Score · AI Placement Quality Score
AI Placement Quality Score: will AI cite this article?
PQS is Spyglasses' article-level placement score. Where AIPVS rates the publisher, PQS reads the specific article, real or prospective, and scores its chance of influencing ChatGPT, Gemini, Claude, and Perplexity answers: substance, structure, brand treatment, and accessibility.
PQS at a glance
- Scores the article itself: substance, structure, brand treatment, accessibility.
- Works before you pitch (hypothetical) and after you land it (real URL).
- Same outlet, two articles, very different odds; PQS shows which.
- Feeds coverage groups, so landed placements get watched for citations.
Product deep-dive
1,800 words · brand featured
News brief
240 words · passing mention
What PQS reads
The article, not the masthead
Two pieces in the same outlet can have wildly different odds of being cited. PQS scores what's on the page: how substantial it is, how extractable it is, how your brand is treated, and whether AI can reach it.
Before and after
Score the pitch, then score the placement
Model a prospective placement before you commit the effort, then score the live URL when it lands and see what the outlet's edit did to your odds.
Pair it with the publisher's AIPVS for the two-number briefing.
AI Placement Value Score →Landed placements go into coverage groups and get watched for citations.
For PR & Comms →Same outlet. Different odds.
Score a placement, real or hypothetical, in seconds.
Frequently Asked Questions
Score a specific brand placement — real or hypothetical — for how likely it is to be cited by AI assistants. The AI Placement Quality Score evaluates document position, chunk containment, sentiment, and link attributes on a single page.
What is the AI Placement Quality Score?
The PQS is a 0–100 score that evaluates how well a specific brand mention on a specific page is structured for AI citation. It combines document position, token chunk containment, placement type, editorial control, link attribute, and surrounding sentiment into a single number.
How is it different from the AI Placement Value Score (AIPVS)?
AIPVS answers 'how valuable is this domain for a placement?' — it's a domain-level metric. PQS answers 'how well does this specific placement on this specific page perform?' — it's a page-level metric. They're complementary: Total Placement Value = AIPVS × PQS / 100. A great domain with a mediocre placement position can score lower than a good domain with an excellent placement.
What's the difference between retrospective and prospective mode?
Retrospective mode scores a placement that already exists — paste the URL and we fetch, tokenize, and score the real content. Prospective mode scores a hypothetical placement — enter the publisher, placement type, and position to evaluate an offer before you accept it. The prospective mode also previews the lift from negotiating for better terms.
How do you detect the brand mention, link attribute, and placement type?
We locate the brand mention by walking the tokenized content looking for your property's name and aliases. We parse the raw HTML to find the rel attribute of any link pointing at your domain. Placement type and sentiment around the mention are classified by an AI model given the three-chunk window containing the mention.
What is key message pull-through?
Pull-through measures how well a property's key messages come through in a placement's content. We compute semantic similarity between each key message and each sentence in the placement and surface the matches. This is informational only — it does not affect the PQS score — but it helps you see whether your messaging is making it into the coverage.
Is it available through the API?
Yes. The /v1/ai-impact/placement endpoint scores a prospective placement given its inputs and optionally combines it with AIPVS for a Total Placement Value. Retrospective scoring is available through the web interface.
Can I score an article before it's published?
Yes. Score a hypothetical placement by describing the outlet and planned angle, then re-score the live URL once it publishes to see how the final edit changed your odds.