The Triangulation gate

Retailer intake figures are rarely neutral: they're anchored on the buyer's enthusiasm, negotiation posture, or last year's comparable.

The core problem isn't "forecast the SKU", it's arriving at a defensible independent number to interrogate the retailer's seasonal volume before you commit cash to a 15-week buy. 

Your job is to build a bottom-up estimate from first principles and triangulate it against

their top-down number, then treat the buy quantity as an economic decision (cost of over- vs under-stocking), not a single forecast.

For licensed goods specifically, you have an advantage most NPI categories lack: the licence itself throws off measurable leading demand signals (search, social, release calendars, franchise momentum) that exist before any EPOS. Exploit them.

The strongest practical approach is a layered one: analogue-based rate-of-sale × distribution math, adjusted by licence-heat signals, expressed as a probabilistic range, and reconciled against the retailer's figure through a structured triangulation gate.

Key analysis, the methods worth using

1. Distribution × Rate-of-Sale decomposition (your backbone)

Never forecast a "season total" directly. Decompose it:

Units = Doors × Rate of Sale (units/store/week) × Weeks on shelf × Sell-through assumption

This forces every assumption into the open and lets you challenge the retailer's number in their language. If their seasonal volume implies an ROS of, say, 4 units/store/week and your best analogue did 1.8, you've found the argument — not a gut feeling, a testable claim.

2. Comparable analogue forecasting (the highest-value, lowest-cost method)

Find 3–5 prior SKUs that are genuinely comparable. Same licence family, price band, format, retailer, and equivalent point in the franchise lifecycle.

Use their actual launch curves and ROS as the anchor.

The discipline is in analogue selection: match on attributes (character, format, price, seasonality, media support), not on "it feels similar".

A structured attribute-weighted approach beats a single hero comparable.

3. Licence-heat / leading-signal overlay (the licensed-goods edge)

This is where you differentiate from a generic planner. Before EPOS exists, demand for licensed product is visible in:

• Release calendar: film/streaming/game launch dates relative to your on-shelf date. Demand for character goods is heavily event-timed; being 6 weeks early or late materially changes the curve.

• Search & social momentum: Google Trends trajectory, TikTok/YouTube view velocity, wish-list/pre-order volumes.

• Franchise lifecycle stage: new franchise (diffusion/ramp), peak (event-driven spike), or evergreen (stable baseline with seasonal lift). Each implies a different curve shape and decay rate.

Use these not as a precise multiplier but to bias the analogue up or down and to set the shape of the curve.

4. Probabilistic range, not a point (essential at 15-week lead time)

Produce a forecast range, not a single number. The spread itself is information: a wide spread on a big commitment tells you to phase the buy or negotiate risk-sharing rather than commit blind.

5. Newsvendor framing for the buy decision

The forecast informs the buy; it doesn't equal it.

Set initial buy by weighing underage cost (lost sales, retailer service failure, relationship damage) against overage cost (markdown, obsolete licensed stock, which is often unsellable once the licence cools). Licensed overstock is often more punishing than generic overstock. It often can't be carried to next season. That asymmetry should shape whether you lean toward or below the mid range forecast.

6. Be sure to de-bias judgement.

Buyer commitment, pre-orders and retailer intent are legitimate inputs but capture them structured and separately from the analogue, so enthusiasm doesn't silently inflate the base.

Record assumptions so you can measure forecast value-add later and build proprietary calibration over time.

Triangulation gate: how to actually make the "initial call comparison":

Step	Action	Output
1	Build bottom-up (doors × ROS × weeks) from analogues	Your independent P50 + range
2	Convert retailer's seasonal volume into implied ROS/store/week	Apples-to-apples comparison
3	Overlay licence-heat signals to adjust curve shape/level	Adjusted estimate
4	Compare the two; quantify and explain the gap	Reconciliation
5	Set buy via over/under cost asymmetry, phase if possible	Initial buy recommendation

Risks / considerations

• Analogue scarcity/bias: poor comparable selection is the biggest failure mode. Garbage in = confident-but-wrong forecast out.

• Licence signals are noisy and can peak before shelf date: social heat ≠ purchase intent and timing decay is brutal for licensed goods.

• Retailer number is not independent of yours: if they've seen your forecast, triangulation is contaminated. Build yours before anchoring on theirs.

• Long lead time removes your safety net: 15 weeks means limited or no in-season replenishment, so error is expensive in both directions. This raises the value of phasing and of range-based decisions.

• Point-forecast temptation: a single number feels decisive to a buyer but hides the risk you're carrying.

Recommended actions (prioritised)

1. Build the decomposition model first (doors × ROS × weeks × sell-through). This alone lets you challenge any retailer number credibly. Highest value, lowest cost.

2. Assemble an analogue library for your key licence families with real launch curves and ROS this becomes reusable IP and improves every future call.

3. Add a lightweight licence-heat scorecard (release timing, search trend, social velocity, lifecycle stage) to bias the base estimate.

4. Express every forecast as range and make the buy decision using over/under cost asymmetry, not the midpoint.

5. Where lead time allows, split the commitment: initial buy + option/top-up to convert forecast risk into a staged decision.

If resources are tight, do 1 and 2 only.

(The decomposition plus a decent analogue is 80% of the value.)

Questions that would improve this:

• Roughly how many doors and what shelf duration are typical for these launches, and is any in-season replenishment possible within your 15-week window?

• Do you have access to prior comparable licensed SKUs' actual sell-through/ROS, or would analogues have to be estimated?

• What's the typical overage penalty? Can licensed overstock be liquidated, carried, or is it a full write-off?

• Is the retailer's seasonal volume a firm commitment (they own the markdown risk) or indicative (you carry it)? This changes the buy plan entirely.