When research cannot support a critical claim, HOLD is the honest result of the work, not proof that the work is unfinished. The proposed action stays blocked. The memo can still be complete: it tells its reader what was checked, which claim prevents the action, who owns the next check, and what result would reopen the decision. That is a deliverable, even when it is not the permission the sponsor wanted.
“More research needed” is a weaker handoff. It leaves the reader to guess whether the work is merely late, whether any action is safe in the meantime, and who has authority to change the answer. HOLD makes those boundaries part of the result. In this sense, HOLD is one of the decision dispositions in TRACE, the method in The Verified Research Loop for building a source-qualified, versioned decision record. It is an operating rule, not a claim that the method produces better outcomes.
What does HOLD decide?
TRACE keeps the evidence record and the decision record linked but separate. A relationship summary describes what the inspected evidence does for a claim: Corroborated, Partial, Unresolved, or Conflicting. A decision disposition states the action boundary. The eight canonical dispositions are Proceed, Conditional proceed, Pilot, Defer, Reject, Specialist review, Accepted risk, and HOLD.
Those labels answer different questions. A Partial claim may support a narrow test while failing to support a major commitment. An Unresolved claim may be peripheral to one choice and decisive for another. The status word does not settle either decision by itself. For a critical unresolved claim, the book’s default disposition is HOLD, and one option is off the table: an unqualified Proceed, because nothing in the record supports it. The owner must say which proposed action depends on that claim.
A research pass can also Continue under the same brief or Narrow through a new version. Those are routes for the work, not permission to act. A team can keep researching while the proposed commitment remains on HOLD. The TRACE framework shows where the decision memo sits in the full method.
What makes a HOLD different from an unfinished job?
The book’s conclusion draws the line in one sentence: “HOLD is an operating state with an owner and a release condition, which is what separates it from indecision hidden in process language.” A note that says only “we still need to check this” has neither an owner nor a release condition. A recorded HOLD gives the reader five things they can use now:
- A bounded decision. Name the act being considered, its scope and the person authorized to decide it. “Launch” is too broad if the live question is whether to sign one service contract for one workload.
- The blocker. List every critical claim ID and the precise gap. “Vendor risk” is a category; “the draft schedule does not cover every required data flow” is a checkable gap.
- The present boundary. State what is held and what narrower work, if any, may continue. Internal review is not the same act as signing.
- The release check. Say what evidence, completed test, or qualified review would let the owner reconsider. “Look into it” cannot be passed or failed.
- The next owner and trigger. Assign the check, set its review point, and say what happens if it fails or access remains closed.
The memo is complete as a record of the current evidence and action boundary. The underlying question may remain open. Do not imply that the research is done because the budget ran out; the book is plain that an expired budget does not make research complete. List the planned searches that were never run, the source bodies not obtained and any access failures. The handoff can be ready while the held act is not.
Which gap is critical enough to hold the decision?
Set the boundary before the inconvenient finding arrives. Chapter 3 of the book puts the critical HOLD condition in the research brief, beside the decision, scope, exclusions, owner, deadline and authority needed to act. This makes a later disposition inspectable. A reader can see whether the team applied its own stated threshold.
For each open claim, ask the question Chapter 6 uses to mark a claim decision-critical: could a material change in this answer alter the choice, its terms, its timing or the right to act? Rights, safety, hard-to-reverse harm, reach and dependent work also count, and there is no fixed number of critical claims. If the answer is no, record the gap without making it a blocker for this decision. If yes, link it to the act it could change. A planned exclusion does not neutralize a risk that could overturn the decision. Give that issue its own research leaf and owner, route it to a specialist, or keep the action on HOLD.
Dependencies matter too. Chapter 8 describes a market-entry decision with market, competitor, regulatory and people research leaves, each with its own claim IDs, specialist owner and disposition. Three complete leaves cannot release the parent memo when the fourth carries a critical HOLD. The parent memo should name that child claim and the dependent act. A count of finished leaves is not a vote.
How do you write the HOLD line?
Put the disposition near the top of the memo, before a polished synthesis can be mistaken for approval. Keep the underlying source and claim records linked. A short working block is enough for a small decision:
Memo ID/version: __________________ Evidence as of: __________
Decision and authorized owner: _______________________________
Disposition: HOLD
Held action and scope: _______________________________________
Blocking claim IDs and exact gaps: __________________________
Permitted work meanwhile, with limits: ______________________
Owner of each missing check: ________________________________
Release condition and observable completion test: __________
Next review/reopen trigger: _________________________________
If a check fails or access stays closed: ____________________The authorized decision owner and the people doing the missing checks may be different. Name both. If two critical claims block the act, give each its own ID, check owner and release condition. A generic “pending review” line makes it too easy to mistake one completed task for clearance of the whole decision.
