Home · Business · Sales · Pipeline & forecasting

YES

As of 13 August 2026, AI can forecast sales from an Excel spreadsheet.

This still needs a person who signs their name to it.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsApollo.io is listed as a commercial AI tool for prospect data and outreach, but no price is provided for sales forecasting.

If this goes wrong: you plan hiring, stock or targets around an inflated forecast and discover the gap after the relevant decisions have been made.

What to actually do

  1. Hand it to a person

    The route this page recommends

    A person who owns the outcome does this end to end, worth it when the failure is dear.

  2. Use a tool built for this

    Second choice

  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open the Excel workbook and check that each opportunity has a consistent revenue value, stage, expected close date and outcome field where available.
    2. Remove test rows, obvious duplicates and closed opportunities that are not genuine sales, then save a copy of the cleaned workbook without overwriting the original.
    3. Write down the forecast period, currency, sales stages, standard stage probabilities if your organisation uses them, and known changes such as pricing, territory or sales capacity.
    4. Upload the cleaned workbook to a chatbot, paste the prompt, and replace each bracketed slot with your forecast period, currency, reporting interval and business context.
    5. Compare the forecast totals and scenario calculations with the workbook and with a previous comparable period, then ask a colleague who knows the pipeline to challenge the assumptions before using it for targets or planning.

    Prompt

    Analyse the attached Excel spreadsheet and produce a sales forecast for [forecast period]. Use only figures present in the file and the assumptions I provide below. First describe the columns, date range, missing values, duplicate records and any obvious data-quality problems. Then state the historical sales trend, conversion rates by stage where the data supports them, average sales cycle where the data supports it, and the assumptions used. Produce a base forecast plus cautious and optimistic scenarios, showing the calculation for each. Separate closed-won revenue, open pipeline and weighted pipeline. Do not treat an opportunity as likely to close solely because its close date is in the forecast period. Flag records or assumptions that could materially change the result. Do not invent missing values, customers, probabilities or revenue. Use [currency] and report totals by [month or quarter]. End with a short list of decisions I should make before using this forecast. Business context and assumptions: [paste context here].

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • AI cannot tell whether a salesperson has overstated an opportunity without evidence from the customer or the account team.
  • AI cannot know that a new competitor, pricing change or internal capacity problem makes historical conversion rates unreliable unless you provide that context.
  • AI can calculate weighted pipeline from stated probabilities, but it cannot establish that those probabilities reflect your current market.
  • AI does not carry responsibility for decisions based on the forecast, so a colleague still needs to approve the assumptions and intended use.

Even on a YES, the friction has a name: judgement under ambiguity, verification cost and stakes of error.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs2
Verification1
Liability1
Effort delta2
Total8 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT forecast sales from Excel?
Yes. It can analyse a suitably structured workbook, calculate trends and pipeline scenarios, and explain the assumptions it used. Check the calculations and ask someone who knows the pipeline to challenge the commercial assumptions.
How accurate is an AI sales forecast?
There is no reliable accuracy figure without testing it against your own historical forecasts and results. The output is only as sound as the opportunity data, stage probabilities, close dates and business context you provide.
What data do I need for an AI sales forecast?
Give it historical sales, opportunity value, stage, expected close date and outcome data where available. Also provide the forecast period, currency, stage definitions and known changes to pricing, territories, capacity or the sales process.
Can AI forecast sales from a messy spreadsheet?
It can identify many missing fields, inconsistent dates and duplicate rows, but it cannot reliably decide what every ambiguous row means. Clean the workbook first and require the model to list unresolved data problems rather than filling them in.

Nearby answers

All Pipeline & forecasting rulings

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.