Content optimization for LLMs & SEO

Put your site
through the Prism.

Prism Optimizer grades and optimizes your website content for both — the AI engines now answering your buyers’ questions (ChatGPT, Gemini, Perplexity, Claude, AI Overviews) and the search rankings that still send the traffic. Ten dimensions, one Prism Score, and the fix simulated before you publish it.

Request a demo Explore the 10 dimensions
AI Overviews People Also Ask ChatGPT Gemini Perplexity Claude Google rankings
Prism Score · example.com Live scan
Prism Score
72
▲ 9 vs last scan
Core Intent
88
Extractable
79
Citable
64
AIO Readiness
42
PAA Coverage
71
AI Overviews
12 ▲4
PAA owned
7 ▲2
LLM citations
23 ▲6
01The framework

One number your whole team can rally behind: the Prism Score.

AI engines don't rank pages — they select sources. The Prism Score grades every page on your site across the ten dimensions that decide whether an engine selects you: from core intent and implied questions to whether an engine can retrieve, extract, cite and reuse what you wrote.

Not a black box. Ten named dimensions, scored deterministically, comparable run to run — so movement means something.
  • ◆Page-level precision. Engines cite URLs, not domains — so Prism scores, grades and fixes at the URL level.
  • ◆A–F grades per dimension. Executives see the headline; writers see exactly which dimension is dragging.
  • ◆Tracked over time. Every scheduled scan updates the score and shows what changed — and why.

The 10 dimensions of the Prism Score

three families · per page
Content Quality
01Core IntentA
02Edge CasesC
03Implied QuestionsB
04Fan-out QueriesC
AI Accessibility
05RetrievableA
06ExtractableB
07CitableB
08ReusableA
Search Visibility
09AIO ReadinessC
10PAA CoverageB

Dimension names are the live scoring rubric. Grades shown are illustrative sample data.

02Live answer-surface data

Fetched live from the engines. Nothing estimated.

Every scan pulls current data from the surfaces where answers actually happen — so you're optimizing against today's results, not last quarter's export.

Live

AI Overviews

Which of your keywords trigger an AI Overview, who's quoted inside it, and whether that's you or a competitor.

Live

People Also Ask

Question-level ownership: the PAA boxes your pages hold today, and the open questions you could claim next.

Live

SERP rankings

Where every page actually sits today — pulled fresh each scan, never carried over from a stale export.

Live

Fan-out queries

AI engines fan one question out into many. Prism fetches the query variants live and maps the ones your pages must answer to stay in the result.

Live

Edge cases

The long-tail phrasings and unusual intents where answers swing — fetched live, so wins there aren't left to chance.

Live

Verified search demand

Keyword-level volume behind every recommendation — never inflated grouped totals.

Every card above is a live fetch, run per scan and scored into the Prism Score. Competitor scores and publishing velocity are fetched live too — see section 05.

03LLM optimization

Optimized for every LLM. Not just one.

Each engine selects sources differently. Prism builds prompt sets from your pages' real intents, runs them live across ChatGPT, Gemini, Perplexity and Claude, and records the outcome — cited, mentioned, or absent.

  • ✓Per-engine, per-prompt results — see exactly where you're a source and where you're invisible.
  • ✓Competitor attribution — when you're absent, see whose URL took the citation instead.
  • ✓Dimension-mapped prompts — every prompt ties back to the Prism Score dimension that influences it, so results point straight at the fix.
  • ✓Crawler access verified — GPTBot, ClaudeBot, PerplexityBot and Google-Extended checked on every scan, plus llms.txt.
"best option for [your category]"1 of 150 prompts
Gemini/guides/comparisonCited
Perplexity/guides/comparisonCited
ChatGPTbrand onlyMentioned
Claudecompetitor citedAbsent
04The workbench

Simulate the score. Then publish.

No other step in your content process lets you see the result before you ship. Prism's Optimize workbench does: it drafts the rewrite in your brand voice, then simulates the new Prism Score — before a word goes live.

  • ✓AI rewrites tuned to your brand profile — upload it once, every draft speaks your language.
  • ✓Simulated lift, dimension by dimension — know which fixes matter before spending writer time.
  • ✓Implementation packets — export a .docx with heading structure and schema markup ready to paste into your CMS.
1
Rewrite
2
Simulate
3
Publish
4
Verify
Current
27
Grade D
→
Simulated
After rewrite
66
Grade B

Illustrative example of the rewrite → simulate flow. Simulated scores use the same deterministic engine as live scans.

05Competitive intelligence

Know how fast your competitors publish. And what.

Visibility is a race for the citation. Prism tracks your competitors' publishing velocity — how many pages they ship per month, what they published this week, and how their Prism Scores compare to yours, run over run.

  • ✓Velocity trends — content output per competitor, month by month.
  • ✓First-seen feed — new competitor pages surfaced the moment a scan finds them.
  • ✓Score matrix with tickers — every competitor's movement vs. the previous run, side by side with yours.

Publishing velocity · pages/month

last 90 days
Your site YOU
12 /mo
Competitor A
27 /mo
Competitor B
18 /mo
Competitor C
8 /mo
Competitor A published “2026 buyer's guide…”2 days ago
Competitor B published “How to choose…”5 days ago

Scheduled scans

Re-scan on your cadence with a "what changed" report after every run — real movement separated from measurement noise.

Score history & trends

Every scan is kept, so the Prism Score is tracked over time per page — you can show movement, not just a snapshot.

Provenance on every number

Verified, modeled or benchmark — each figure is labeled with its source. No silent estimates, in the app or the boardroom.

06The deliverable

A complete content audit report. Not a dashboard to interpret.

Every scan produces one comprehensive audit — page by page, dimension by dimension, with the live evidence behind each number and the fixes ranked by what actually moves the Prism Score.

  • ◆Every URL, fully graded. A Prism Score plus A–F on all ten dimensions for each page in scope — not a domain-level average that hides the problem.
  • ◆The live evidence attached. AI Overviews, People Also Ask, rankings, fan-out queries, edge cases and search demand shown as the data each grade was built from.
  • ◆Competitive context. Where you sit against tracked competitors on score and publishing velocity, in the same document.
  • ◆Ranked fixes, not a list of problems. The highest-impact change per page first — with the simulated score it should reach.
  • ◆Provenance on every figure. Verified, modeled or benchmark, labeled throughout. Nothing silently estimated.
  • ◆Client-ready on export. A branded PDF your stakeholders can act on, plus per-page .docx packets a writer can work straight from.

Content Audit Report

example.com · 128 URLs · scan of 12 Aug 2026
Sample
01Executive summary & Prism Scorep. 1
02Score by dimension — all tenp. 3
03Page-by-page grades & findingsp. 6
04Live answer-surface evidencep. 24
05LLM citation results by enginep. 31
06Competitor matrix & velocityp. 38
07Prioritized fixes & simulated gainp. 44
08Methodology & data provenancep. 52
Contents illustrative — sample reportPDF · .docx packets
07Insights

Research on the answer economy.

The blog home — sourced, practical writing on AI visibility. Cards below are samples.

View all articles →
Sample
AI Overviews

What actually gets a page quoted in an AI Overview

Structural patterns shared by pages AI engines cite — and the ones they pass over.

8 min read
Sample
Framework

Inside the Prism Score: why ten dimensions, not one

How a composite, deterministic score keeps optimization honest — and comparable over time.

9 min read
Sample
Competitive

Publishing velocity: when speed matters — and when it doesn't

What competitor output actually predicts about citation share, from live tracking data.

6 min read

See your Prism Score.

A live walkthrough on your own site — your pages, your citations, your gaps across every AI surface.

Request a demo