Growth amplifiers

A growth amplifier multiplies the forecast by (1 + boost) over a date range. A 30% boost means “expect 30% more sales than the model would predict, every day inside this range.” Useful when something the model can’t see in history is about to drive demand: campaigns, launches, known events.

The same popover also shows your shop’s promotional events: holidays and sales that come back every year and lift every view’s forecast once you switch them on.

What an amplifier does

The amplifier multiplies the forecast model’s daily prediction by (1 + boost) for each day inside the range. The boost is a percentage; 30% means 1.30x; 100% means 2x.

Because the forecast feeds the reorder math, an amplifier shifts everything downstream: the reorder point gets larger, the reorder date moves earlier, the reorder quantity goes up.

When to use one

A one-time marketing campaign. For a holiday or sale that repeats every year, such as Black Friday, add a promotional event instead: you set it up once for every view and every year.
A product launch (new collection drop, influencer collaboration).
A seasonal push the historical data doesn’t reflect: your first ever Christmas in business, or a major one-off promotion.
A known one-off (a TV mention, a viral moment you can plan around).

Not useful when the forecast model can already see the pattern in history. Recurring seasonality (yearly cycle, weekend uplift) is the model’s job, not yours.

Setting one up

Click the Amplifiers button in the Forecast toolbar. The button turns green when amplifiers or promotional events are in effect, and its number counts them. When a promotional event lifts the forecast, the button also gets a soft green glow and a ★ before the number. Under This view’s amplifiers, click Add period and set:

Boost % (15% to start with, at most 100%).
Start date.
End date.

A new period starts today and runs for two weeks until you change its dates. Click Apply. The forecast reloads with the boost, and the view shows Unsaved changes until you save it.

Setting one from the Supply view

In the Supply view, you can set an amplifier straight from the period headers:

Drag across the headers. Press on a period header and drag across the periods you want. When you let go, the Apply Growth Amplifier dialog opens with those dates. Set the Amplifier and click Apply Amplifier. Esc cancels the drag.
From one period’s card. A plain click on a header opens that period’s column card instead. Its Growth amplifier for this period button opens the same dialog for just that period.

Periods with an amplifier show a green chip in their header, for example +20%. When amplifiers overlap, the chip shows the largest combined boost on any day of the period. The column card repeats it as Growth amplifier +20% · set on this view, and a cell’s card names the days it covers, for example Growth amplifier +20% · Fri–Sun (or all week).

Status badges

Each amplifier shows one of three states:

active: today is between start and end.
upcoming: start is in the future.
past: end is behind today.

Past amplifiers stay on the list; they’re history. Delete them when you’ve collected enough data to know whether the campaign worked.

Stacking

Two amplifiers covering the same day add up. Two +15% amplifiers means +30% for that day. Promotional events add up the same way, with each other and with the view’s amplifiers. Useful when you want to compose effects (“baseline launch boost” + “extra Black Friday boost”).

Amplifiers also stack on top of whichever model the view uses, including History replay, so “replay last year, plus 10%” is a single, readable setup: pick History replay for the model and add a +10% amplifier over the range you want lifted.

What it doesn’t change

Historical data: amplifiers only affect the forward forecast.
The forecast model itself: Auto still uses your sales history to pick the model. The amplifier rides on top.
The simulation steps: the day-by-day inventory simulation just runs against the amplified daily-demand vector, so all downstream reorder columns shift accordingly.

Removing one

In the popover, click the bin icon on the amplifier’s row, then Apply. Save the view to keep the change.

Promotional events in the forecast

Above the view’s own amplifiers, the popover lists every switched-on promotional event under Promotional events (marked shop-wide), also the ones that change nothing in this view. Each row shows the event’s dates within the forecast horizon (the next two years) with their uplift, for example Nov 27, 2026 +40 % · Nov 26, 2027 +40 %, the first two and how many more. An event measured per product shows no percent there, as each product gets its own. A tag on each row says what the event does here:

