FRE (Forecast Recast Events)

Amazon & D2C glossary · Harpy Media

A Forecast Recast Event (FRE) is the deliberate scrapping and rebuilding of your demand forecast after a structural shock — a viral spike, a sudden collapse, a competitor exit, a platform change — because the old baseline no longer describes the world, and planning on it will take you straight into a stockout or an overstock.

What is FRE?

A Forecast Recast Event (FRE) is the deliberate scrapping and rebuilding of your demand forecast after a structural shock — a viral spike, a sudden collapse, a competitor exit, a platform change — because the old baseline no longer describes the world, and planning on it will take you straight into a stockout or an overstock.

Normal forecasting works by projecting the past; an FRE happens when the past stops being a guide. The skill isn’t predicting the shock — nobody does — it’s recognising quickly that one has occurred and rebuilding on new ground instead of defending the old curve.

Recognising when a recast is due (not just a blip)

The distinction that matters: noise is what forecasting absorbs; structure is what invalidates it. Signals of a genuine recast: weekly sales deviating repeatedly by large margins from the model’s ranges (not one spike — a new level that persists), a discrete event changing the market’s fundamentals (a competitor’s stockout or exit, a category trend inflection, a platform policy change, a viral moment), or the reverse — demand that collapsed and stayed collapsed. The recast protocol: freeze the old model, build a fresh projection from the new reality (current run-rate as the new baseline, with explicit assumptions about what caused the shift and whether it’s structural or temporary — each implies different inventory action), and decide the position you’ll take: chase it (order up, air freight if needed), ride it conservatively (small tranches), or hedge it (option to scale both ways).

Acting on a recast under uncertainty

The recast happens before anyone knows if the new level holds — so the discipline is staged commitment: first a small expedited batch to capture the moment while the signal is fresh, then a scaled order proportional to how long the new level persists, and a pre-written exit (returns, discounts, other channels) if it doesn’t. The two classic failures sit on either side: the seller who dismisses a structural shift as “temporary” and stocks out of the best month of their year, and the seller who treats a viral week as a permanent throne and builds inventory for customers who never arrive. The recast mindset is neither — it’s committing at the speed of evidence.

In practice

A kitchen brand’s gadget gets picked up by a creator and weekly sales jump from 400 units to 3,100, holding for four weeks. The recast: new baseline built at 2,400 (discounted from the spike’s peak for conservation), a first air batch of 5,000 to hold the wave while it’s hot (rank implications included in the math), an ocean order staged behind it sized to persistence, and a written exit path (bundle it into a set, funnel demand to related SKUs) if the level decays. The wave holds at ~1,900; the staged plan means no stockout during the moment, no warehouse of 20,000 units after it.

How Harpy Media helps

Event-driven re-forecasting — recognising level changes, staging orders, and pre-writing exits — is how we help clients catch demand waves instead of watching them.

FRE FAQ

What is a Forecast Recast Event?

A deliberate rebuild of a demand forecast after a structural shock — when the old baseline no longer describes current reality.

How do I know when to recast?

Persistent multi-week deviation from your model’s ranges, or a discrete market event that moves fundamentals (competitor exit, viral moment, policy change) — not one-off spikes.

What’s the safest way to act?

Stage commitments: a small expedited batch first, scaling with evidence of persistence, plus a written exit path if the new level doesn’t hold.