Optimization scoring: how AdAstra measures a rewrite

AdAstra's optimize tool does not just rewrite content, it measures the rewrite. Here is exactly what changes, what gets scored before and after, and where every number actually comes from.

~6 min read . Optimize

Optimize is the one AdAstra tool that both rewrites content and grades its own work. Turn on the right toggles and a single run produces a new draft, a structured change log, and a before and after comparison across SEO, LLM readiness, conversion strength, and real projected search volume and CPC.

What "optimize" actually changes

optimize_content runs one holistic rewrite pass, not a stack of separate edits. Whichever toggles you enable, on-page SEO (enableSeoOptimization), LLM readiness (enableLlmOptimization, with a style such as FAQ Style, In-depth Analysis or Key Takeaways and its own intensity level), conversion optimization, and search volume and CPC optimization (enableVolumeCpc), all get applied together, against the same original text, in the same pass.

Two toggles change how the rewrite behaves rather than what it targets. useOriginalWording tells the model to make only minimal, surgical edits: preserve the original wording, length and sentence structure, and change only what is needed to weave in high-value keywords. useAdvancedAi swaps in a stronger model for the rewrite step itself, for runs where you want more editorial firepower behind the same settings.

You can also hand the run extra context: an optimization goal and level (0 to 100) that set what the rewrite is chasing and how far it may deviate from your original text, real ad copy to align against, specific keywords to integrate, a content framework, and a selection of SEO or conversion standards drawn from AdAstra's own standards library.

The before/after score model

Optimize does not report a single new score, it reports a pair: one for the original content and one for the rewrite, both produced by scoring the two versions the same way, so the comparison is apples to apples rather than a guess dressed up as a measurement.

That before and after pair is the actual shape of the result: a scores object carrying seoScore and llmOptimization, a conversionScore when conversion optimization is on, and a separate projectedMetrics object carrying search volume and CPC, each with its own before and after.

Which of those actually appears depends on which toggles you turned on. scores shows up when SEO or LLM optimization is enabled, conversionScore shows up when conversion optimization is enabled, and projectedMetrics shows up only when volume and CPC optimization is enabled. Nothing you did not ask to optimize gets a score attached to it.

SEO and LLM scores

The scores object carries two fields, seoScore and llmOptimization, each 0 to 100. seoScore reflects the same on-page fundamentals covered in AdAstra's SEO checklist, keyword relevance and structure read together. llmOptimization reflects how well the content would work as a source an AI system can extract from and cite, the same FAQ, depth and key-takeaways read behind AdAstra's GEO scoring, and enabling LLM optimization is also where you choose which of those three styles the rewrite leans into.

Those two numbers are illustrative only, a reminder of the scale and of the fact that this is a before and after, not a claim about any real piece of content. llmOptimization moves the same way alongside it, scored on the same 0-to-100 range.

Conversion score

conversionScore is a 0-to-100 read of the content's persuasive strength, included when conversion optimization is enabled. It is the same kind of signal covered in AdAstra's landing page quality scoring: does the copy make a clear case for the sign-up, subscription or purchase you want, not just describe a topic. It moves with a rewrite the same way the SEO and LLM scores do, reported as a before and after.

As with the SEO and LLM pair, both numbers here are placeholders, not a promise about what any specific rewrite will score. What they represent is real: two independent reads of the same content, on the same scale, taken before and after the same rewrite.

Projected search volume and CPC

projectedMetrics is the one comparison that never touches an AI model's own guess about numbers. When enableVolumeCpc is on, AdAstra extracts roughly 15 keywords from your original content and, separately, roughly 15 keywords from the rewritten content, then looks both sets up for real against Google Ads keyword data for your target country and language, the same lookup used throughout AdAstra.

The before figure is the summed monthly search volume and the average CPC across the original content's keywords; the after figure is the same calculation run on the rewrite's own keywords. CPC is reported in EUR, converted from the publisher-side cents figure (the revenue-share-adjusted number covered in AdAstra's keyword economics article) so it reflects what the traffic is worth to your site. Nothing here is estimated by the model: it is a live lookup, run twice, once against each version of the text.

The change log

Alongside the rewrite itself, optimize always returns changesMade: a structured log of what actually changed, as an array of category and description pairs (for example, category "SEO" or "Keyword Analysis", with a short, first-person description under 13 words). It is generated in the same pass as the rewrite, not reconstructed afterward by diffing text.

The log is written to cover every optimization you enabled: if volume and CPC optimization is on, a Keyword Analysis entry explaining how keywords were used is mandatory, not optional. Descriptions are written in your interface language regardless of what language the content itself is in, so the log stays readable even when the rewrite is in German or French.

The short version

• optimize_content runs one holistic rewrite pass, applying whichever toggles you enable together against the same original text.
• useOriginalWording forces minimal, surgical edits; useAdvancedAi swaps in a stronger model for the rewrite step.
• Every comparison is a real before and after: the same version of content and the same rewrite, scored the same way twice.
• scores (seoScore, llmOptimization) appears when SEO or LLM optimization is enabled; conversionScore appears when conversion optimization is enabled.
• projectedMetrics (search volume and CPC) appears only when volume and CPC optimization is enabled, and comes from a live Google Ads lookup run once on each version of the text, never a model guess.
• changesMade is a mandatory, structured log of what was changed and why, generated alongside the rewrite, written in your interface language.