Applied: the event lifts this view’s forecast.
Applied, outside this view’s date range (Supply view): the event’s next dates come after the view’s last column. Reorder dates and quantities further out still include it.
Off in this view: you switched it off for this view.
Would double-count, not applied: see When an event is not applied.
No upcoming dates: the event has no dates within the forecast horizon.
No uplift: the event’s uplift is 0 % on every date, so it lifts nothing. This happens when Fixed % is left empty, or when From past years has not been analyzed and has no fallback uplift, or measured no increase. Set an uplift on the event.
No past dates to measure (measured per product, No uplift for products without sales history): there is no past occurrence to measure each product’s uplift at yet, so no product is lifted.
Sales history still being prepared (measured per product, No uplift for products without sales history): your products’ sales history is still loading, so no product can be measured yet and none is lifted.
No sales data for the comparison period (measured per product, No uplift for products without sales history): no product on this page has sales for the whole comparison period before the last occurrence, so none on this page is lifted. A note on the row says so, for example No product on this page has sales for the whole comparison period before Black Friday 2025, so none of them gets an uplift. With Use the shop-wide uplift, the event stays Applied and the note names the uplift every product uses instead.
Needs the Enhanced plan: promotional events lift forecasts on the Enhanced and Elevate plans only. The events are listed, but the forecast does not use them, and their switches are locked. The popover links to Plans and billing.
Plan update pending: your shop is on a plan that includes promotional events, but forecasts still see your previous plan, so the events do not change the forecast yet. Contact support if this does not clear.

To leave an event out of one view, switch off the switch on its row. To leave every event out, switch off Apply promotional events in this view. Then click Apply and save the view. This only changes this view: other views keep the event, and the event stays switched on. The view keeps leaving the event out even while the event is switched off or archived, so it doesn’t come back here when you switch it on again. An event you add later applies to every view until you switch it off there. Click an event’s name to open it.

In the supply view, the column headers of periods with an applied event are tinted blue with a ★ 1 badge below the dates (the number counts the applied events). Point at the badge for the event’s name, for example Black Friday +40 % applied here, or Black Friday applied here, uplift per product for an event measured per product. A grey ☆ 1 badge marks events that are listed but not applied, for example Black Friday +40 % off in this view, Black Friday +40 % not applied: would double-count or Black Friday not applied: uplift is 0 %. A cell’s card names the event and the days it covers, for example Black Friday +40 % applied here · Fri. The column card also counts the products it lifts, for example Black Friday +40 % · Fri · lifts 10 of 12 variants, and names the ones it doesn’t.

When an event is not applied

Some views already repeat last year’s sales. Adding the event on top would count last year’s spike twice, so the event is left out there:

A view whose history period is Rolling (Past Year + Lead Time) applies no events. The popover says: Not applied: this view already repeats last year’s sales, so applying an event would double-count.
A product forecast with History replay over a whole year (365 days, give or take a week, or a whole number of years) gets no events either. The popover counts them, for example Not applied to 12 variants on this page: History Replay already repeats last year’s sales for them., and the product’s reorder date hover says Not applied to this variant: History Replay already repeats last year’s sales.

When an event may count twice

The forecast learns from the view’s history period. If that period contains last year’s event, the base forecast may already include part of its spike, and the event adds its uplift on top. In that case the event is still applied, and the popover warns you, for example:

Your history window (last 90 days) contains Black Friday 2025. The base forecast may already include that spike, so applying events on top may double-count.

When the view picks a history window per variant, the warning counts the variants instead, for example The history window of 12 variants on this page contains Black Friday 2025. The warning shows only on rows tagged Applied or Applied, outside this view’s date range.

If the warning matters for this view, lower the event’s uplift, switch the event off for this view, or pick a history period that doesn’t include the event.

Per-product uplift

An event whose uplift is measured for Each product variant (shown as Per product variant) measures each product’s own uplift at the event’s most recent past occurrence. A product’s own measurement counts in proportion to how much it normally sells in that period. A product that sells only a few units leans mostly on the event’s shop-wide uplift or its fallback (or on 0% if products without sales history get No uplift). A steady seller keeps nearly all of its own measured lift.

The popover explains it per event, for example Per product variant: uplift measured from each product’s own sales at Black Friday 2025; products without enough sales use +40 %. Before the event has run once, it says so instead: Per product variant: there is no past Black Friday to measure from yet, so every product uses +40 %. A row tagged Would double-count, not applied has no such line, as nothing applies there.

A product’s reorder date hover shows its own result under Promotional events, for example Black Friday: measured +233 % (4 vs 1.2 expected at Black Friday 2025), weight 29 % → +95 %: 4 units sold against 1.2 expected, a measurement that counts 29 %, and the uplift the forecast uses. A product that couldn’t be measured says why, for example Black Friday: no sales history before Black Friday 2025 → +40 %. See One uplift or one per product.