Update or Rewrite? An AI Content Decision Guide

Update a page when its core purpose, audience, and useful material are still sound. Rewrite it when the search intent, central promise, evidence, or structure is fundamentally wrong.

That distinction matters more than the page’s age or traffic decline. A newer article can need a rewrite, while an older evergreen guide may only need a few verified corrections.

AI can make the assessment faster by organizing performance data, comparing content, and locating unsupported claims. It should not make the final decision without human review.

The short decision guide

Use these four outcomes:

Finding	Best response
The page still serves the right reader and intent, but contains isolated gaps or outdated details	Update
The topic is still valuable, but most of the page no longer answers the reader’s task well	Rewrite
Several pages compete to answer substantially the same need	Consolidate
The page has no useful purpose, relevant audience, or suitable replacement	Retire

Do not choose an update simply because it is faster. Do not choose a rewrite merely because an AI tool produced a more polished draft.

Google recommends creating content primarily to benefit people and asks whether readers will leave feeling that they learned enough to achieve their goal. Its guidance also warns against changing dates or removing old content merely to make a site appear fresh (Google Search Central).

What counts as an update?

An update preserves the page’s main purpose and much of its useful content. Typical changes include:

• Correcting outdated facts, prices, screenshots, product names, or procedures
• Replacing broken or weak sources
• Adding a missing section that current readers need
• Improving an unclear introduction or explanation
• Updating examples while retaining the original argument
• Strengthening author information, citations, or review notes
• Improving internal links and removing irrelevant ones
• Aligning the title and description more closely with the existing content

The result should still be recognizably the same resource, only more accurate and complete.

An update is usually appropriate when most existing sections remain useful after fact-checking. There is no universal percentage that separates an update from a rewrite; the significance of the changes matters more than the number of edited words.

What counts as a rewrite?

A rewrite rebuilds the page around a new or heavily revised brief. Some verified material may survive, but the structure, argument, or answer changes substantially.

Consider a rewrite when:

• The page targets the wrong search intent
• The introduction promises something the article never delivers
• The advice depends on obsolete products, laws, interfaces, or practices
• The content summarizes other sources without contributing useful analysis
• Important claims cannot be verified
• The structure reflects keywords rather than the reader’s actual task
• Several sections repeat the same point while essential questions remain unanswered
• The article was produced at scale with little editorial or subject-matter input
• Fixing individual paragraphs would leave an inconsistent or confusing whole

Google does not prohibit content because AI helped create it. Its stated focus is the quality and purpose of the result. However, producing many AI-generated pages without adding value may violate its policy on scaled content abuse (guidance on generative AI content, spam policies).

A rewrite should therefore improve substance, not simply replace human-sounding sentences with different human-sounding sentences.

Diagnose the page before changing it

1. Confirm what the page currently ranks for

Open Google Search Console and filter the Performance report to the exact URL. Compare a meaningful period with the preceding period or the same period in a previous year when seasonality applies.

Review:

• Clicks
• Impressions
• Click-through rate
• Queries
• Countries
• Devices
• Search appearances

Treat average position carefully. Google describes it as an averaged value based on the topmost result and recommends focusing more on trends in clicks and impressions than on position alone (Search Console metric definitions, Performance report guidance).

Look for patterns rather than one headline number:

• Impressions are stable, but CTR falls: The search result presentation or query alignment may need a focused update.
• Impressions decline across the page’s main queries: Intent, demand, competition, technical issues, or content quality may have changed.
• The page appears for unexpected queries: Its focus may be unclear.
• Several URLs appear for the same queries: Consolidation or clearer differentiation may be more useful than rewriting one page in isolation.

A traffic drop does not prove that the content is poor. Google advises checking technical problems, seasonality, algorithm changes, and shifting search interest before drawing that conclusion (debugging Search traffic drops).

2. Check for search intent drift

Review the current search results for the page’s primary topic. Record the dominant result types and the tasks they help users complete.

For example, determine whether results now favor:

• Definitions
• Step-by-step guides
• Product pages
• Comparison pages
• Templates or tools
• Recent news
• First-hand reviews
• Videos or visual demonstrations

Do not assume that every visible competitor represents a ranking requirement. Use the results to understand the range of likely needs, then compare those needs with your audience and business purpose.

For a repeatable process, use the internal guide to How to Audit Search Intent Drift With AI in 45 Minutes.

3. Audit the substance

Evaluate each major section against four questions:

1. Is it accurate?
2. Is it necessary for the reader’s task?
3. Is it supported by an appropriate source or clear first-hand evidence?
4. Does it add something beyond a generic summary?

Mark each section as:

• Keep
• Correct
• Expand
• Replace
• Remove
• Verify with a subject-matter expert

A page with several isolated “correct” or “expand” labels is an update candidate. A page dominated by “replace,” “remove,” and “verify” labels probably needs a rewrite.

4. Evaluate trust and accountability

Check whether readers can understand:

• Who created or reviewed the content
• How important conclusions were reached
• Which claims come from external sources
• Which observations are original
• When the information was meaningfully reviewed
• Where limitations or uncertainty remain

Google suggests considering accurate bylines where readers would reasonably expect them. It also says AI or automation disclosures can be useful when readers would reasonably want to know how the content was created (Google’s AI-content guidance).

For related editorial checks, see 7 Ways to Build Trust Signals Into AI Content and How to Turn AI Drafts into E-E-A-T Content in 7 Days.

