A frozen query sample is adequate only when it passes DECIDE-6: decision fit, essential material-stratum coverage, controlled observation comparability, information gain, disclosed residual gaps and exact replayability. The result is not a score. It is one release state: Accept, Qualify, Expand or Refuse.
Status: Founder reviewed by Pero Baldjiev. Founder reviewed by Pero Baldjiev.
DECIDE-6: the adequacy gate
STRATA-6 builds, tests and freezes a defensible sample. DECIDE-6 is the parent framework’s six-control decision gate for determining what that frozen sample may support.
DECIDE-6 decision path
From a frozen sample to permitted wording
- Frozen STRATA-6 sample
The sample, exclusions, observation settings and evidence record are fixed before adequacy is judged.
- Decision and consequence
Name the exact decision the sample must support and the consequence of being wrong.
- DECIDE-6 binding controls
- D
- Decision fit
- E
- Essential-stratum coverage
- C
- Controlled comparability
- I
- Information gain
- D
- Disclosed residual gaps
- E
- Exact replayability
- Release-state hierarchy
- AcceptRelease the decision.
- QualifyNarrow the wording.
- ExpandAdd missing evidence.
- RefuseStop the requested claim.
- Permitted wording
The first binding failure controls what may be claimed, qualified, expanded or refused.
- Released decision and reopen trigger
Record the decision, its limits, accountable owner and the change that would force reassessment.
Adequacy is decision-relative. The same sample may be adequate for exploratory mapping and inadequate for a funding recommendation.
DECIDE-6 binding adequacy controls
DECIDE-6 is a mnemonic layer over the six formal controls. The controls remain conjunctive: a strength in one cannot average away a binding failure in another.
| Control | Pass condition | Failure effect |
|---|---|---|
| Decision fitDecision fit | Every sampled query maps to the stated decision, and unsupported adjacent decisions are named. | Refuse the requested conclusion or narrow its scope. |
| Essential coverageMaterial-stratum coverage | Every buyer, intent, stage or market cell capable of reversing the decision is represented or explicitly excluded. | Expand the sample before release. |
| Controlled comparabilityObservation comparability | Geography, language, device, date and result environment are controlled closely enough for comparison. | Refuse cross-query or cross-market inference. |
| Information gainInformation gain | New queries produce diminishing decision-relevant change within the represented strata. | Continue sampling; saturation is not demonstrated. |
| Disclosed gapsResidual-gap exposure | Known omissions and unknowns are visible, and their ability to reverse the decision is assessed. | Qualify, expand or refuse according to materiality. |
| Exact replayReplayability | The exact sample, sources, selection rules, exclusions, settings and version can be reconstructed. | Refuse any reproducibility or audit claim. |
How to read the original scorecard: the scorecard remains the article’s practical wide diagnostic image. DECIDE-6 is the governing semantic text model: a total score cannot average away a binding failure, and only Accept, Qualify, Expand or Refuse controls the permitted wording.
Decision consequence changes the adequacy threshold
Adequacy becomes stricter as the cost of being wrong increases. The sample must be judged against the consequence of the next action, not against an arbitrary keyword count.
| Decision type | Minimum acceptable state | What remains prohibited |
|---|---|---|
| Exploratory mapping | Qualify or Accept | Market-wide claims, definitive competitor rankings or budget commitments. |
| Comparative recommendation | Accept, or a narrowly bounded Qualify | Declaring a definitive winner while a material stratum remains unsampled. |
| Funding or execution decision | Accept | Allocating spend, ownership or sequencing when any binding control fails. |
Composite failure scenario: a large sample that still fails
Evidence boundary: this is a fictional composite designed to demonstrate the framework. It is not a client case or a claim about observed commercial outcomes.
A payroll-software team freezes 500 queries and concludes that its competitor set is stable. Most queries describe basic payroll administration. Migration, compliance, integration and switching questions are barely represented.
| Control point | What the team believed | What the review found | Adequacy effect |
|---|---|---|---|
| Decision fit | The list represented the buying market. | Most queries described early information seeking rather than provider selection. | Qualify the market claim. |
| Material coverage | Five hundred queries meant broad coverage. | Migration, compliance and integration strata were materially absent. | Expand the sample. |
| Information gain | New exports had stopped adding value. | The same tool kept returning near-duplicate informational variants. | False saturation; continue through new sources. |
| Decision outcome | The competitor set could be released. | Specialist providers appeared only after the missing commercial strata were sampled. | Move from Accept to Expand. |
Applied takeaway: a large sample can fail because it repeats the same view of the market. Adequacy depends on material coverage and decision consequence, not volume alone.
Where the framework loses validity
| Failure mode | Why it matters | Required response |
|---|---|---|
| Tool-export convenience | The first exported list silently becomes the market definition. | Return to STRATA-6 and diversify candidate sources. |
| Head-term dominance | High-volume terms crowd out commercially decisive long-tail, local or specialist queries. | Expand the missing material strata. |
| False saturation | Near-duplicate additions create the appearance that no new information remains. | Change source, language or stratum before claiming diminishing information gain. |
| Intent collapse | Different buying stages are compared as if they supported the same decision. | Separate the decision cells and re-evaluate coverage. |
| Silent coverage gaps | Missing locations, devices or result environments disappear from the conclusion. | Qualify, expand or refuse according to their reversal risk. |
What this framework owns—and what STRATA-6 owns
STRATA-6 and DECIDE-6 are complementary but non-interchangeable. STRATA-6 constructs and freezes the evidence object; DECIDE-6 evaluates whether that object is sufficient for the stated decision.
