> ## Documentation Index
> Fetch the complete documentation index at: https://docs.retailgrid.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Sales forecast & price impact

> See the forecasted units, value, gross profit, and margin impact of a price change - over 7, 30, 90, or 180 days - before you approve it.

Retailgrid forecasts how a product will sell at both its current price and a proposed price, so you can see the financial consequences of a price change before you commit to it. The forecast shows up where you make pricing decisions - most visibly in the [Price Approval](/agents/price-approval) queue.

## What it estimates

For each product and horizon, the forecast projects:

* **Units** - expected quantity sold.
* **Value** - expected sales value (revenue).
* **Gross profit** - expected margin in money.
* **Margin %** - expected margin rate.

Each is projected at the **current price** and at the **proposed price**, so the difference is the estimated impact of the change. The impact is reported as the change on each measure - with the margin change shown in **percentage points**. Forecasts are computed both catalog-wide and per grid/store.

## Horizons

Pick the horizon that matches the decision: **7, 30, 90, or 180 days**. Short horizons react to recent momentum; longer horizons smooth over week-to-week noise and suit structural pricing calls.

## Where it appears

* **Price Approval** - each proposed price shows its forecasted units, value, gross profit, and margin over the selected horizon, so you can judge a change on its financial impact, not just its size. See [Review and approve price changes](/agents/price-approval).

## What drives it

The forecast is built on each item's recent sales rate (base demand), its price sensitivity (elasticity), and seasonality - the same demand signals behind [Price Optimization](/agents/price-optimization). Items with sparse history borrow demand signal from a cluster of similar items rather than dropping out. Every forecast row also carries a **confidence** level - High, Medium, or Low - so you can see at a glance how much to trust it.

## Common pitfalls

* **Blank or skipped forecast** - a row can come back without numbers when there's **no recommendation**, **low confidence**, or **missing inputs** for that item. That's signal about data coverage, not a bug.
* **Very short horizons on sparse items** - a 7-day forecast on a slow seller is noisy; widen the horizon.
* **Treating it as exact** - use the forecast to compare options and size impact, not as a promise of a specific number. Read the confidence level alongside the numbers.

## Related

* [Review and approve price changes](/agents/price-approval)
* [Price Optimization](/agents/price-optimization)
* [Metrics glossary](/reference/metrics)