Placements

What This Dashboard Does

The Placements dashboard scores a specific brand mention on a specific page. Where the AI Placement Value Score rates a whole publisher, the AI Placement Quality Score (PQS) rates an individual placement — how well it's positioned in the page, whether the surrounding context is coherent enough for an AI to extract as a citation, how favorable the sentiment is, what kind of link attribute it has, and more.

Use this dashboard to:

Audit existing placements by pasting a live URL and letting Spyglasses fetch, tokenize, and score it automatically
Evaluate placement offers before accepting them by entering publisher, placement type, and position as a hypothetical scenario
Negotiate better terms by previewing how much the score lifts when you upgrade the slot, the link attribute, or the placement type
Check key message pull-through — see whether your property's key messages actually come through in the content (informational only, not part of the score)

Retrospective vs Prospective Mode

Placements have two modes. You choose the mode when creating a new placement.

Retrospective (live URL)

Paste the URL of an existing placement. Spyglasses fetches the page, isolates the main article content, tokenizes it, and walks the tokens looking for your property's name and aliases. Once it finds the brand mention, it:

Computes where it sits in the document (first 10%, middle third, final 10%, etc.)
Evaluates the 128-token chunk containing the mention plus the preceding and following chunks for self-containedness
Parses the raw HTML to detect the rel attribute of any link pointing at your brand domain
Classifies the placement type (dedicated article, roundup position, passing mention, etc.) via an AI model given the three-chunk window
Analyzes sentiment around the mention — balanced / mildly positive / strongly negative
Computes key message pull-through via embeddings if your property has key messages configured

Retrospective scoring runs as a background job and typically takes 15–30 seconds. You'll see a “Scoring placement…” indicator while the job runs.

Prospective (hypothetical)

Evaluating a placement offer before it exists? Choose prospective mode and enter the scoring inputs manually — publisher, placement type, editorial control, link attribute, and where in the document you expect the mention to land. Prospective placements are scored instantly (no URL fetch, no LLM calls).

Prospective mode also shows a “What if you negotiated better terms?” panel that re-scores the placement under several improved scenarios — moving to the first 10% of the document, upgrading to a dofollow link, promoting from a passing mention to a dedicated article — so you can quantify how much to push in negotiations.

Creating a Placement

Open the Placements tab in your property
Click New placement
Choose Retrospective or Prospective
Retrospective: paste the URL and click Score placement. The page is fetched, the brand mention is located, and the score appears within about 30 seconds
Prospective: fill in the publisher, placement type, editorial control, link attribute, and position slider, then click Create placement

For retrospective placements, you don't need to fill in any of the scoring fields upfront — Spyglasses detects them automatically from the live page. You can still correct any of the auto-detected values after scoring completes if they came out wrong.

Reading the Score

Each placement result shows:

PQS (0–100) with a tier label (Premium 75–100, Strong 50–74, Moderate 25–49, Limited 0–24)
Total Placement Value when the publisher is enriched — AIPVS × PQS / 100. This is the headline number that combines the domain-level and page-level metrics per the AIPVS × PQS formula
Five sub-score bars: Position, Chunk containment, Placement type, Editorial control, Link attribute — with each factor's weight shown alongside
Sentiment pill: the category (strongly negative → strongly positive) and the multiplier applied to the weighted sum
Scoring inputs summary: the detected placement type, link attribute, position percentage, and matched alias

When sentiment applies a multiplier less than 1.0x — especially a strongly-negative cancellation — the panel shows both the raw weighted score and the sentiment-adjusted score so you can see what the multiplier did.

The Content Viewer (Retrospective)

For retrospective placements, the dashboard renders the placement content with three overlays:

Green highlights on every occurrence of your property's brand name and aliases (same visual language as the AI Visibility Report)
Violet highlights on sentences that match one of your property's key messages above the similarity threshold — key message pull-through
Boundary markers around the three-chunk placement window (containing chunk + preceding + following) so you can see exactly which tokens we scored

Key Message Pull-Through

If your property has Key Messages configured, retrospective placements automatically run a semantic similarity check. Each key message is compared against every sentence in the placement content using an embedding model, and sentences scoring above the threshold are treated as a pull-through.

The pull-through panel shows every key message with its best-matching sentence and score as a percentage, even if the score is below the threshold. Matched entries are highlighted in the content viewer in violet; below-threshold entries are labeled “No match” so you can see how close the content came.

Pull-through is informational only. It does not affect the PQS score. It exists to tell you whether the coverage delivered your key messages, which is a different question than whether the coverage is well-positioned for AI citation.

Combining with AIPVS

When the placement's publisher has been enriched with AIPVS data, the score panel displays the Total Placement Value alongside the PQS. This is computed as AIPVS × PQS / 100 and answers the combined question: is this domain worth it AND is this placement well-positioned?

For example, a Tier 1 publisher (AIPVS 85) with a mediocre mid-article passing mention (PQS 55) delivers a Total Placement Value of 47 — solidly Tier 3. A Tier 2 publisher (AIPVS 65) with an excellent dedicated-article lead position (PQS 90) delivers 59 — higher, despite the lower-tier publisher.

Click Full AIPVS on the score panel to jump to the Publisher Lookup dashboard for the publisher's full three-layer AIPVS breakdown.

Click Share on any scored placement to mint a public share token. The share link renders the full placement report (score panel, content viewer, pull-through panel) in read-only mode at /placements/<token> — no authentication required. Share links are revocable by deleting the placement.

Tips

Audit existing backlinks quarterly: PR agencies can run retrospective scoring across a client's existing backlink portfolio to identify which placements are delivering AI citation value and which are effectively dead weight.
Score before you pitch: Use prospective mode to rank placement targets — if two outlets offer similar positioning, the PQS × AIPVS combination tells you which one is worth pushing for.
Use the lift preview: When negotiating placement position, run the prospective dry-run with and without your asks to show the client exactly what the upgrade is worth.
Key messages pay off: Configure your property's key messages before scoring. The pull-through analysis is one of the most useful pieces of feedback on whether your messaging actually made it into the coverage.
No brand mention found? If retrospective scoring fails with “No brand mention found in content”, check that your property's aliases include the exact form used in the article (e.g., “International Business Machines” vs “IBM”).

Placements

On this page