Business Management · Using data, company functions, people skills, influence and clear communication

Analysing and Presenting Data for Decision Making

Updated 11 October 2026 · Fact-checked

Analysing data for decisions means turning raw figures into a clear, honest message. Check data quality, pick summaries that fit the question, choose a chart that shows the comparison, test for bias, state the uncertainty, and end with a recommendation and next step the audience can act on.

Understand Analysing and Presenting Data for Decision Making

Data on its own does not decide anything. A decision-maker needs an answer to a question: should we raise the premium, enter a new market, or change a claims process? Your job is to move from the question to the data, and from the data back to a clear recommendation.

Start with the question and the audience. A board member wants the conclusion, the size of the effect and the risk of being wrong. A technical peer also wants the method and assumptions. The same analysis is presented differently to each.

Next, summarise. Use the mean for symmetric data, the median when there are outliers or skew (claim sizes and incomes are usually skewed), and a measure of spread such as the standard deviation or range. A single average hides variation, so give a spread alongside it. Use percentages and rates for comparing groups of different sizes, and say what the base is.

Then choose a visual that matches the message. Use a line chart for trends over time, a bar chart for comparing categories, a histogram for the shape of a distribution, a scatter plot for the relationship between two variables, and a table when exact values matter. Avoid pie charts with many slices, truncated axes that exaggerate changes, and decoration that adds no information.

Finally, be honest about reliability. Data can be biased through how it was collected (selection bias, survivorship bias), through how we read it (confirmation bias, anchoring), or through mistaking correlation for causation. Any estimate from a sample carries uncertainty, so give a range or a sensitivity test and state what could change the conclusion.

Key rules to remember

Mean: x̄ = Σx ÷ n; Sum of the values divided by the number of values. Sensitive to outliers.
Median: Middle value of the ordered data (average of the two middle values if n is even); Better than the mean for skewed data such as claim amounts.
Sample standard deviation: s = √[ Σ(x − x̄)² ÷ (n − 1) ]; Measures spread around the mean. Uses n − 1 for a sample.
Coefficient of variation: CV = s ÷ x̄; Compares spread between data sets with different means. Meaningful only when the mean is positive.
Percentage change: (New − Old) ÷ Old × 100%; Always state the base. A rise from 2% to 3% is a 1 percentage point rise but a 50% relative rise.
Approximate 95% confidence interval for a mean: x̄ ± 1.96 × s ÷ √n; A rough guide for a reasonably large sample. It reflects sampling error only, not bias.

How to solve Analysing and Presenting Data for Decision Making questions

Use this order for any question that asks you to analyse, present or interpret data for a decision.

1State the decision or question and who the audience is.
2Check the data: source, period, completeness, definitions, and any obvious errors or outliers.
3Choose summaries that fit the data shape: mean or median, plus a measure of spread, and rates or percentages for comparisons.
4Choose a chart or table that shows the key comparison, and label axes, units and the base.
5Look for bias and limits: sampling, missing data, survivorship, confirmation, and correlation versus causation.
6Quantify uncertainty with a range, scenario or sensitivity, and say what would change the conclusion.
7Give a clear recommendation in plain language, with the main caveat and a next step.
8Write for the audience: lead with the conclusion, keep jargon out, and put technical detail in an appendix.

Quickest way: Question, Data, Message, Doubt, Action

When to use it: Use when you have only a few minutes on a written or case-study question about data and decisions.

Question: write the decision in one line.
Data: note one quality check and one limitation.
Message: name the summary and chart that show it best, and why.
Doubt: name one bias and give the uncertainty.
Action: finish with a recommendation and a next step.

