Analysis can be overwhelming. Sometimes even the word “analysis” will cause panic in folks. Where do I start? How the heck do I pull insights out of all these numbers? Below you’ll find a few tips to help guide you in your analysis adventures.
Sort out the details before pulling data
Gathering data can be time-consuming, and you don’t want to have to do it twice. Before pulling data from your various sources, consider the following:
- Scope – data sources, time frame, etc.
- