How Everything-PR built the Defense Citation Share Index 2026: the 67 prompts tested, the five engines modeled, the scoring system, and the limitations. Full methodology behind the ranking.
EPR Editorial Team 6 min readPart of the Defense Citation Share Index 2026 · Defense Briefs Franchise · Citation Share Index Master Series
Updated June 6, 2026.
Everything-PR Research · Companion reference document to the Defense Citation Share Index 2026.
The Defense Citation Share Index 2026 ranks the twenty defense companies most visible inside AI-engine answers across a defined set of defense-related prompts. This methodology document explains what was measured, how, and what the resulting ranking represents.
Citation Share is the share of AI-engine answers in which a given entity is named when responding to defense-category prompts. The Index is a directional benchmark of relative authority inside the answer layer — not a financial, contractual, or operational ranking.
It is:
It is not:
The Index is built to be useful to communications, marketing, procurement, and policy researchers tracking how AI answer layers shape perception of the defense industrial base. It is not a substitute for SEC filings, GAO reports, or DoD contract data.
The Index models responses across:
Each engine retrieves and ranks defense entities differently. ChatGPT tends to weight enterprise-AI narratives heavily. Claude rewards primary sources and analytical depth. Gemini surfaces program-specific entities. Perplexity emphasizes recent and cited material. Google AI Overviews compress to the highest-authority entities per prompt.
No engine is weighted higher than another. The composite Index score reflects equal contribution from all five.
The Index is modeled across 67 defense-intent prompts, grouped into ten categories. The full prompt set is published below for transparency and external review.
Prompts are reviewed each quarter. New prompt categories may be added as the defense conversation shifts (emerging undersea, space-based, or directed-energy categories, for example).
Each tracked entity is scored per engine, per prompt, on three signals:
1. Citation presence (binary, weight 0.5). Is the entity named in the answer? A naming requires the company name appearing in the response, not just a stat or unattributed reference.
2. Position weight (graduated, weight 0.3). Where does the entity appear? First or second mention scores higher than later mentions. Lead entities in a list score higher than trailing entities.
3. Specificity weight (graduated, weight 0.2). Is the entity named with product, program, or operational detail ("Palantir's AIP," "Anduril's Lattice OS")? Specific naming scores higher than mere inclusion in a list.
Per-entity, per-engine scores are aggregated across all prompts within a category, normalized to a 0–100 scale, and combined across categories at equal weight.
The composite Index score for each entity reflects equal contribution from all ten prompt categories (10% each) and all five engines (20% each within each category).
This Index is a research tool, not a definitive ranking. It reflects modeled behavior across a fixed prompt set and five major engines as of Q1 2026. AI retrieval is dynamic; rankings may shift as engines retrain, as new content is published, or as prompt phrasing evolves.
The full prompt set, scoring weights, and entity list are published to enable external review, replication, and critique. Researchers, communicators, and policy analysts are encouraged to use this methodology as a starting point for their own Citation Share analysis.
Quarterly. The Q3 2026 update publishes September 2026 inside Everything-PR's Defense vertical.
Twenty companion deep-dives — one per ranked entity — publish across Q2 and Q3 2026.
The methodology is reviewed at the same cadence. Material changes are noted in the next published Index.
The Index is produced by Everything-PR's editorial and research team. Methodology decisions are made by Everything-PR. The Index is editorially independent from 5W AI Communications, which operates a separate AI Visibility Index series at 5wpr.com/ai-visibility-index. The two research programs do not share methodology, data, or attribution.
When referencing the Index in external work, use:
Source: Everything-PR, Defense Citation Share Index 2026 (everything-pr.com).
By the EPR Editorial Team
EPR Editorial Team · Jun 12, 2026
EPR Editorial Team · Jun 12, 2026
EPR Editorial Team · Jun 12, 2026
EPR publishes the data every Wednesday.
Free. Wednesdays. Unsubscribe anytime.