How to Decide Refresh vs New Article With AI From Your Content Inventory

The problem this solves and who it is for

This workflow is for editors, content leads, and strategists who manage an existing content library. The common problem is not generating more ideas. It is deciding whether a topic needs a new article, a refresh of an existing page, a merge, or a tighter repositioning.

The output is a decision brief for one topic. It helps you assign the right kind of work instead of defaulting to write a new article every time.

Prerequisites

• A Gemini or ChatGPT account with file upload support
• A content inventory export with titles, URLs, publish dates, and target topics if available
• A small set of live search observations for the topic you are evaluating
• A note about the business value of the topic

How to capture or gather the source material

1. Export a simple content inventory into CSV or spreadsheet format.
2. Filter it to the topic cluster you are evaluating.
3. Add columns if helpful: current ranking page, last updated date, conversion relevance, and owner.
4. Run a quick live search for the topic and capture two or three screenshots or notes about what the current search field looks like.
5. Save those notes or screenshots in the same folder as the sheet.

Step-by-step workflow

1. Upload the filtered content inventory. Include only the rows relevant to the topic cluster.
2. Add the live search notes. Paste or upload a small summary of what you saw in the search field.
3. Ask for overlap and gaps. Have the model identify which existing pieces already serve the topic, which pieces partly overlap, and where the current field suggests a different angle.
4. Ask for the decision brief. Request a verdict of refresh, new article, merge, or reposition, plus the reasoning.
5. Use the brief to assign next steps. The point is to make the assignment decision cleaner, not to produce a full draft yet.

Tool-specific instructions

Primary recommendation: Gemini

Gemini is a practical choice here because Google documents file analysis in Gemini Apps, including documents and spreadsheets. That makes it good for a focused content-inventory review paired with a small set of search notes.

Practical setup:

• Filter the sheet before upload so the model is not distracted by the whole site.
• Include one business-value note so the recommendation is not based on traffic patterns alone.
• Ask for a single verdict with reasons, not a list of possibilities.

Alternative: ChatGPT

ChatGPT is a very good backup for sheet-based analysis. Use a Project if you will run several topic decisions in the same working space.

Alternative: NotebookLM

NotebookLM is more useful when the audit becomes source-heavy and you want to keep inventory notes, search screenshots, and revision memos together over time. For a one-off decision, Gemini is usually faster.

Copy and paste prompt blocks tailored to the workflow

Inventory analysis prompt

{
  "role": "content strategist",
  "task": "analyze a filtered content inventory against a live search snapshot",
  "goal": "decide the right assignment type for one topic cluster",
  "instructions": [
    "Use the uploaded content inventory and the pasted or uploaded live search notes.",
    "Identify which existing content already overlaps with the target topic, which pieces are partial matches, and where the current search field suggests a different or missing angle.",
    "Do not write the final content brief yet."
  ],
  "output_format": {
    "clear_existing_matches": [],
    "partial_matches": [],
    "content_gaps": [],
    "signs_the_current_field_has_shifted": []
  }
}

Decision brief prompt

{
  "role": "editorial planner",
  "task": "decide refresh versus new article versus merge versus reposition",
  "goal": "produce one clean assignment decision for the topic under review",
  "instructions": [
    "Use the inventory analysis already completed in this chat.",
    "Choose one verdict: refresh, new article, merge, or reposition.",
    "Explain why that is the best assignment choice right now.",
    "Include what asset should be updated or created, what the assignment goal should be, and what research or verification is still needed before writing."
  ],
  "output_format": {
    "verdict": "",
    "why_this_is_the_best_choice": [],
    "asset_to_update_or_create": "",
    "assignment_goal": "",
    "research_or_verification_still_needed": []
  }
}

Quality checks

• Confirm the verdict is one clear action, not a vague maybe.
• Make sure the recommendation reflects both the inventory and the current search field.
• Check that the business-value note actually influenced the decision.
• Remove any recommendation based on missing or assumed performance data.

Common failure modes and fixes

Failure mode: The model recommends a new article too quickly.
Fix: Ask it to prove why existing content cannot be refreshed or merged first.

Failure mode: The inventory is too broad.
Fix: Filter to one topic cluster before upload.

Failure mode: The verdict ignores business goals.
Fix: Add one explicit note about conversion relevance or strategic importance.

Failure mode: The search notes are too thin.
Fix: Capture two or three screenshots or a short manual summary before re-running.

Sources Checked

• https://support.google.com/gemini/answer/14903178 (accessed 2026-03-25)
• https://help.openai.com/en/articles/8555545-file-uploads-faq (accessed 2026-03-25)
• https://help.openai.com/es-es/articles/8437071-data-analysis-with-chatgpt (accessed 2026-03-25)
• https://support.google.com/notebooklm/answer/16215270 (accessed 2026-03-25)

Quarterly Refresh Flag

Review by 2026-06-23 to confirm tool interfaces and supported file workflows still match the live products.