Forecast

Forecasts that show their work.

Per-SKU, explainable forecasts that decompose into base rate, trend, seasonality, and events. Every number reconciles line by line, before you buy on it.

  • Every figure reconciles to its drivers
  • 80% confidence band on every SKU
  • Anomalies excluded and shown
HTX-41 · demand forecastSample
212

units

next 30 days

80% confidence bandreorder 168 u22 d cover

how it is built

Base rate128 u/mo
128
Trend+6%
136
Seasonality · Jul×1.40
190
Event · SUMMER-15+22
212
TODAY
Made for Shopify brands
ShopifyBrightpearlCSV / StockyKlaviyoSoonAmazonSoonNetSuiteSoon

What it does

Everything a forecast should tell you, on every SKU

Not a category average and a shrug. A decomposed, defensible number for every SKU and variant you sell.

Explainable drivers

Open any forecast to its base rate, trend, seasonality, and event lifts. No black box. Every figure reconciles to the one above it.

Confidence, not false precision

Every SKU comes with an 80% range, so you can buy against a band you understand instead of a single guess you can't.

Seasonality & events

Peaks, promos, and holidays are modeled per SKU. Add a sale or a launch and the shape of the year updates with it.

New-product cold-start

Forecast products with little or no history from comparable SKUs and category signals, so launches don't start blind.

Plan vs. actual, live

A board that replans as sales land, with lockable closed periods so your history stays honest and your plan never drifts.

Per-SKU & per-variant

Variant-level demand at size, color, and pack level, not a blended parent number that hides the SKU about to stock out.

Explore

See forecast in the product

Tab through the capabilities, or let it play. Every panel is a live sample surface.

Explainable

Every number reconciles, line by line.

Most tools hand you a forecast and ask you to trust it. Sellcast shows the arithmetic: base run-rate, the trend on it, the seasonal multiple, the event lift, and the anomalies it threw out. If you don't agree, you can see exactly where to push.

  • Base → trend → seasonality → events, each with its running total.
  • Anomalies (stockouts, one-off spikes) excluded and shown, not silently smoothed.
  • The same math your team can audit in front of the board.
forecasts / HTX-41
Sample

Forecast, HTX-41 · Jul 2026

212uAI forecast · next 30 days

shaded = 80% range

  • 1Base rate128 u/mo × 1= 128
  • 2Trend+6%= 136
  • 3Seasonality (Jul)×1.40= 190
  • 4Event lift, SUMMER-15+22= 212
  • ·Anomalies excluded2
Forecast, next 30 days212 units

Explainable AI

See the reasoning behind every number.

Sellcast doesn’t just predict. It explains. Every forecast keeps the signals it used, the anomalies it excluded, and the confidence it holds, so you can defend the buy and adjust any input you disagree with.

  • Grounded, Recommendations come from your own sales, inventory, and promo history.
  • Auditable, Every figure traces back to the data and the driver behind it.
  • Yours to adjust, Override any input. Nothing the AI produces is final until you approve.
forecast / why this number · HTX-41
Sample

212u Jul forecast

80% CI · 190–236u

Signals used

  • Base run-rate· trailing 90 days128 u/mo
  • Seasonality· 3 prior Julys×1.40
  • Trend· 12-week velocity+6%
  • Event lift· SUMMER-15 on calendar+22u
  • Excluded, 2 stockout weeks (Mar)anomaly
Audit. You set trend to +4%, then reverted · 2 days ago

How it works

Live in a day, not a quarter

1

Connect Shopify

Your sales history, products, and inventory sync in minutes, or import from Stocky in one click.

2

The model decomposes demand

Sellcast learns base rate, trend, seasonality, and your event calendar, per SKU and variant.

3

Forecast you can open

Every number arrives with its drivers, a confidence band, and the buy it implies, ready for your approval.

Why it's different

Forecasting the old way vs. Sellcast

 Spreadsheets & exportsSSellcast
Where it livesStatic exports, re-keyed weeklyLive on your Shopify data
ExplainabilityA number you can't questionEvery figure reconciles to its drivers
UncertaintyA single point guessAn 80% confidence band per SKU
New productsBlind until they have historyCold-start from comparable SKUs
UpdatesManual, when someone has timeReplans as sales land
The next stepRe-type it into a POThe buy is drafted for you

One AI loop

Forecast is one stage. Sellcast closes the loop.

Each stage feeds the next. The forecast drives the buy, the stock position drives the promo, and what won't sell gets gated for markdown.

Your systems

ShopifyBrightpearlCSV / StockySuppliersComing soon

One AI loop

Decisions, back to you

  • PO drafted→ Brightpearl
  • Promo scored→ you run in Shopify
  • Markdown gatedat your margin floor
  • Forecast refreshed+ cash-flow
  • Stockout alert12 days of cover

FAQ

Forecast, answered

What data do the forecasts use?

Your own Shopify sales history, product velocity, seasonality, and your promo calendar. If you use Brightpearl or import from Stocky, purchasing and cost history come in too. Every forecast expands to show which signals drove it.

How accurate is it, really?

We don't wave a single accuracy number at you. You can't verify it. Instead every forecast ships with an 80% confidence band and is backtested against your own history, so you can see the range and judge it for yourself before you buy.

Can it forecast a brand-new product with no history?

Yes. Cold-start forecasting uses comparable SKUs and category signals to give a defensible starting number, then sharpens it as the product's own sales come in.

Can I override or adjust a forecast?

Always. You can adjust any driver or override the number outright, and Sellcast keeps an audit trail of the change. Nothing the AI produces is final until you say so.

How often does it update?

Continuously. As sales land, the plan-vs-actual board replans the forward periods; closed periods you've locked stay immutable so your history never gets rewritten.

See forecast on your own data.

Connect Shopify and your first forecasts appear immediately. Book a demo to watch the whole loop run on a sample of your catalog.