Skip to content
Browse topics

Predictive analytics - sales forecast, what will run out and expected collections

Every night Nori works out how much of each item you will sell, what runs out before your purchase arrives, how much you will collect each week and which customers bought less than expected.

Updated November 17, 2026

On this page

Predictive analytics uses your own sales and payment history to tell you what is coming. Nothing to set up and no other service to hire: Nori calculates it every night from what you already record.

Go to Analysis › Predictive analytics. At the top you see when it was last calculated. If you just recorded a lot, use Recalculate and you will have the new numbers in a few minutes.

Sales forecast

Shows how much you expect to sell of each item in the coming weeks.

  1. Go to Sales forecast. Items are listed by what sells most.
  2. Search an item by code or name and click it.
  3. At the top choose the warehouse (or all) and whether you want it weekly or monthly.

The chart shows what you sold as a solid line and the forecast as a dashed line. The colored band is the likely range: 8 times out of 10 actual sales land inside it. Below is the table with the forecast and the likely low and high of each period.

What counts as a sale: invoices, what was delivered and not invoiced yet and, subtracting, returns and credit memos. A delivery invoiced later counts once. It is the same count the sales reports use. Everything is in the item's inventory unit.

To plan purchasing and production with this forecast, load it into an MRP forecast with Fill with the calculated forecast (see Material planning).

How the method is chosen

Nori tries several methods and keeps the one that missed the least:

  • Average of the last weeks: for steady sales or little history.
  • Average that weighs recent weeks more: follows changes without jumping with every week.
  • Trend: for sales going up or down.
  • Season: repeats last year's pattern. Needs two years of history.
  • Occasional sales: for items sold now and then.

To try them, Nori hides the last 8 weeks, forecasts them from what came before and compares with what actually sold. The average error is how far off it was: units over or under divided by units sold. 20 % means it was off by 2 units for every 10 sold. If the item has little history you see Little history and the average of the last periods is used.

Why: every item sells differently. Instead of one formula for all, Nori picks the one that would have worked best for that item.

Running out

Lists the items that run out before your next purchase arrives, per warehouse, within the next 60 days.

For each one you see:

  • Available: stock on hand less what is committed in sales orders.
  • Daily sales: what is expected to sell per day according to the forecast.
  • Lasts: available divided by daily sales, in days, and the date it runs out.
  • Purchase arrives: the delivery date of the nearest open purchase order (or its date, if it has none) and how much it brings.
  • Short by: the units you will not be able to sell until the purchase arrives.
  • Why: Runs out before it arrives or Nothing on order.

Filter by warehouse at the top right. Click an item to see its forecast in that warehouse.

Expected collections

Tells you how much you will collect in each of the next 8 weeks and which customers are most likely to pay you late.

For each open invoice, Nori takes its due date and adds the days that customer usually takes to pay. If the customer has no history, it uses your customers' average. What is already overdue is expected this week. What falls after 8 weeks shows as Later.

The risk of paying late goes from 0 to 100 and comes from three things:

  1. How many of its invoices of the last year it paid after the due date. With few invoices it leans toward one half, so nobody is judged by a single one.
  2. How much it owes overdue today and for how many days.
  3. Whether over the last 3 months it pays later than before.

Below 35 is Low, 35 to 59 Medium and 60 or more High. The Why column shows the reasons with their numbers, for example Paid 4 of 6 invoices late (67 %).

For the detail of what you are owed by days past due use the aging report.

Sales alerts

Shows the items and customers whose last 8 weeks went off what was expected, for example López Groceries bought 40 % less than expected.

An alert only appears when the change is 30 % or more and falls outside the likely range of those weeks. That way the usual ups and downs do not flood you with notices. Items count units; customers, the amount before tax.

Filter between Items and Customers. Click an alert to open the item or the customer.

Ask the assistant

The AI assistant reads the same as these screens, with your permissions. You can ask it:

  • What will I run out of?
  • How much will I collect in the next weeks?
  • Sales forecast of item CAFE-500
  • Which customers are buying less?

Who sees it

It takes the permission See predictive analytics (in Users and security › Roles and permissions). It comes on in the roles that see reports. Each screen also asks for the permission of what it shows: items for the forecast, inventory for what runs out and invoices for collections. A user limited to some warehouses only sees those warehouses.

Related articles

Can't find something?

Tell us what you need to do and we'll get back to you. If you're already a customer, include your company's name.

Contact us