Skip to content

Signal Flags

Every finding produced by a specialist is tagged with two attributes: signal strength and priority level.

Signal Strength

Reflects how well-established a finding is.

Flag Color Meaning When to use
confirmed Green Directly observed or verified against a primary source robots.txt read directly, schema validated at validator.schema.org, Bing site: check run live
unconfirmed Amber Inferred or reported — needs verification Competitor behavior inferred from citations, platform behavior reported not tested
emerging Red Pattern observed in limited cases — not yet a consistent signal New platform behavior seen in one or two instances, not yet documented across sessions

Source Credibility

Independent of signal strength, every finding also carries a credibility tier — how trustworthy the underlying source is. Strength is "how sure are we it's true"; credibility is "how good is the source." A finding can be confirmed from a weak source or unconfirmed from a strong one, so both are stated: [finding] — [strength] · [tier] · [priority].

Tier Source type
T1 Official platform documentation / first-party announcements
T2 Direct, reproducible observation (live query, robots.txt, schema validator, Bing site:)
T3 Reputable empirical study with stated sample and date
T4 Vendor / practitioner analysis — directional
T5 Anecdote / community signal

Weighting rule: when sources conflict, the higher tier governs; a T4/T5 claim is never reported as fact without a T1–T3 corroborator.

Tool fidelity: the researcher's own page fetch (HTML→markdown) strips <script> JSON-LD, so a "no schema" result from its own read is non-evidence — unconfirmed until verified at validator.schema.org or the Google Rich Results Test. Full detail in references/source-credibility.md.

Priority Level

Reflects urgency and impact on the client's AI visibility.

Level Color Meaning
critical Purple Blocks citation capability entirely — fix before anything else
high Red Significant impact on retrieval — high effort justified
medium Amber Notable gap — address after critical and high
low Gray Worth noting, low urgency — useful for backlog

How They Work Together

A finding tagged confirmed · critical is a verified blocker — act on it immediately.

A finding tagged unconfirmed · high is worth investigating further before acting — the impact is real if confirmed, but you need verification first.

A finding tagged emerging · medium is a trend to watch — not urgent enough to redirect resources, but worth tracking across future sessions.

Investigator Confidence

Separate from signal strength and source credibility, findings also carry an investigator confidence level — how certain the investigator is about the implication or recommendation that follows from the evidence.

Level Meaning
High Implication is unambiguous; inline tags are sufficient
Moderate A plausible alternative interpretation exists; consequence statement required
Low Evidence supports multiple interpretations; directional only; primary investigation needed before acting

When confidence is Moderate or Low, the finding appends a consequence statement: [finding] — [strength] · [tier] · [priority] — Confidence: [level] — [what changes if the alternative is correct]

High-confidence findings do not require the suffix. Full format and examples in references/source-credibility.md → Investigator Confidence.

Session Close Rule

A session that produces findings must close with two mandatory elements before the conversation ends:

  1. Research Ledger — confirmed findings, unconfirmed inferences, and what would change the conclusion (format in references/source-credibility.md → Research Ledger)
  2. Next Research Move — one specific action, naming the exact query, tool, or check, and what it will resolve

Neither is optional. A session that ends without surfacing its uncertainty and naming the next move has not completed its investigation — it has paused it without a handoff.

In the Research Brief

The Priority Matrix sorts all findings by priority level (critical → low) regardless of domain. This gives you a single ranked action list across the entire research session.