Optimize · Citation Optimizer
Citation Optimizer finds and fixes where AI drops your content.
Citation Optimizer is Spyglasses' content scoring and rewriting tool. It measures your pages against the four stages of AI's retrieval pipeline: query fan-out, search ranking, content extraction, and relevance scoring. Score a page, rewrite with stage-level guidance, and re-score until it survives to the answer.
Find out where your content falls out of the pipeline:
Query fan-out
6 in · 5 out
Search ranking
5 in · 4 out
Content read
4 in · 3 out
Relevance scored
3 in · 2 out
- /guides/tent-buying-guideCited
- /compare/ridgeline-vs-bastion
- /blog/winter-camping-tips
- /products/ridgeline-2Cited
- /faq/tent-care
- /press/2026-gear-awards
/compare/ridgeline-vs-bastion · Outscored by a fresher comparison
/blog/winter-camping-tips · Thin extractable copy
- of the AI retrieval pipeline, each scored
- 8 stages
- for every score, not just a number
- Stage-level reasons
- guided rewrite, then re-score the same page
- Rewrite loop
- for every fan-out query AI actually runs
- SERP ranks
Why it works
The answers vary. The pipeline doesn't.
AI responses change between sessions; that part really is a dice roll. But the retrieval steps that feed them follow known, published algorithms. Those steps are deterministic, repeatable, and measurable. That's where you work.
Stage by stage
What happens between a prompt and an answer?
Query fan-out
AI turns one prompt into several web searches
Search ranking
Your page has to rank for those searches
Content read
AI fetches the page and extracts what it can
Relevance scored
Extracted passages compete for the answer
The working loop
Score it. Rewrite it. Score it again.
Every scored page gets stage-level reasons, not just a number. Rewrite with that guidance, re-score, and watch the same page climb. This example took two rewrites.
/blog/winter-camping-tips · 2 rewrites
Grounding search rankings
Which searches does AI actually run?
Spyglasses shows the fan-out queries behind your tracked prompts and where you rank for each one. If you don't rank for the searches AI runs, nothing downstream can save you. This is the roadmap.
- best 4-season tents 20264 platforms#3
- ultralight tent comparison3 platforms#7
- tent waterproofing ratings explained2 platforms#1
- expedition tent wind rating3 platformsUnranked
- budget backpacking tent reviews2 platforms#12
Rather have us run the loop for you? That's the AI-focused SEO & rewriting service.
Explore services →Works over MCP too: score and revise from Claude or any MCP client.
See the MCP server →Stop guessing why AI skips you.
Score a page in minutes and get stage-level reasons you can act on today.
Frequently Asked Questions
Score any page against a real ChatGPT search query, see every step where it falls out of the citation pipeline, and rewrite it to get cited. Built on the deterministic retrieval steps you can actually control.
What is the AI Search Citation Optimizer?
It's a tool that models ChatGPT's citation pipeline as a sequence of gates, scores your page against each one, and shows you where your content would survive AI retrieval and where it would fall out. For every weakness it gives a specific recommendation, and it can rewrite the page for you and re-score it.
How is this different from a prompt tracker?
Prompt trackers run prompts and report what came back, which varies session to session. The Citation Optimizer scores the retrieval pipeline instead: the search, ranking, reading, and synthesis steps that follow published, repeatable algorithms. Those steps are measurable, so you can optimize for them and see the result.
How much does it cost?
Scoring is free for everyone, including unlimited re-scoring. A rewrite costs 10 credits and includes unlimited re-scoring of that page. Brand plans include 10 rewrites per month and Agency plans include 50; beyond the allowance, rewrites are billed per use.
Which AI platforms does it cover?
This release scores against the ChatGPT research pipeline, which is the best-documented. The same gate framework extends to Google AI Overviews, ChatGPT Business, Pro and Deep Research, Claude, and Google AI Mode and Gemini, which are planned for Q3 2026. Until a platform is modeled, the optimizer scores against ChatGPT and labels others as not yet scoreable.
Does it write content from scratch?
No. It rewrites content you already have to make it more citable: front-loading the answer, tightening passages, and restructuring for AI retrieval, plus a new title, meta description, and structured data. It never invents facts, statistics, or credentials; it works with what your content already supports.
Can I drive it from an AI assistant?
Yes. The whole score, rewrite, and re-score loop is available through the Spyglasses MCP server, so a connected assistant like ChatGPT or Claude can optimize your content in a conversation. Everything is scoped to your own properties and nothing is published on your behalf.