A practical update-or-rewrite scorecard

Score each statement as yes, partly, or no:

1. The page still serves the same audience.
2. Its primary search intent remains relevant.
3. Its central answer is accurate.
4. Most sections help readers complete their task.
5. Its important claims can be verified.
6. Its structure makes the answer easy to find.
7. It contains useful expertise, evidence, or analysis.
8. Its topic still fits the site’s purpose.
9. The existing URL accurately describes the subject.
10. The page has a distinct role that another page does not already serve.

Interpret the pattern rather than calculating a supposedly scientific score:

• Mostly yes, with a few partly answers: update.
• Several no answers involving accuracy, intent, or structure: rewrite.
• A no on distinct purpose because another page answers the same need: consider consolidation.
• No answers on audience, relevance, and site fit: consider retirement.

This scorecard is an editorial framework, not a documented Google ranking model.

How to use AI during the decision

AI is most helpful as an analyst and drafting assistant. Give it evidence rather than asking it to guess why a page lost traffic.

Useful inputs include:

• The existing article
• Search Console query data
• A manually reviewed summary of current search results
• Approved primary sources
• Product or policy documentation
• Editorial standards
• Intended audience and conversion goal

A practical prompt might be:

Compare this article with the supplied query data, reader goal, and approved sources. Classify every section as keep, correct, expand, replace, remove, or human verification required. Identify contradictions and unsupported claims. Recommend an update or rewrite, but explain the evidence for the recommendation. Do not infer ranking factors or invent missing facts.

Review the output carefully. In particular, verify every factual correction against the original source. AI can misunderstand source scope, merge unrelated facts, or state an uncertain conclusion too confidently.

Before publishing AI-assisted work, apply a consistent human QA process such as the checks in Stop Publishing AI Content Without These SEO Checks.

Workflow for a focused update

When the diagnosis supports an update:

1. Save the current page and its performance baseline.
2. List every claim, link, image, and instruction that needs verification.
3. Correct factual problems using primary sources where possible.
4. Add only the missing information required by the current reader task.
5. Tighten the introduction so it answers the main question early.
6. Improve headings, navigation, and formatting where they reduce effort.
7. Add relevant internal links without forcing keyword-rich anchors.
8. Check the title, description, canonical tag, structured data, and indexability.
9. Run a final human review for accuracy, usefulness, and tone.

Internal links should support genuine relationships between pages. The workflow in How to Build AI-Driven Internal Links in 30 Minutes can help identify those relationships without automating sitewide placement.

Workflow for a full rewrite

When the old page cannot support the right answer:

1. Write a fresh brief based on the intended reader and current task.
2. Extract only the old material that has been independently verified.
3. Build the new structure without copying the old heading order by default.
4. Gather current primary sources and original expert input.
5. Draft the central answer before supporting sections.
6. Add limitations, exceptions, and uncertainty where they affect decisions.
7. Compare the finished draft with the brief and reader task.
8. Perform editorial, legal, brand, and technical review as applicable.
9. Check every internal link, external citation, image, and structured-data field.
10. Record what changed so later editors can understand the rewrite.

If the topic remains the same, rewriting on the existing URL is often the clearest editorial choice. Changing the URL is a separate technical decision, not a requirement of rewriting.

If pages are being merged or a URL must change, use a relevant permanent redirect. Google treats permanent redirects as a signal that the destination should be canonical and advises against sending many unrelated old URLs to one irrelevant destination (redirect guidance).

Do not fake freshness

Change the visible “last updated” date only when the page has been significantly updated. Google recommends consistent visible and structured dates and says its systems estimate when a page was published or significantly changed (publication-date guidance).

For article structured data:

• Preserve the original datePublished
• Set an accurate dateModified after a meaningful change
• Keep structured dates consistent with visible dates
• Include the correct time zone
• Do not use a future date

Correcting punctuation or changing a few links is maintenance, not necessarily a reason to present the article as newly updated.

Hypothetical example

Suppose a page titled “AI Content Optimization Checklist” still receives impressions for relevant checklist queries. Most steps remain accurate, but several tool screenshots are obsolete, two citations are broken, and the article lacks a section on source verification.

The reader, topic, and central workflow remain intact. That points to an update.

Now suppose the same page is mostly a list of keyword-density instructions, treats automated text as inherently optimized, and does not address audience needs, evidence, review, or current search policies. Correcting individual paragraphs would not repair its underlying method. That points to a rewrite.

If the site already has a stronger checklist covering the same intent, consolidation may be better than maintaining two similar resources.

Measure the result without promising an outcome

Record the publication date and the changes made. Then monitor the same page and query groups used in the diagnosis.

Review:

• Organic clicks and impressions
• CTR by important query group
• Landing-page engagement
• Relevant conversions
• Rankings as a secondary trend
• Indexing and canonical status
• Feedback, corrections, or support questions

Compare equivalent periods where possible and account for seasonality, campaigns, site changes, and broader demand. Search performance is dynamic, and Google explicitly states that content improvements do not guarantee a noticeable search impact (core update guidance).

Conclusion

Choose an update when the page’s foundation remains useful and the problems are limited. Choose a rewrite when its intent, evidence, structure, or central answer no longer works. Consolidate overlapping pages and retire content that has no defensible purpose.

AI can accelerate classification, comparison, and editing, but the decision should rest on verified evidence, reader needs, subject expertise, and careful human review.

References

• Google Search Central: Creating helpful, reliable, people-first content
• Google Search Central: Guidance on generative AI content
• Google Search Central: Spam policies
• Google Search Console Help: Performance report tasks
• Google Search Central: Debugging Search traffic drops
• Google Search Central: Publication dates