What this method produces. One strength verdict per signal — watch, test, or act — built from three scores you can defend out loud. Prerequisite: the signal already has Source Cards; you cannot score what has no passport.
Step 1 — List the ORIGINS, not the mentions. Write down every place the signal appeared, then collapse echoes: reposts, reaction videos and articles quoting the same moment all fold back into one origin. What remains is your true source list.
Step 2 — Score INDEPENDENCE (0-2). 0 — one origin. 1 — two or three unrelated origins. 2 — four or more origins that could not have copied each other (different cities, different scenes, no shared trigger).
Step 3 — Score REPETITION (0-2). Look at the dates on your Source Cards. 0 — one moment in time. 1 — recurs across several weeks. 2 — sustained or growing across months. A tall one-week spike still scores 0 here: height is not duration.
Step 4 — Score SPREAD (0-2). 0 — lives in one bubble (one city, one platform, one micro-scene). 1 — crossed one border: a second country, a second platform, a second age group. 2 — crossed the commercial border: from early adopters into mass-market drops, mainstream retail or mainstream media.
Step 5 — Verdict by total. 0-2: WATCH — set a check-back date, spend nothing. 3-4: TEST — a capsule, a content probe, a limited buy; define in advance what result upgrades it. 5-6: ACT — the evidence is broad; the strategic question is no longer "is it real" but "is the window still open for us".
Step 6 — Write the verdict as one defensible line. "Independence 2, repetition 1, spread 1 = 4: test" — plus the single fact that most limits confidence ("all origins are European; no data outside EU").
Common mistakes. Scoring loudness instead of counting origins — a viral week can be 0-1-0. Rounding up because you personally love the signal — taste chooses what to score, never the score itself. Skipping the limiting fact in step 6 — a score without its weakest point is advertising, not analysis.