correct.

Illustrative record · Regional workflow exception

Status: needs validation

01 · Decision, trigger, people
Decide whether to approve an alternate regional workflow when the standard route conflicts with a documented local constraint. Affects the requester, regional operator, reviewer, and decision owner.
02 · Authoritative source
Candidate source: the current workflow standard and regional addendum. Governing authority remains to be confirmed.
03 · Rationale and protected goal
Preserve the control intended by the standard workflow while allowing a legitimate regional constraint to be handled explicitly.
04 · Known exceptions and limits
The alternate route applies only to the named constraint. It does not create a general regional exemption or override another required review.
05 · Expert-judgment boundary
An expert must assess unfamiliar constraints, conflicting obligations, or cases whose effect cannot be compared with the standard route.
06 · Source, confidence, validation
Source expert: regional operations lead. Confidence: provisional until the workflow owner confirms authority and the exception criteria.
07 · Owner and escalation
Accountable owner: workflow owner. Escalate disputed authority or novel conditions to the designated policy reviewer before approval.
08 · Trade-off and uncertainty
The alternate route may protect the regional constraint while adding review work. It remains uncertain whether the constraint applies to this case.
09 · Last review and reopening trigger
Last review: not yet validated. Reopen when the standard, regional constraint, owner, or observed result changes—or when evidence contradicts the record.
10 · Intentionally excluded
General regional history, biographies, and background that do not change this decision. Exclusion keeps the record bounded, not comprehensive.
A source reference does not prove accuracy, completeness, authority, or currency. Validate the decision basis with the accountable source expert.

Choose what happens next

What does the evidence support now?

These four outcomes are equal options—not steps up a ladder. Choose the one the evidence supports, name who acts next, and record what would prompt another review.

01

Ready to rely on

When: people can retrieve it successfully, and there is credible support for its basis, limits, authority, owner, and expert validation.

Next: use it and name the review trigger. The owner reviews it if something changes or evidence contradicts it.

02

Needs validation

When: there is still disagreement about authority, known exceptions are missing, or the evidence does not yet support confidence.

Next: assign the smallest check needed. The owner reviews the outcome after an expert or source check.

03

Keep expert judgment

When: the work is rare, high-stakes, variable, or unsafe to simplify beyond a clear route to an expert.

Next: make the expert’s role in the decision clear. Review this choice if the work becomes more frequent, ownership changes, or new evidence appears.

04

Retire or replace

When: the material is out of date, duplicated elsewhere, contradicted by evidence, or no longer helps with a current decision.

Next: stop relying on it or link to its replacement. Review it again only if the decision or obligation returns.

Guardrails

Make context portable without pretending people are replaceable.

Knowledge holders are collaborators. Agree how their contribution will be represented, who may use it, and where their judgment remains necessary.

Consent and attribution
Agree what may be captured and how the holder or source will be credited.
Confidentiality and access
Keep restricted, personal, regulated, or contractual context within its permitted audience.
Retention and authority
Do not preserve material beyond its allowed life or present an informal workaround as governing policy.
Maintenance and roles
Name who owns the basis, who validates it, who decides, and what event requires review.
Expert judgment
Show where interpretation begins; a record should route judgment, not conceal or impersonate it.
AI assistance
AI may organize, summarize, or draft validated material. It must not invent workflows, rules, authority, assumptions, or exceptions, and it cannot replace subject-matter review.

Continue independently

Test one decision before you design a knowledge program.

Choose a real reoccurring decision, ask a capable colleague to retrieve its basis without coaching, and record the first break—or stop when the path already works.

Return to the retrieval test Browse the independent library