Information Gain
Missing Comparison Criterion Audit: Add What Changes the Choice
A missing comparison criterion audit finds the absent dimension that could change a reader’s choice. A table can list many features yet omit suitability, total effort, evidence, risk, support, compatibility or lifecycle needs. Start with the decision the comparison must support.
Inspect the platform comparison example

Worked example: feature coverage is not process fit
An SEO platform comparison lists keyword counts, dashboards and data output formats. A team choosing a system for a complete project also needs to know how evidence, page ownership, review and maintenance move from one person to another.
| Missing criterion | Evidence that could answer it | How it changes suitability |
|---|---|---|
| Process continuity | Trace a sample task from project context through research, drafting, review and delivery. Record which steps need another system. | A team needing one continuous process may exclude an option with unsupported handoffs; a specialist doing one task may not need that breadth. |
| Evidence provenance | Inspect a sample claim and its source passage, date, scope and revision history. Distinguish documentation from a demonstrated record. | A team that must review claims needs traceable evidence, not just a link field or a promise of accuracy. |
| Page responsibility | Check how a sample task identifies its destination, content owner and overlapping answers. | A large content program may need explicit responsibility; a single analyst may be able to manage it outside the platform. |
| Review and exceptions | Follow an unresolved claim or failed check through assignment, repair and acceptance. | A controlled delivery process needs a clear return route. A visual status label alone may not meet that need. |
| Maintenance and transfer | Inspect how a changed source or departure of a team member is handled, including access to retained records. | Ongoing work may favor clear update ownership and portable records over more initial dashboard views. |
These are candidate criteria for this scenario, not claims about any named product. Each requires evidence before a comparison can rate an option. The audit should expose the reader’s requirements rather than turn a seller’s preferred functions into a universal definition of quality.
Define the reader before choosing the rows
State who is choosing among which options, for which outcome and under which constraints. A criterion is missing only in relation to that choice; not every possible dimension belongs in every comparison.
- Suitability and function. Define the task and the operation the option must complete. Separate advertised scope, demonstrated operation and work that needs another resource.
- Cost and effort. Use a consistent scope for price, setup, training, connections to other systems, recurring work and exit effort. Do not compare an entry price with a complete delivery cost.
- Evidence and risk. Assess how assertions are supported, what can fail and how the reader can respond. Record the source and the uncertainty instead of treating marketing language as verified behavior.
- Support and compatibility. Name the relevant assistance, systems, versions, permissions and environment. Intended support and a test of a particular combination answer different questions.
- Lifecycle and maintenance. Include updates, ongoing ownership, retained access and replacement when they affect the decision. The initial purchase is only part of the task for a reader maintaining the system.
Keep criteria independent enough to avoid counting the same benefit twice. For example, “process control,” “review control” and “approval process” may overlap unless their definitions distinguish different operations. A long table can give one selling point repeated votes without making that weighting visible.
A criterion can also be redundant. If it cannot change suitability, explain a meaningful tradeoff or choose an onward route for the stated reader, consider removing it. The aim is a decision the reader can inspect, not the largest possible catalog.
Make every retained criterion answerable
Define the row before scoring the option. State what counts, which unit or observation is used, where the evidence comes from and how an unknown result will be handled.

| Criterion role | How to assess it | What to do with incomplete evidence |
|---|---|---|
| Mandatory condition | Test the stated minimum against the same scope for every option. Explain why it is required for this reader. | Mark it unresolved and defer the affected recommendation if the condition controls eligibility. Do not silently pass it. |
| Weighted preference | State why the reader cares about it and how its weight relates to other distinct criteria. Use comparable units or a defined assessment scale. | Show the missing evidence and test how plausible alternatives would affect the verdict. Avoid a precise total built on guesses. |
| Tradeoff or context | Describe the circumstances in which the dimension improves or worsens fit. A numeric score may be unnecessary. | Qualify the comparison and state the evidence needed to resolve the tradeoff. |
| Not relevant to this reader | Explain the reader condition that makes it irrelevant. Keep that separate from a poor result. | Do not confuse “not needed,” “not supported” and “not yet verified.” They imply different decisions. |
Record units, definitions, source date, market or version, sample and evidence type beside the criterion. If one option has a test result and another only has documentation, disclose that difference. Provider statements, independent comparisons and direct trials can support different wording; a common row label does not make them equivalent.
Weights are an editorial model of the stated reader’s priorities, not objective facts about a product. Mandatory conditions should not be averaged away by unrelated strengths. Test the verdict after changing a disputed weight or resolving an unknown; if the preferred option changes easily, describe the recommendation as sensitive to that input.

Show the decision effect, not just the new row
Return to the worked example after adding the missing criteria. The useful contribution is the explanation of how the reader’s choice changes under specific conditions.
For a team that needs a traceable project from context to maintenance, a missing source history or unresolved review handoff can prevent a recommendation even when an option lists more keywords and dashboards. The next action is to obtain evidence of that operation or consider a supported alternative.
For an analyst doing one bounded research task, the same broad process criteria may serve a supporting role or be irrelevant. A narrower resource can remain suitable when the analyst has another supported way to manage review and maintenance. This does not establish a winner; it explains why the two readers need different verdicts.
- Record the changed choice. Name the new criterion, the evidence added and the effect on fit, tradeoff, verdict or next route. If nothing changes, explain why retaining the row helps the decision.
- Keep the record maintainable. Retain the audience, options, constraints, definitions, units, sources, weighting, unknown handling and exceptions.
- Assign review. Name the accountable owner, accepted or unresolved state, publication placement and next destination. Review again when pricing scope, product version, terms, task or evidence changes.
- Check the whole answer. Read the direct answer, comparison, evidence, terminology, page boundary and onward routes together. Remove a headline verdict that the revised table no longer supports.
When evidence is absent, hold, qualify, replace or remove the affected comparison claim. Do not add a seller friendly dimension solely to favor one option. A useful repair may be a row or explanation inside an existing owner; it does not require a new page.
Questions about missing criteria
Can a criterion be useful without a numeric score?
Yes. A clear prerequisite, bounded comparison or explanation of an incompatible environment can change the decision without a combined score. Use numbers when their definitions and evidence support them.
How many criteria are enough?
Enough to resolve the stated choice and its material constraints. Feature count does not measure completeness. Add a dimension for a reason and remove redundant or irrelevant rows.
How is this different from a missing process step?
A comparison audit checks the dimensions used to choose between options. A process audit checks the actions and dependencies needed to complete a method. A missing review step inside an option can become evidence for a comparison criterion about handoffs.
What if the verdict remains unchanged?
The new evidence may strengthen a bounded choice or reveal a material tradeoff. Explain that effect. Adding a familiar feature row without changing the decision or its support does not establish a useful contribution.
Related decisions and sources
Information Gain
Choose an addition that changes the reader’s decision or its support.
Missing Process Step Audit
Trace the actions behind a claimed process function.
Format Based Information Gain
Choose a format that makes the comparison usable.
Google: creating helpful content
Use the public self assessment questions as a broader editorial reference.
The examples are teaching scenarios, not records of work performed for a client. The illustrations explain record separation and revision; they are not evidence of the assertions described. These editorial methods do not guarantee rankings, indexing, traffic, rich results, citations or AI feature inclusion.
