Build a simple forecast from your own numbers with AI
You have the numbers, you just never pull the answer out of them, so you run on a hunch about money. Here is how to use AI to build a simple, sense-checked forecast from figures you already have.
You have the numbers. Months of sales sitting in a spreadsheet, every figure real. And yet when someone asks what next quarter looks like, you find yourself shrugging.
Not because the answer is not in there. It is. You just have never had the time, or frankly the appetite, to sit and pull it out.
So you run the business on a gut feel about money, which works right up until the month it does not, and a surprise you could have seen coming lands anyway.
Why forecasting feels out of reach
Forecasting sounds like something for finance teams and clever software. It carries a whiff of spreadsheets with formulas you half remember and mostly avoid.
So a small business runs on the bank balance and a hunch. That is understandable, and it leaves you reacting to your numbers instead of reading what they are quietly telling you.
You already have the raw material
Here is the good news. You do not need clever software or a finance degree. You need the numbers you already have and a way to ask a plain question of them.
Export your monthly sales, or your bookings, or whatever the number is that matters, into a simple table. A date and a figure is enough to begin.
Hand that to a model that can actually work with data, and now you can just ask, in words, the questions you have been avoiding.
Ask it like a person
Upload the figures and talk plainly. Here are my monthly sales for two years. Based on the trend and the seasonal pattern, what might the next three months look like? Show me your working.
That last line is the important one. You are not after a magic number. You are after a sensible estimate you can see the reasoning behind, so you can judge whether it holds up.
It will spot the things you feel but never quantified. That every January dips. That summer lifts. That the trend under the noise is gently up, or quietly down.
A simple forecast is really just a trend line plus a bit of seasonality, and the maths is old and well understood. What is new is that you can now ask in plain English and the model writes and runs the calculation for you, in a tool like ChatGPT's data analysis mode, where it works in actual code behind the chat rather than guessing numbers in its head. That distinction matters. When it runs code on your figures the arithmetic is real and repeatable. When a chat model estimates a total from memory it can drift, so the habit worth keeping is to make it show the working, the assumptions and the formula, so you can sense-check the story rather than trust a number on faith.
A forecast is a conversation, not a fact
Treat the number as a starting point, not a promise. No forecast is right, and the useful ones are roughly right in a way that changes what you do this week.
Push on it. What if my best client leaves? What if I raise prices five percent? Now you are not forecasting, you are planning, testing decisions against evidence instead of nerves.
What you get back
A rough sense of what is coming is worth more than a perfect view of what already happened. It turns money from a thing that surprises you into a thing you can steer.
I work alongside these tools every day, and watching a business owner go from shrugging at next quarter to quietly planning for it is one of the more satisfying shifts to be part of.
If you would like to stop guessing at next quarter and actually read what your numbers are telling you, that is a great thing to set up together. Book a quick chat and we will build one.
Book a quick chat →Related: Better decisions from your own data.
Common questions
Do I need finance skills to forecast with AI?
No. You need the numbers you already have, a date and a figure per row, and a model that can work with data. You ask your questions in plain English and it does the maths. The skill is not in the spreadsheet, it is in asking sensible questions and sense-checking the answer.
What numbers do I need to build a simple forecast?
Less than you think. A monthly figure that matters, sales, bookings, or revenue, going back a year or two, laid out as a simple table of date and amount. The more history you have the better it reads the seasonal pattern, but even a year gives the model something useful to work with.
Can I trust an AI forecast?
Trust it as a rough guide, not a promise. Ask it to show its working and its assumptions so you can sense-check the story rather than a bare number. Use a tool that runs actual calculations on your figures rather than estimating in its head, and treat the result as roughly right, not exact.
What is the point of a forecast if it is not exact?
A forecast that is roughly right still changes what you do this week. It warns you about a slow month before it arrives and lets you test decisions, like raising prices or losing a client, against evidence instead of nerves. Being approximately prepared beats being precisely surprised.