Forecasts
What will sell next month, what will run out, and which accounts are slipping, each with a range and a recommended action. Built on the same numbers as your dashboard.
What a forecast gives you
Sales by month and by line
Next quarter's revenue by product line, with the range it is likely to land in, so purchasing and staffing follow the numbers instead of last year's gut feel.
Stockout dates by part
Which parts run out and when, at the current sales rate and lead time, with a reorder quantity attached.
Accounts at risk
Which customers are buying less than their pattern says they should, early enough to call them.
The action, not only the number
Each forecast carries a recommendation: reorder 40, call these three accounts, hold the hire a month. You decide, and the page shows the reasoning.
What it needs
About two years of history in a system we can read, and enough volume for a pattern to show. A shop with fifty invoices a month has a pattern. A business with five big deals a year does not, and we will say so.
The numbers have to tie to the books first. If they do not, the Starter dashboard comes before the forecast, because a forecast built on doubtful data is worse than none.
How we build it
We fit the model to your history, then test it against last year as if we had not seen it, and show you how close it came. You see the error before you trust the output.
Every figure carries a range, never a single number dressed up as certainty. The model reruns monthly inside a data team plan, so it learns from each new month.
What it costs
From $6,000, quoted fixed after a paid scoping step that checks the history first. The scoping fee comes off the build if you go ahead within 60 days.
Monthly reruns and tuning sit inside the data team plan. Every price is on one page.
Before we start
What we ask
- How far back the history goes, and in which system
- How many records a month: invoices, orders, parts moved
- The decision the forecast drives, and who makes it
- How often it has to update
- What changed in the business inside that history: a new location, a lost customer, a price change
What we need from you
- About two years of history in a system we can read
- Numbers that already tie to the books, or a Starter dashboard first
- The decision maker on a 30-minute call before scoping
- Lead times for the parts you want stockout dates on
What will not work
- Fewer than two years, or fewer than two seasons, of history
- Data that does not tie out yet
- One-off events: a hurricane, a one-time contract
- A single exact number with no range