In the short and varied time that I’ve been analyzing web and ecommerce data, I’ve found that the same 4 questions can still drive the vast majority of your analysis.
These questions work, because they force you to focus your diagnoses, interpretation and communication around action and executive summary.
Thus, I present the only 4 web analysis questions you need.
#1 – What Happened?
Simply describe the trends in terms of your KPIs–revenue, conversion rate, traffic, etc. The goal here is to pull out the most important and meaningful trends.
These are usually either notable bits of data that stand out or they relate to some activity you’ve recently taken (new campaign, site redesign, etc.)
#2 – Why Did It Happen?
“What” is fairly shallow. “Why” is much deeper. This is the analysis.
It’s a matter of looking at past your KPIs to your diagnostic measures.
It’s moving away from averages and toward segmentation.
There’s no perfect guidebook for this. You just have to keep going until you can clearly explain the root cause.
#3 – What Does That Mean?
Answering “What happened?” and “Why?” still doesn’t give you the whole picture. For example:
- What?: Revenue increased 10% month over month
- Why?: Email drove the boost. While visits stayed steady, conversions increased with the new free shipping offer.
- What does that mean?: Existing customers may be more responsive to shipping offers vs. price reductions.
Sometimes the data tells you the whole story, but many times you need hypothesize about the meaning.
#4 – What Do We Need to Do About It?
Analysis and action are meant to be together. This last question is all about drawing that connection between data, insight and action.
- Data - A metrics based view of the situation
- Insight - Human interpretation of the data
- Action - Next steps to improve
In answering this question, you’re connecting the dots between what your business hopes to achieve and what the data tells you to do. It should be obvious why you recommend what you do and how it will benefit your organization.
If you could only ask 4 web analysis questions, what would they be?