Data Analysis for Business Analysts: How to Read It, Interpret It Honestly, and Make Better Decisions

DAY 5 | The Data Analysis Toolkit: Your Practical Reference Guide

This is a reference guide, not an argument. It pulls together everything from this week into a single place you can return to whenever you are working with data. If you followed the series from Monday, this is the distilled version. If you are arriving here first, each section stands on its own well enough to be useful immediately.

The Mindset

Before any technique, the foundation of good data work for a Business Analyst is a single distinction: the difference between reading data and understanding it. Reading data describes what is in front of you. Understanding data tells you whether to trust it, what it means, and what it leaves out. The organisation does not need more summaries. It needs judgement. Everything below serves that goal.

Reading Data: The Foundations

Know your data type

Qualitative data describes categories. Quantitative data measures quantity. Within quantitative, discrete data takes specific values and continuous data takes any value in a range. The type determines which analysis is valid. You cannot meaningfully average a set of category codes, however willingly the software does it.

Choose the right measure of central tendency

The mean is the everyday average and works for evenly distributed data, but it is distorted by extreme values. The median is the middle value and tells a truer story when the data has outliers or a long tail. The mode is the most common value and is the right measure for categorical data. When given an average, always ask whether it is the mean or the median and whether extreme values are present.

Look at the spread, not just the middle

Central tendency tells you where the middle is. It says nothing about how spread out the data is. Two processes with the same average completion time can have completely different reliability. The spread, measured by range or standard deviation, often carries more useful information than the average.

Understand the distribution

The shape of the data determines which measure to trust. Evenly distributed data suits the mean. Skewed data with a long tail suits the median. You do not need statistical tests to benefit from this. Simply looking at the shape tells you which summary is honest.

Questions to Ask of Any Dataset

• Where did this data come from and how was it collected?

• What time period does it cover, and is that period representative?

• How complete is it, and is the missing data random or systematic?

• What is the sample size, and is it large enough to trust the pattern?

• What does this data not capture?

The Traps That Mislead

• The truncated axis: a vertical axis that does not start at zero stretches a small change into a dramatic one. Check the axis before believing the trend.

• The cherry-picked time period: any trend can be reversed by choosing the right start and end points. Ask why a particular period was chosen and what a longer view shows.

• The misleading average: an average can hide a distribution that tells a different story. Ask about the spread and the outliers.

• Correlation as causation: two things moving together does not mean one causes the other. Ask what else could explain the relationship.

• The missing denominator: a raw number means little without the total it came from. Always ask, out of how many?

• Survivorship bias: data that only includes the cases that made it through a filter hides the ones that did not. Ask who or what is not in the dataset.

The Habits of Honest Interpretation

• Separate what the data shows from what you want it to show. Name your expected conclusion so you can notice when you are bending toward it.

• Actively look for the alternative explanation. If you cannot rule it out, your conclusion is weaker than it looks.

• State the limitations alongside the findings. A recommendation that acknowledges its limits is more trustworthy, not less.

• Be willing to deliver the inconvenient interpretation. When the data does not support what everyone wants, saying so clearly and with care is the most valuable thing a BA can do with data.

From Data to Decision

Data informs decisions. Judgement makes them. A good decision weighs the data alongside the factors the data does not capture. To build a recommendation the data genuinely supports, work through four parts in order:

• The question: what decision is this recommendation supporting?

• What the data shows: the relevant findings, honestly interpreted, in plain language.

• What the data does not show: the limitations, the assumptions, the factors outside the dataset.

• The recommendation: what you recommend and why, following logically from the evidence and its limits.

When you present data, use the same tools the misleaders use, but use them honestly. Start axes at zero. Choose representative time periods. Give raw numbers their denominators. Show the data that complicates your recommendation as well as the data that supports it. Present data the way you would want it presented to you if the decision were yours and the outcome mattered.

When Data and Stakeholders Disagree

Deliver the honest interpretation with clarity and with care. Clarity means not hiding the inconvenient finding inside qualifications. Care means respecting the stakeholder and framing the finding as useful information rather than a verdict. Separate the finding from the decision. Your job is to make the interpretation honest. The decision belongs to the decision-maker, who may weigh other factors. You are not telling them they are wrong. You are giving them the clearest possible picture.

Your Free Data Analysis Toolkit

To accompany this series, I have put together a free Data Analysis Toolkit available to download from www.flotogbainsights.com.

The toolkit contains three resources.

1. A Data Interpretation Checklist that takes you through the questions to ask of any dataset before you trust it, in a format you can use on a live project.

2. A Common Data Traps Guide that summarises each of the misleading techniques from Wednesday with a plain explanation of how to spot it.

3. And a Data to Decision Template that structures the four parts of an evidence-based recommendation so you can build one that holds up.

All three are designed to be used immediately, on real projects, without adapting or configuring them.

Six Principles to Carry From This Week

• Reading data and understanding it are different skills. The organisation needs judgement, not summaries.

• The average is one measure among several, and often not the most honest one. Always look at the spread and the distribution.

• Most misleading data is not a lie. It is a partial truth presented by someone who believes it. Your job is to see the whole picture.

• Correlation is not causation, a raw number needs its denominator, and a trend needs a representative time period. These three checks catch most data errors.

• Data informs decisions. It does not make them. Present it as the strongest available evidence, with its limitations stated, as one input into a judgement that belongs to the decision-maker.

• The BA who interprets data honestly becomes the organisation's honest broker, a trusted source of interpretation precisely because it is not bent toward a predetermined answer. That reputation is one of the most powerful assets a Business Analyst can have.

Thank you for following this series. Data confidence is one of the skills that most directly determines whether a Business Analyst shapes decisions or merely documents them. I hope this week has given you the foundations to shape them.

Subscribe at www.flotogbainsights.com to receive next week's series directly.

Go out and be successful.

Oluwatosin Ogunkoya | Flotog BA Insights | www.flotogbainsights.com

Next week: a new series to build another core area of Business Analysis practice. AI for Business Analysts. What to hand to AI, what to never hand over, and why the role gets sharper rather than smaller. Day 1 launches Monday, 15th June 2026.