Home · Business · IT, Data & Security · Data analysis & reporting
As of 13 August 2026, AI can only partly track your sales KPIs.
This still needs a person who signs their name to it.
Can you do it?
15 minutesto a draft.
1 hourto something you’d act on.
Cost, all in£0
Skill neededpower-user
Who has to check ita colleague
What the alternative costsA purpose-built reporting setup or analyst is the alternative; the supplied tool information gives no price for either.
If this goes wrong, you act on a plausible but incorrect sales picture because a source, definition or calculation was wrong.
What to actually do
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.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, power-user skill, and roughly 1 hour until you can act on the result.
How to actually do it
- Open your CRM or sales spreadsheet and export the relevant opportunity, order or revenue records with dates, amounts, stages, owners and any fields used in your KPI definitions.
- Write down the KPIs you currently use, their reporting period, target and exact definition, including how you treat cancellations, refunds, duplicates and deals still in the pipeline.
- Remove unnecessary personal data from the export, save the file with its extraction date, and record which source system and filters produced it.
- Open Julius AI, upload the export and paste the prompt above, then answer its questions about ambiguous KPI definitions before accepting any calculations.
- Compare the report's row counts, revenue totals, closed-won totals and sample KPI calculations with the CRM or source spreadsheet for the same period.
- Ask a colleague who knows the sales process to check the definitions, exclusions, unusual movements and charts before circulating the report.
- Save the confirmed KPI definitions and update procedure, then repeat the same export, upload and comparison process for the next reporting period.
Prompt
I need to track sales KPIs for [team, business or territory] using the attached data files. Use only the data provided and do not invent missing values. First list the available columns, reporting dates, row counts and any duplicate, missing or inconsistent records. Then propose KPI definitions using the available fields, including the exact numerator, denominator, time period and exclusions for each KPI. Ask me to confirm any definition that is ambiguous before calculating it. After confirmation, calculate the KPIs by [week or month] and, where the data supports it, by [salesperson, team, product or region]. Show the underlying totals as well as the percentages and label every result with its data period and source file. Create clear tables and charts, identify material changes from the previous period, and separate observed results from explanations or predictions. Flag stale data, missing fields, unusual values and any result that cannot be checked from the supplied data. Do not describe the figures as live unless the source data includes a current timestamp. Finish with a repeatable update procedure for the next reporting period and a short list of checks a colleague should perform before using the report.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- AI cannot know whether your organisation's definition of conversion rate, pipeline or revenue is the one management actually uses.
- It cannot make the figures live unless the source data is connected and updated, and it cannot prove that an export contains every relevant record.
- It cannot reliably distinguish a genuine sales change from a CRM process change, missing data or a change in how staff enter records.
- It cannot take responsibility for decisions made from an incorrect dashboard or report.
- It cannot replace a colleague who understands the sales process and can challenge an implausible result.
What caps this at PARTLY: real time truth, verification cost and context depth.
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.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 1 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 2 |
| Total | 7 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT track my sales KPIs?
- Partly. It can calculate and explain KPIs from a spreadsheet or export, but it cannot guarantee live data, correct business definitions or complete CRM records. You need to verify the totals and have a colleague check the logic before using the report.
- What data does AI need to track sales KPIs?
- It needs consistently structured sales records, reporting dates, amounts or values, stages, owners and any fields required by your chosen KPIs. Give it the KPI definitions, targets, exclusions and source-system date as well, and remove unnecessary personal data.
- Can AI connect to my CRM and update sales dashboards automatically?
- Some purpose-built analytics products can work with connected or repeatedly uploaded business data, but the exact connection depends on your CRM and setup. Treat a dashboard as current only when you can see a successful refresh time and reconcile its figures with the source system.
- How do I know if an AI sales report is accurate?
- Reconcile its row counts, revenue totals and closed-won totals with the CRM for the same period, then manually recalculate a sample of each KPI. Have a colleague check the definitions, exclusions, missing data and any large movement before the report is shared.
Nearby answers
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.