SnowFlurry

The question the system answers

Which television series are worth a licensing conversation this quarter — and what is the evidence?

Scope

SnowFlurry is an internally-built decision-intelligence dashboard that ranks a 300-series baseline catalogue by physical-merchandise and licensing opportunity. Nine raw signals per series feed six weighted criteria plus a confidence score; every rank is recomputed by an automated weekly run, so the licensing team starts each Monday from a current, comparable field instead of a stale deck.

Data and research used

There is no single feed for “TV merchandise demand”, so every signal is labeled inside the product with how automated it actually is — the honesty is the feature.

• Title catalogue metadataAutomated

  Refreshed by the weekly job with no human involved.

• Search interest and streaming top-10 chartsSemi-automated

  Pulled weekly where the endpoint allows; the previous value is carried forward, and flagged, when it does not.

• Public social and community footprintsSemi-automated

  Fandom size and engagement direction, refreshed on the same cadence.

• Licensing desk research, design review, ratings reconciliation, storefront checksManual research

  Analyst judgement on a research cadence — and the product says so on every figure they feed.

• Commercial demand-measurement feedsNot yet connected

  Sources the system would use under a data contract, listed openly rather than silently imitated.

The scoring model

Six weighted criteria and a confidence haircut, published in full — inside the product and here.

series_score = Σ (criterion_i × weight_i) × (0.95 + 0.05 × confidence/100)

raw_demand    (26%) = 0.55·reach + 0.30·engagement + 0.15·trend_index
actionability (20%) = 0.40·rights_clarity + 0.35·licensor_receptivity + 0.25·window_score
whitespace    (14%) = 50 + 0.85·(raw_demand − merch_supply)
visual        (14%) = 0.55·iconicity + 0.45·toyetic
licensing     (14%) = 0.40·licensor_receptivity + 0.30·rights_clarity + 0.30·program_evidence
momentum      (12%) = 0.70·trend_index + 0.30·cadence_score

confidence = 0.75·data_quality + 0.25·stability

The confidence multiplier is a haircut of up to 5% where the data is thin. Whitespace and Licensing deliberately move in opposite directions on merchandise supply: a saturated category leaves nothing to take, but it proves the rights holder can transact. The engine is pure and deterministic — it never reads a previous score, so every re-run is reproducible and historical runs stay comparable.

Features shipped

• Sortable, filterable rankings with all seven ranks and spreadsheet export
• Per-series profiles: score-contribution waterfall, cited evidence, risks, seasonality, comparables
• Automated weekly refresh writing one immutable score row per series, preserving week-over-week movement
• Human-verified commerce links with who-verified-and-when recorded
• Methodology page inside the product: every formula, source, and automation level
• Invite-based access control with an activity log

Decisions supported

• Which series justify a licensing conversation this quarter, with the evidence attached
• Whether a category's whitespace justifies entering it at all
• When to time outreach against a series' next release window
• Which signals need human verification before a pitch meeting

Limitations and assumptions

• Scores are directional research estimates for prioritising outreach — not forecasts of sales, and not guarantees that a deal will close.
• Several inputs are analyst judgement on a research cadence; the product labels exactly which, per source.
• Demand indices are internally built and reconciled against public source families, not licensed vendor measurements.
• Semi-automated signals carry forward their previous value when an endpoint fails, and are flagged when they do.