Does any of this sound familiar?
We have published a lot. Now we don’t know where to start.
Your content library has grown, perhaps with help from AI, and keeping it useful is becoming difficult. The starting point is to identify pages that matter to buyers, establish what is actually wrong, and organise the work.
Which group of pages deserves attention first—and why?
Our articles explain the topic, but not why someone should contact us.
A reader can understand an idea without seeing how your business can help. We examine the connection between their problem, your explanation, your expertise and the next step.
What needs to be clearer for an enquiry to make sense, without turning every guide into a sales pitch?
These pages look similar. Should we merge them?
Similar titles are not enough to justify a merger. Compare the needs each page serves, what it adds, its links and the evidence available.
Keep distinct roles, improve an individual page, or plan a consolidation supported by evidence.
Illustrative example: a useful guide with no clear next step
This is a fictional teaching example, not a MindVerve client story or a report of measured results.
- Before
- A company publishes a guide to choosing a marketing tool. Readers learn about features, but cannot tell which ones fit their organisation. The page ends with “Contact us”.
- What is missing
- The guide explains the tools but does not yet help readers choose in their own circumstances. It never explains how the company’s expertise could support that decision.
- Proposed change
- Keep the useful explanations and add practical selection criteria: the task to improve, existing tools, constraints and necessary approvals.
- After — sample wording
- Struggling to choose because a comparison table doesn’t reflect your constraints? Tell us which task you want to improve and which tools your team already uses. We can discuss what support would fit your situation.
- Why this is clearer
- Readers can see when specialist help is relevant and what to explain when they get in touch. That is not evidence of more enquiries; any effect still needs to be observed after publication.
What to apply to your own website
- Give each page a clear audience and a problem to explain or a decision to support.
- Address an observed gap rather than just changing words or adding a button.
- Before merging or removing a page, understand what it contributes and what could be lost.
Use the worksheet below to record your observations. It supports your judgement; it does not interpret the evidence or make the decision for you.
A decision register for your content audit
Start with the reader’s task and evidence you can verify. Missing data does not mean zero performance: this register proposes decisions; it does not publish, redirect or delete anything.
- Evidence
- Decision
- Next action
Keep
Keep a page when it still serves a reader need and no necessary correction has been established. If usage is unknown, keeping it provisionally does not demonstrate success.
Improve
Fix a specific, verified gap or obstacle, then check the correction. Changing the date alone does not improve the content.
Consolidate
Confirm that the pages serve the same intent, preserve their useful differences, and plan the destination and affected links. Shared keywords alone are not enough.
Retire
Treat this as an exceptional decision: review the current need, dependencies and any replacement. Age or low traffic alone is not sufficient; decide the technical treatment separately.
Investigate first
When evidence is missing or conflicting, record the question, its owner and the evidence needed to make a decision.
Distinguish proposed, approved and implemented. Record the execution date and checks. Retiring content is an editorial decision, not a synonym for noindex, canonicalisation, redirection or deletion.
Fictional teaching examples. These example.com pages are not audited sites, clients or measured results.
Reference page — fictional
- Page, language and reader need
- example.com/en/reference-format; supporting reference
- Evidence, source and date or period
- Another resource links to it; fictional observation
- Unknowns and dependencies
- Usage data unavailable
- Proposed action
- Keep provisionally
- Reason and value to preserve
- Supporting information may remain useful
- Destination and links
- Keep URL; review references
- Owner, next step, validation and status
- Owner to assign; verify need; proposed
Similar guides — fictional
- Page, language and reader need
- example.com/en/choose-format and example.com/en/compare-formats
- Evidence, source and date or period
- Similar introductory questions; fictional observation
- Unknowns and dependencies
- Audience, distinct information and intent not compared
- Proposed action
- Investigate first
- Reason and value to preserve
- Shared words do not prove duplicate need
- Destination and links
- No destination or redirect chosen
- Owner, next step, validation and status
- Owner to assign; compare intent; proposed
Outdated instruction — fictional
- Page, language and reader need
- example.com/en/instructions
- Evidence, source and date or period
- Hypothetical conflict with official documentation
- Unknowns and dependencies
- Verify source URL, excerpt, access date and corrected step
- Proposed action
- Improve after verification
- Reason and value to preserve
- Preserve the useful parts of the guide beyond the specific correction
- Destination and links
- Keep URL; check affected links
- Owner, next step, validation and status
- Owner to assign; validate correction; proposed
Old resource — fictional
- Page, language and reader need
- example.com/en/old-resource
- Evidence, source and date or period
- Removal suggested because of age; fictional situation
- Unknowns and dependencies
- Current need, support, dependencies, replacement unknown
- Proposed action
- Investigate first
- Reason and value to preserve
- Age is not evidence of uselessness
- Destination and links
- No removal or redirect decided
- Owner, next step, validation and status
- Owner to assign; document evidence; proposed
Your blank decision register
Download a blank file and fill it in on your own device. Do not include personal data, form messages or sensitive URL parameters.
Sources for the method
An editorial decision aid, not a scoring system prescribed by Google.
Our method
An audit should connect each URL to a purpose, an audience and observed performance. AI can organise patterns, but the decision needs context.
Combine a URL inventory with available search and conversion data. Review intent, freshness, duplication and internal links. Have the model propose classifications with reasons, then inspect important pages and ambiguous recommendations before acting.
How to assess progress
Avoid deleting content solely because it has low traffic or a model dislikes it. Check seasonality, strategic value and redirects. Prioritise changes that improve a real user journey.
Exercise
Decision
Decide which user journey the audit should improve before ranking URLs. A product launch, outdated offer or fragmented topic needs different priorities from a site-wide traffic review. Define what each page helps a visitor decide, and what would be lost if the page disappeared.
Exercise
Sample thirty URLs across high traffic, low traffic, conversions and older content. Give the model page purpose, available performance data and the actual text. Ask for retain, improve, combine or retire proposals with evidence. Manually inspect all proposed removals and a sample of the other categories, then note where missing context changed the recommendation.
Deliverable
Create an action register with URL, audience need, observed issue, proposed action, destination if combined, owner and dependency. Separate editorial work from measurement or technical fixes. Prioritise a small batch that completes one journey, with a before snapshot and a date to review the effect of the changes.
Pitfall
Low traffic does not prove a page has no purpose. A reference page may support a few valuable conversations or answer a rare but essential question. Check internal links, existing references, seasonality and conversion assistance before retiring content or assigning a replacement destination.
You can see the problem. Need help deciding what to do?
Expertise helps when several things need to be considered together: buyers’ questions, what your pages actually explain, your offer, the links between pages and the evidence you have.
Within an agreed scope, support can be organised around three elements:
An evidence-based diagnosis of selected pages
Separating observed problems from information that is still missing.
Prioritised recommendations
Explaining the reason for each action and what needs to be preserved.
An initial set of changes agreed together
Including the approvals and checks the work requires.
Page count, necessary access, deliverables and any implementation are agreed before work starts. Traffic, rankings and sales are not guaranteed.
Which page or group of pages is causing you difficulty right now?
Discuss the pages you want to improveShare your website and what is holding you back: choosing priorities, making a page clearer or deciding between overlapping content. Do not send confidential information.
Choose a useful next step
Understand how AI search finds sources, review your pages, then separate visibility observations from visits and enquiries. Use those findings to choose what to improve in your editorial workflow—not as a promise of citations.
- AEO and GEO: a practical approach to AI search visibility
- Measure visibility in AI answers responsibly
- Help buyers find your expertise in search and AI answers