The release condition is a test for reconsidering the disposition. It is not a timer that changes HOLD to Proceed. The owner still has to inspect the new record, its scope and any remaining risk before recording a new disposition.
Worked example: the service contract on HOLD
Chapter 7 of the book works through a composite memo (MEMO-VEN-01-v1, under BRIEF-VEN-01-v2), not a documented vendor case. A team must decide whether to sign a proposed service contract or defer. The vendor’s dated product documentation describes the required regional configuration. A draft contract schedule contains related language but is incomplete. Both are vendor-controlled. An internal test covers one nonproduction configuration and does not establish every proposed data flow.
The memo’s inference is narrow. The team has a plausible implementation route, but the current record does not establish a complete contractual and technical commitment for the named workload. The memo states no legal conclusion. Its action line can be read without guessing:
| Field | Recorded boundary in the composite (MEMO-VEN-01-v1) |
|---|---|
| Proposed decision | Sign the proposed service contract or defer. |
| Disposition and blocker | HOLD on signature. Blocking claim C-VEN-01-v1 (summary Partial in Chapter 6’s ledger): contract language and data-flow coverage incomplete. |
| Permitted use | Internal contract and technical review only. |
| Action and owners | Finalize the schedule and compare every required flow with the approved architecture; contract owner plus technical and privacy reviewers. |
| Completion test | A signed review record linking contract clauses, architecture version, exceptions and current product documentation. |
| Decision owner | Not named in the composite header. A live memo must name the person authorized to make the new decision. |
That last field matters. Finishing the reviewers’ task is not itself authority to sign. The book records only that the completed check may permit Conditional proceed; the evidence relation remains separately recorded. The authorized owner must review the result and any residual risk before issuing a new disposition.
Can a deadline or caveat lift the HOLD?
No. A deadline can end the current research pass. It cannot produce a missing source body, complete a technical test or give someone authority they lack. A review date can prompt the owner to inspect the record; it cannot stand in for the completion test. Chapter 3 is explicit: a deadline cannot silently turn HOLD into “Partial with caveat.”
A caveat is useful only when it changes how the claim is interpreted or what action is permitted. “Subject to verification” under an unconditional recommendation does neither. State the blocking claim and the action it prevents. Chapter 7’s review rule is equally direct: a change in tone cannot repair a failed check.
There may be a different authorized decision. A team might narrow the act, defer the commitment to a named trigger, reject it, or send the issue for specialist review. Record that new scope and disposition rather than treating the original hold as if it vanished. Accepted risk is another possible disposition only when an authorized owner can take it. It records the owner, date, reason, affected claim versions, bounded use, safeguards, expiry and review trigger. It does not turn weak evidence into strong evidence. In the contract composite, that route exists only where the unresolved issue is legally waivable, the owner has delegated authority to accept it, and the safeguards and expiry do not bypass a mandatory review or duty. A real team must establish those conditions for its own case.
How does a HOLD end?
Run the named check against the blocking claim. Record the new source or review, its version, scope and date. If it addresses only part of the gap, leave the remainder open. If it conflicts with an earlier body, keep the conflict visible. If access stays closed, record that failure rather than smoothing it into a caveat.
Then update the claim record and memo version. The decision owner reviews the new evidence state, the proposed scope, the remaining risk and their authority. They may keep HOLD, choose a narrower disposition, or authorize a different action if the record supports it. If an earlier memo was distributed, name who needs the corrected state. A release condition is useful because it says exactly what can reopen the decision, not because it predicts the answer.
This method makes the record inspectable; it cannot prove that every relevant source was found or that the final choice is correct. Corroboration is not a source count explains why several agreeing documents may still share one origin. What is TRACE? covers the full research chain. This page concerns the disposition at its handoff.
Try this today (15 minutes)
Choose one open decision you are researching. Write the proposed act and its authorized owner in one sentence. Choose its current disposition from the eight named above; “nearly ready” is not one of them. List the claims the act depends on and mark any critical claim that the present record cannot resolve.
If there is a critical unresolved claim, write one HOLD line with its ID, the action it blocks, the check owner, an observable release condition and what work may proceed meanwhile. Add the date or event that will bring the record back to the decision owner. If several claims block the act, list every ID and the check for each.
Give the line, without the rest of your research, to a colleague. Ask what they would stop today and what completed check they would need before reopening the choice. Working alone, put the source notes aside and answer the same questions from the line. If either answer is unclear, repair it. That gives you a usable handoff today; it does not clear the held action.
Cite this:HOLD is a research result, not an unfinished job.Len P. van der Hof. https://lenvanderhof.com/en/blog/hold-is-a-research-result/ ·