Common mistakes in Analysing and Presenting Data for Decision Making

Quoting only the mean for skewed data.
The mean is the familiar default and is easy to calculate.
Fix: Check the shape first. For skewed data give the median and a spread, and explain why.
Choosing a chart because it looks good, not because it fits the message.
Students focus on appearance rather than the comparison the audience must see.
Fix: Match the chart to the purpose: line for trend, bar for categories, histogram for distribution, scatter for relationships.
Treating correlation as proof of cause.
Two series moving together feels like an explanation.
Fix: Say the data shows an association, suggest other explanations such as a third factor, and propose a test or further data.
Giving a single number with no uncertainty.
Students think a range looks weak.
Fix: Give an interval or scenarios and explain what they mean for the decision. A stated range builds trust.
Ignoring how the data was collected.
Data in a spreadsheet looks objective.
Fix: Ask who is included and who is missing, for example only surviving policies or only customers who replied to a survey.
Writing for a technical reader when the audience is non-technical.
Students want to show method and use jargon.
Fix: Lead with the conclusion in plain words, explain terms briefly, and move the technical detail to an appendix.

Worked examples

Example 1

A health insurer reviews seven claim amounts (₹ thousands): 12, 14, 15, 16, 18, 20, 95. Management asks for one figure to describe a typical claim. Which should you give, and how would you present it?

Show the solution

Sum = 12 + 14 + 15 + 16 + 18 + 20 + 95 = 190.
Mean = 190 ÷ 7 = 27.14 (about ₹27,143).
Order the data. With n = 7 the median is the 4th value = 16 (₹16,000).
The value 95 is an outlier. It pulls the mean well above six of the seven claims.
Recommend the median as the typical claim, and report the mean separately because large claims matter for total cost.
Present with a simple bar or dot plot of the seven claims, with the outlier labelled, and note that seven claims is a small sample.

Answer: Use the median of ₹16,000 as the typical claim, and mention the mean of about ₹27,143 and the large claim of ₹95,000, with a caution about the small sample.

Example 2

A survey of 100 customers who renewed a policy shows 80% satisfaction. The sales head says this proves the service is excellent for all customers. Comment, and give the approximate 95% confidence interval for the proportion, using 1.96 × √[p(1 − p) ÷ n].

Show the solution

Standard error = √(0.8 × 0.2 ÷ 100) = √0.0016 = 0.04.
Margin = 1.96 × 0.04 = 0.0784, so the interval is 80% ± 7.84%, about 72.2% to 87.8%.
The interval covers sampling error only. It does not fix bias.
Only customers who renewed were surveyed. Unhappy customers who lapsed are missing. This is survivorship or selection bias.
So the 80% likely overstates satisfaction across all customers.
Recommend surveying lapsed customers too, then present the results by group in a bar chart with the interval shown.

Answer: The approximate 95% interval is 72.2% to 87.8%, but the sample excludes lapsed customers, so it cannot support a claim about all customers. Extend the survey before concluding.

Exam tips

In written and case-study questions, always link the analysis to the decision. Marks go to the recommendation, not just the calculation.
Name the bias specifically, such as survivorship or confirmation, and say how it would change the result.
When asked for a chart, give the type, what goes on each axis, and why it suits the message and audience.
State assumptions and limits in one or two lines. Examiners reward honest uncertainty.
In MCQs on this topic, watch for options that claim causation from correlation or treat a sample result as certain.

Practice questions from Using data, company functions, people skills, influence and clear communication

Analysing and Presenting Data for Decision Making: frequently asked questions

How do I present data to non-technical stakeholders?

Lead with the conclusion and what it means for their decision. Use one clear chart per message, plain words, and a short note on the main risk. Put method and detail in an appendix.

Which chart should I use for business data?

Use a line chart for change over time, a bar chart for comparing categories, a histogram for the spread of values, and a scatter plot for the link between two variables. Use a table when exact figures matter.

What are common biases in data analysis?

Selection bias and survivorship bias come from who or what is in the data. Confirmation bias and anchoring come from how you read it. Also watch for treating correlation as causation.

How do I draw conclusions when the data is uncertain?

Give a range or scenarios instead of one number, and say what the range means for the decision. State key assumptions and what new data would change the conclusion. Then still make a clear recommendation.