10 Rules for Measurement From 7 Years of Analytics Experience
Whether your job involves analytics or not, everyone gets asked to measure performance.
How well did it work?
This is the question that every leader wants to know — and the one that causes us the most anxiety. Fortunately, there are some simple rules that can help you out:
- Measurement isn’t analysis. Your goal is to answer one question: Did it work? Analysis happens later.
- Measuring performance requires a standard to measure against. Without a goal or a target, you’re not measuring, you’re just telling what something did.
- Too many KPIs mean you don’t know how to define success, which means you can’t comply with rule #2.
- Measurement isn’t about highlighting good performance, it’s about accountability.
- Reporting bad performance makes good performance meaningful. Without bad performance, there’s no insight.
- Measure what matters, not what’s easily available. Sometimes these are the same, often they aren’t.
- The ability to do #6 is 90% integrating siloed datasets. Leverage tools to make this easier (Supermetrics, Open Bridge, Xplenty, Stitch). If tools are too expensive use R Studio or the time-tested VLOOKUP.
- Analytics tools like Power BI, Tableau, and Looker save time but require adoption. Excel is time-consuming but universal.
- Visualizations are cool, (bar graphs, pie charts) but often don’t give enough information. Numbers and tables are usually better.
- If you’re not sure what to measure, try turning off stuff — what’s most important will become quickly apparent.