Every number in the report has a paper trail.

1. What we read

An assessment starts with your own footprint: the website and its key pages (about, values, visit, programs, giving, contact, press), the sitemap, and your feeds. The engine then discovers the channels you link to and measures the ones that expose numbers to a logged-out visitor: YouTube (subscribers, total views, videos, recent upload cadence, views per upload), Instagram (followers, following, posts, bio), TikTok (followers, likes, videos), LinkedIn (followers, listed size), Apple Podcasts (episodes and rhythm from the show feed), and Google reviews when a Places key is configured. Facebook and X hide their numbers behind a login; the report says so rather than guessing.

Then it searches for what everyone else says. Twelve to eighteen searches per assessment, tuned to your kind of organization: reviews, community discussion, news and press, video, podcasts, controversy, leadership, partnerships, events, and the category around you. Pages are fetched, deduplicated, and classified as owned (yours), earned (independent media), or community (people talking). Each page is checked to confirm it is about you and not another organization with a similar name; mismatches are excluded.

2. Claims, not vibes

From every usable page the engine extracts discrete claims: a fact, a piece of public evidence, or an inference, each with the excerpt it came from, a confidence level, the code it bears on, and a numbered id. A typical 360 carries two to four hundred claims from thirty to sixty sources. Nothing in the report may rest on anything else. Every sentence that makes an assertion cites claim ids, and the ids link to the evidence appendix and the source audit trail.

3. The Seven Core Codes

Each code is scored from three to five subdimensions with fixed weights. The scorer sees the evidence set and the rubric, scores each subdimension only from cited claims, and explains its reasoning. Software, not the model, computes the weighted score. A subdimension with no evidence is marked insufficient and left out of the weighting, unless the rubric says absence is itself the signal (a church with no visit page, a company with no pricing). The report shows every subdimension, its weight, its score, its status, and its rationale.

Scoring anchors: 90 to 100 is exceptional and independently corroborated; 75 to 89 strong with minor gaps; 55 to 74 a solid foundation with clear gaps; 35 to 54 weak or inconsistent; below 35 largely absent or working against the organization. Scores are absolute. Colors rank priority: the two lowest codes are red, the middle three yellow, and the top two green only when they clear 90, so a strong organization still sees where to focus.

4. The critic

Before you see anything, a separate pass reads the draft with one job: find what does not hold. It checks every cited id against the evidence, flags conclusions that outrun their support, lowers confidence or rewrites text where needed, and can adjust a subdimension. Its verdicts and every change it made are listed in the report's methodology section, and the whole report carries a confidence rating.

5. Benchmarks

Your composite and each code are placed on a percentile among organizations of your kind. Until enough organizations in a category have been assessed, the percentile is an estimate from a distribution we set for that category, and the report labels it estimated. As real assessments accumulate, they are blended in and eventually replace the estimate. Percentiles exist to point you at the right problem, not to rank you against a neighbor.

6. Disputes and human review

Anything in the report can be disputed with one click. A strategist reviews the claim, corrects the report if needed, and you see the resolution. Executive-tier reports are reviewed by a Redbird strategist before delivery; their notes appear in the report. Pulse subscribers get a monthly signal update and a quarterly rescan with a what-changed section that shows score movement, new evidence, and which recommendations were acted on.

7. Limits

• The engine reads the public web as a logged-out visitor. Private groups, closed communities, and platforms that hide numbers behind a login are not measured, and the report says so.
• Research runs on a date. The report is a snapshot; Pulse keeps it current.
• Organizations with a thin public record get honest low-confidence scores rather than confident guesses. Insufficient means insufficient.
• Sentiment is drawn from what was findable. A handful of reviews is a signal, not a census.

Open a sample report to see all of this on a real organization, or start your own.