Advanced · Data and analysis · module 136 of 155
From question to analysis plan
Turn a loose question into an analysis plan with definitions, source, period and method, before you let AI process figures.
After this lesson
You formulate a sharp analysis question, set out definitions and source, and let AI help with a plan you can check.
In view · Step by step
From question to analysis plan
Metric, group, period, comparison and the decision that depends on it.
What counts, from which export and of what date.
Set down calculations, chart and pitfalls before you look at the figures.
Part 1 of 3 · A sharp question
Your preference for less movement is on. Choose each part yourself.
A vague question gets a vague answer
If you ask AI “how are our sales doing?”, you get a general answer that could apply to anything. An analysis question is concrete: which measure, for which group, in which period and compared with what? For example: how many new registrations did we get per month in 2025, compared with 2024? Also write down why you want to know and which decision depends on it. That determines how precise the answer needs to be (lesson 78).
Set out definitions and source
What counts as a new registration: every application, or only those who have paid? A different definition gives a different figure, even if the question seems the same. Choose the source, for example an export from your membership system, and note the period and the date of the export. First check the data for empty fields, duplicate rows and strange values (lessons 44 and 107). AI can help to detect such problems, but does not decide by itself what happens to them.
Let AI think along, not decide in advance
Ask AI for an analysis plan: which calculations, which comparisons, which chart and which pitfalls. Do not let it guess an outcome before the data is there. Fix the plan before you start, so that afterwards you do not keep searching until something interesting happens to turn up. Report the definitions, the source and the limitations with the result. A small group or a short period makes any conclusion more tentative (lesson 139).
New practice example
From explanation to a usable result.
A good analysis starts with a sharp question, fixed definitions and a known source, before any figure appears.
Fictional practice situationFictional: a sports club wants to know whether the new training slot is working. The chair asks: “Are the members more satisfied?”
View a possible solution
This is how it can be done
Sharper question: how many members who started in September still came to train at least twice a month in October and November, compared with last year’s group? Source: the attendance lists. Definition: present according to the trainer’s list. Limitation: the weather was exceptionally bad in November.
Why this works
Satisfaction is hard to measure; attendance is a concrete indication. The question shows what the figure does and does not say.
Make it slightly more difficult
Come up with a second definition of new members. How could the outcome change?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
Satisfaction is hard to measure; attendance is a concrete indication. The question shows what the figure does and does not say.
A step further
Come up with a second definition of new members. How could the outcome change?
Try & discover · no obligation
Put the steps in a workable order
Move the steps with the arrows. Check whether each step has its necessary input and control.
What does a step depend on?
- Formulate a sharp question must precede Set down the definitions
- Set down the definitions must precede Choose source and period
- Choose source and period must precede Check the data for empty and duplicate rows
- Check the data for empty and duplicate rows must precede Carry out the planned calculations
- Carry out the planned calculations must precede Report with source and limitations
Move a step and then check the sequence. The arrows also work with your keyboard.
This model checks dependencies, not duration, capacity, or actual execution. With multiple agents, the human decision remains a separate step.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Help me make an analysis plan. Fictional question: “Are more young people coming to our library than last year?” First ask me questions about the definition of young people, the source, the period and the decision that depends on the answer. Then make a plan with the calculations, one suitable chart and three pitfalls. Do not give an outcome yet.
Apply it
A team leader asks AI to find “something interesting” in an export of all customer data for the annual meeting. The AI finds a link between the day of the order and the amount.
Which approach best fits this situation?
- Present the link straight away as the main conclusion.
- First formulate a concrete question with definitions and a source, and treat the link you found as a hunch that you still need to test.
- Keep searching until a more striking link turns up.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
Which analysis question is best suited to being worked out?
- How are our customers doing?
- How many new customers ordered for the first time in the first quarter of 2026, compared with the same quarter in 2025?
- Let AI look for interesting patterns in all our data.
Extra practice
Choose a question from your work that you want to answer with figures. Write down the question, the definition, the source, the period and the decision that depends on it, before you open the data.
Question: how many borrowers aged between 12 and 18 borrowed at least once from September to December 2025, compared with the same months in 2024? Definition: a borrower with at least one loan; age on 1 September. Source: export from the lending system of 3 January 2026. Decision: do we keep organising the youth programme on Wednesdays? Pitfall: in October 2024, the library was closed for two weeks.
