Optimize existing content

Use Optimize existing content when you already have an article or page and want to improve how it answers a specific monitored prompt. PromptEye imports the source, compares it with related query phrases, and builds a brief for an updated article. It does not silently overwrite your website; you review the imported source and the generated version before using it.

Add the source text

Choose New article → Optimize existing content. Provide the article by one of these methods:

• Paste text if the article is in a document or editor you can copy from. The preview preserves common Word and Google Docs formatting when pasted.
• Fetch from URL if the page is public and accessible to PromptEye’s fetcher. The fetch flow is intended to extract article text without page navigation, footers, or advertisements.
• Upload file to use a DOCX, PDF, or TXT source.

Before accepting imported content, inspect the beginning and end. A fetched page or file can include a menu or footer, or omit part of the article. The screen shows extraction details such as character count and detected headings. Continuous text without headings does not block optimization, but headings can make it easier to check that the right material was extracted.

A page may block the fetcher. If the URL cannot be fetched, paste the text manually or upload a supported file. You can adjust the content in the preview before continuing. Choosing Use this content confirms the source for the workflow; the import preview is not saved to your site and does not replace the live page.

Choose what the article should answer

After reviewing the source, select a monitored prompt or choose the unlinked article mode. A monitored prompt ties the optimization to a question whose visibility and citations can be checked later. In unlinked mode, the article can still be improved, but PromptEye does not connect it to a monitored-prompt measurement.

The wizard shows the source character count and uses the chosen prompt to calculate query fan-out and topic coverage. Query fan-out is a set of related questions and phrases around the main prompt; read Create an article and review its brief for a full explanation. The aim is to find ways to strengthen this article’s answer without making it a catch-all for every related subject.

Understand source fit

The source-fit view estimates how many fan-out phrases already have coverage in the imported text. A higher fit means more of the planned subject matter is present in the source, so the model can edit and expand what is there. A lower fit means the selected prompt and existing content may not match; much of the resulting article could be new.

If the fit warning appears, you can go back and choose a more suitable prompt or continue. Use the warning to decide whether you are truly refreshing this article or asking the model to create substantially different content. If the prompt is wrong, changing the prompt before generation is usually easier to review than trying to repair an off-topic outline afterward.

Source fit measures phrase coverage, not writing quality, factual accuracy, SEO performance, or the value of the page. A low percentage does not automatically mean the article is poor. It says that the source may not cover the themes associated with the selected target prompt.

Add the right sources and style

Continue through the Knowledge and Style steps. Select project documents that support this article’s subject. The wizard also lets you include the project’s Brand input. Choose only relevant materials; the fact that a document exists in the project does not mean it belongs in every article.

Writing preferences help control tone, audience level, structure, brand presence, perspective, and call to action. They do not turn off the brief’s shared editorial rules. If the optimization needs to preserve terminology, avoid a particular claim, or follow a specific editorial approach, add a clear note and verify the result against your approved facts.

Review the brief before generating

The brief turns the target prompt and its fan-out into an outline, phrases, direct-answer notes, and potential separate-article recommendations. For existing content, it can show source fit and warn when much of the article is likely to be new. Review the title, heading order, section notes, selected phrases, source documents, and suggestions before continuing.

The outline can retain some source sections, edit them, add new sections, or omit sections that do not fit. The editor later explains changes such as kept or omitted sections, added headings, and FAQ questions. Use these details to check whether the output still reflects the source’s purpose and whether anything important was removed.

You can finish on the brief and save it to the library if you want to agree on a plan before generating. Export it for review by a writer or subject-matter expert, then return to generate the updated version.

Review the original and optimized version

The content editor separates Original, Brief, and Optimized views for an optimization. The original source is read-only; it preserves the imported text so you can compare it with the new version. The brief records the decisions used to generate that version. The optimized view is editable and saves automatically.

The article editor may explain what changed: whether it adjusted the title to the target prompt, added tables or comparison sections, expanded the FAQ using unused fan-out phrases, inserted direct answers, retained some source sections, or omitted off-topic material. Review each change. Automated generation can keep, alter, add, or remove content according to the brief, and its summary is not a substitute for comparing factual claims with the source.

When the editor offers AEO Review, it analyzes the current text for issues such as clarity, structure, SEO, AEO, tone, and evidence. It can group findings by severity and issue type. Marking an issue resolved changes its review state; it does not itself edit the article. Use a suggestion as an editorial lead and verify it against the actual page and your standards.

If text changes after an analysis, the review can become stale. The screen explains when an analysis cannot be run again for that content version or when a completed analysis is unavailable. Make edits before running a review when possible, and treat the review as evidence for the text version it analyzed. Do not assume the review automatically reruns after every edit.

Revise a passage or regenerate

You can use the editor to make direct edits, adjust headings, insert links, and format text. The AI rewrite action works on a selected passage and replaces that fragment in place. Select a short passage, state the change you want, and read the replacement in context. The interface limits the selection length and shows remaining rewrite uses; the generated passage still needs review for accuracy and tone.

Regenerating an article creates a new version from the brief. It may discard manual edits from the current version. Before proceeding, choose whether to keep the previous article in the library or replace and delete it, as offered by the confirmation dialog. Keeping the previous version preserves it as a separate library item; replacing it removes the current item as part of regeneration. Export or copy any work you need before choosing to replace it.

The original imported text remains a read-only reference where available. Discarding a pending optimized rewrite restores the prior editor version; accepting it applies the proposed version. Read the confirmation and compare views before accepting or replacing content.

Prepare the updated page for the loop

After review, use Mark as ready when the article is suitable for your publishing process. This changes its PromptEye workflow status; it does not publish the page. Publish the revised content through your CMS, then add or update its live URL in the library. PromptEye uses that URL for indexing and citation tracking. If you change the live URL, update the record so later signals are associated with the right page.

Common questions

Does optimizing edit my live website?

No. PromptEye imports or receives the source text, then produces an editable version in the application. You publish approved changes through your own website or CMS process.

What if my article does not have a heading structure?

You can continue. The wizard explicitly says that continuous text with no detected headings does not block optimization. Check that the correct text was imported and review the outline generated from the selected prompt.

What if the chosen prompt does not fit the source?

Use the source-fit warning to return and select a more appropriate prompt, or continue knowing that the output may be largely new. Compare the final version with the original before using it.

Does AEO Review prove the article is ready or will be cited?

No. It reports potential improvements in the text version it analyzed. Readability, structure, and answer visibility are separate outcomes; review the page and later check its citations in the Content library.

What happens to my current edits when I regenerate?

The regeneration confirmation asks whether to keep the previous article in the library or replace and delete it. Review that choice carefully before continuing because the replace option removes the current article item.