The task
You have a chunk of operational data (Zendesk tickets, Mixpanel events, sales call logs, support transcripts) and a vague feeling something is happening. You need an answer that’s actionable: not “engagement increased,” but “engagement increased 12% on Android, mostly from session 2-3, likely the new onboarding.” The risk is asking “analyse this” with no question; AI then produces a beautiful summary nobody can use.
When AI helps — and when it does not
AI is excellent at pattern-finding on structured or semi-structured data and at writing the “so what”, once you give it a question and a decision. It is poor at picking the question for you. The two-line rule: tell AI what you’re trying to learn, and what decision the answer informs. Otherwise you get a polished restatement of the data.
What to feed the AI
- Aggregated data or a sample (CSV, table, ticket text)
- The specific business question you are answering
- The decision the analysis informs
- Time window and segmentation
- Known data caveats (sampling, instrumentation issues, recent changes)
- What “good” looks like in this metric
Copy-ready prompt
Analyse operational data.
Question (what I want to learn): <line>
Decision (what I will do with the answer): <line>
Time window: <line>
Segmentation: <list>
Known data caveats: <list>
What "good" looks like: <line>
Data (sample or aggregated):
"""
<paste>
"""
Return:
1. One-line answer to my question (in plain language)
2. Three supporting data points — each with a number and the segment it came from
3. One caveat about the data — sampling, recency, instrumentation
4. The single next analysis I should run before acting
5. A "would I bet $X" check — at what confidence level should I act on this?
6. The action I should take if confident, paused if uncertain
Do not produce a summary. I want an answer, supported by data.For multi-question analysis: “Now produce a 200-word executive memo from this analysis — the message, the action, the next step.”
Recommended output structure
A one-line headline answer, three supporting points, a caveat, a next analysis, and the action decision. Skip executive summaries; readers want the answer, not the data tour.
How to check the output is usable
- The headline answer addresses your question, not a related one
- Supporting points have numbers and segments, not vague claims
- The caveat is real (sampling, instrumentation, recent changes)
- The next analysis is specific (a query you can run, not “look into it”)
- The action is binary: do, do not, or defer until data X
Common mistakes
- Asking “analyse this” with no question. You get a tour, not an answer
- Skipping data caveats. Your reader will quote the number without them
- Not stating the decision. AI optimises differently when stakes are clear
- Trusting AI on numbers. Verify the headline metric
- Acting before the next analysis. Your confidence level was lower than you thought
Practical depth notes
For How to Use AI to Analyse Business Data: From Operational Noise to a ‘So What’, the difference between a usable AI result and a generic one is the input packet. Give the model the audience, the current draft or raw material, the desired format, the decision you need to make, and two examples of what good and bad output look like. Ask it to preserve facts first, then improve structure or wording second.
After the first response, do a separate review pass. Look for missing constraints, invented details, weak calls to action, and language that sounds plausible but does not match the real situation. The best final output should be easy to use immediately: clear owner, clear next step, and no hidden assumption that someone else has to untangle. A stronger version of this workflow also defines the handoff. Decide who will use the output, what they should do next, and what information would make them reject it. If the deliverable is copy, test whether it has a single clear action. If it is analysis, test whether it separates observation from recommendation. If it is planning, test whether dates, owners, and tradeoffs are explicit enough for someone else to execute. One final check: compare the finished result against the original goal in a single sentence. If that sentence is hard to write, the output is probably polished but unfocused. Tighten the goal, remove decorative language, and rerun only the weak section instead of regenerating the entire piece.
FAQ
- What about confidence intervals? Ask AI to estimate sample size and noise. Always do back-of-envelope sanity check.
- Can AI write SQL? Often yes, for verification, not as final query.
- Should I pipe AI into a BI tool? Useful for read-only exploration. Production analytics still need stable queries.
Related
- Excel analysis prompts — analysis on spreadsheet data
- Survey analysis — when the data is survey responses
- Funnel analysis readout AI — funnel-specific
- Retention cohort readout AI — cohort-specific
- Experiment interpretation — when data is A/B test
- Chart table explanation AI — communicate the analysis
- ChatGPT data analysis workflow — broader workflow
Tags: #AI writing #Data analysis #Finance #Business analysis