Scoping, sourcing, resolving duplicates, assembling, weighting, testing and freezing the query sample itself.
Source reliability, evidence freshness, claim wording, exclusions and the reproducibility of the wider analysis.
Founder-origin note: this framework was developed to separate sample construction from permission to make a decision claim. That is a first-party methodology statement, not external validation or evidence of commercial outcomes.
Query Sample Adequacy Radar
How to read it: select a checkpoint to isolate that part of the method. The canvas shows the relationship; the inspector explains what the checkpoint changes in the decision.
Apply the DECIDE-6 adequacy gate
| Stage | Control question | Required output | Stop rule |
|---|---|---|---|
| 1 | What decision will this sample support, and what is the consequence of error? | A bounded decision statement and consequence level. | Refuse if the decision remains undefined. |
| 2 | Which frozen sample version, exclusions and observation settings are being evaluated? | A versioned evidence packet. | Refuse if the sample cannot be replayed. |
| 3 | Which of the six binding controls pass, fail or remain unresolved? | A control-by-control adequacy record. | No average score may override a binding failure. |
| 4 | Which release state follows from the first material failure? | Accept, Qualify, Expand or Refuse. | Do not release stronger wording than the state permits. |
| 5 | What evidence would invalidate this decision later? | Permitted wording, owner, review date and reopen trigger. | No release without an accountable owner. |
Messy reality: conflicting evidence, shared ownership or reviewer disagreement
Run the controls in order, stop at the first binding failure and carry the resulting release boundary into every downstream claim.
When the adequacy decision must be reopened
- The decision moves from exploration to recommendation, funding or execution.
- New first-party language exposes a missing high-value intent or buyer stratum.
- A location, device, language or result-type change alters the observed competitor set.
- New queries continue changing roles, territories or confidence after saturation was declared.
- The sample version, selection rules or observation environment can no longer be replayed.
- A residual gap previously judged non-material becomes capable of reversing the action.
Retain the prior release state, the changed evidence, the reviewer, the date and the decision affected. Reopening is part of the framework, not evidence that the original review failed.
Adding more queries can reduce adequacy when the additions amplify the same source bias while leaving the material gap untouched.
Decision-level conclusions
- Purpose: apply DECIDE-6 to evaluate whether a frozen query sample is sufficient for one stated decision.
- Threshold: decision consequence determines how much unresolved risk can be tolerated.
- Release: the output is Accept, Qualify, Expand or Refuse—not a universal score.
- Boundary: one binding failure cannot be averaged away by strengths elsewhere.
- Review: reopen the decision when material conditions, evidence or consequence change.
A query sample is adequate only for the decision it can support without hiding a material gap.
Adequacy does not prove that all search demand has been observed and does not predict rankings, traffic, leads, revenue or commercial outcomes.
Reviewer questions
Is there a universal minimum number of queries?
No. Quantity follows the coverage requirement and the consequence of the decision. A smaller stratified sample may be adequate while a larger convenience list fails.
Can a sample pass with known gaps?
Yes, but only when the gaps are visible, assessed as non-reversing for this decision and reflected in qualified wording.
When is expansion mandatory?
Expand when a material buyer, intent, stage or market cell is missing, or when new queries are still changing the decision.
How is this different from STRATA-6?
STRATA-6 explains how to build and freeze the sample. DECIDE-6 is the parent adequacy gate that evaluates the frozen sample and controls what decision may be released.
How the framework connects to adjacent controls
The adequacy decision sits between sample construction and downstream competitive or investment analysis.
The adequacy decision record
Calibration template: this record demonstrates the required fields. It is not a live market sample, a demand estimate or evidence that any listed coverage is representative.
The decision record makes the release state reviewable. It should be stored with the frozen sample rather than reconstructed later from memory.
| Record field | Required evidence | Release question |
|---|---|---|
| Decision and consequence | The exact question, audience, action and cost of error. | What will change if this conclusion is wrong? |
| Frozen sample version | The exact query list, source set, selection rules, exclusions and timestamp. | Can another reviewer load the same sample? |
| DECIDE-6 result | Pass, fail or unresolved status with evidence for every binding control. | Which first failure determines the release state? |
| Residual gaps | Every unsampled material cell, unknown and observation limitation. | Could any gap reverse the decision? |
| Release state | Accept, Qualify, Expand or Refuse, with the deciding rationale. | What action is permitted now? |
| Permitted wording | The strongest claim the evidence may support and the wording that remains prohibited. | Does every summary, chart and recommendation respect the same boundary? |
| Lifecycle | Owner, review date, stop condition and material reopen trigger. | Who decides when the sample is no longer fit for purpose? |
Adequacy is therefore not a permanent property of a query list. It is a governed relationship between a versioned sample, a specific decision and the evidence conditions observed at a particular time.
Carry the sample into the next control
Once the sample is accepted or qualified, carry its release state and residual gaps into opportunity, investment, execution, monitoring and retrieval decisions.
Authoritative context and review sources
- UK Government — The AQuA Book
- UK Government — Verification and validation for analytical work
- NIST — Guidelines for evaluating and expressing uncertainty
These sources provide external context for analytical quality, appraisal, evaluation, delivery or platform eligibility. They do not certify the proprietary SERPIFAI framework or validate an unstated outcome.