People often ask AI to “analyze this spreadsheet” and receive a polished chart or paragraph. The easy-to-miss question is whether the rows, units, and arithmetic are right. A model can be excellent at suggesting what to compare and still make a numerical error. This lesson gives you a small, human-checkable practice set.
You do not need spreadsheet expertise. You will ask the assistant to organize a few numbers, show the calculation, and name what the numbers cannot tell you. Then you will check the result with a calculator or spreadsheet. The point is not to catch the AI out; it is to build a reliable habit before using larger data.
Fictional support example
Week one
100 × 12 minutes
20 hours
Week two
120 × 10 minutes
20 hours
Begin with units and the question
Imagine a fictional support team handled 100 attendee emails in week one and 120 in week two. Average handling time was 12 minutes in week one and 10 minutes in week two. Ask a narrow question: “How many staff hours were spent on those emails in each week?” The answer is 100 × 12 ÷ 60 = 20 hours, then 120 × 10 ÷ 60 = 20 hours. The team worked faster per email but did not release total handling hours because volume rose.
This is the sort of distinction a smooth summary can miss. “Time per email improved” is true. “The team saved four hours” is not. Ask for the formula, units, and whether the comparison mixes changes in volume with changes in speed.
Try this
Copy a tiny-data prompt
The figures are fictional. Check the answer yourself with a calculator.
Analyze only these fictional figures: Week 1 had 100 attendee emails at 12 minutes each. Week 2 had 120 attendee emails at 10 minutes each. Calculate total handling hours for each week and the change in minutes per email. Show each formula and its units. Then write three sentences for a manager: what improved, what did not improve, and what extra information we would need before claiming cash savings. Do not invent costs or staffing changes.
Why it works: This asks for transparent arithmetic and guards against turning faster work into an unsupported cash claim.
Ask the assistant to show its work, but verify it elsewhere
A visible formula is easier to audit than a bare result. It is not automatically correct because the assistant wrote steps. Recalculate the two totals in a spreadsheet or calculator. Check that minutes became hours by dividing by 60 and that the same unit is used on both sides of the comparison.
With real data, inspect missing rows, duplicates, outliers, and definitions. Does “email” mean one customer issue or every reply in a thread? Does the average include escalations? A calculation over the wrong group can be precise and useless. If the data set is large, ask the assistant to propose checks before asking for a conclusion.
Translate a number into a decision carefully
Suppose the assistant suggests hiring fewer people because the average fell by two minutes. The supplied data does not support that. It says nothing about peak hours, quality, backlog, or whether the team was already short-staffed. Ask what would have to be true for a staffing claim to follow, then look for those records.
If you later use the site’s ROI calculator, remember the same rule: released hours are capacity until a budget action turns them into cash savings. A useful analysis ends with “what to measure next,” not just a tidy percentage.
Practice assignment
The one-table analysis
Use the two-week fictional support figures in the prompt.
- 1Ask AI for a table with email count, minutes per email, and total hours for each week.
- 2Check each total independently.
- 3Write one correct conclusion and one tempting but unsupported conclusion.
- 4Add the two measures you would need before changing staffing or budget.
What to save
A checked two-row table and a four-sentence management note.
You are ready when
Both weekly totals equal 20 hours, and the note does not claim cash savings from the lower per-email time alone.
Further reading
Sources and guidance behind this lesson, not required reading to finish it.
- NIST: Generative AI Profile
Explains risks including confident false statements and fabricated citations.
- AI Adoption Trends: AI ROI decision lab
Interactive practice for keeping released time separate from cash savings.