Learn about business analytics

Explore how businesses use data to make decisions, from KPIs to dashboards.

Time:: 60 min
Cost:: free
Difficulty:: moderate
Where:: Indoors
Best:: Solo

Most business analytics effort produces dashboards nobody acts on. The failure happens before any tooling is involved: the question was never phrased so that an answer would change a decision. An hour spent on question framing and the difference between a useful metric and a flattering one is worth more than an hour learning any analytics platform.

Why learn about business analytics is worth an hour

A metric earns its place only if some value of it would make you do something different. Applying that test eliminates most of what organizations routinely track. Understanding the distinction between leading and lagging indicators does the rest, because lagging numbers tell you what already happened while leading ones let you intervene.

How to learn about business analytics step by step

1
Write a question that changes a decision
"Which channel should we stop funding?" beats "how is marketing performing?" The first has consequences attached.
2
Apply the so-what test
For each metric, name what you would do if it doubled or halved. If nothing, it does not belong on the dashboard.
3
Separate leading from lagging
Revenue is lagging. Trial signups are leading. Only leading indicators give you time to change the outcome.
4
Check the denominator
Absolute counts flatter and mislead. Rates and per-unit figures are usually the honest version of the same number.
5
Look for the confounder
Before believing a change caused a result, ask what else changed at the same time. Usually something did.

Tips to get more out of it

Be suspicious of any metric that only ever goes up. Cumulative totals are vanity metrics by construction.
Segment before concluding. An average hides two opposite trends more often than it reveals one real one.
Write down what you expect before looking. Prediction first is how you avoid explaining any result after the fact.

Ways to mix it up

If you already frame questions well, spend the hour on experiment design instead, since correlation is where most business analysis quietly goes wrong. If you are building a dashboard, halve the number of metrics on it; almost every dashboard improves by deletion.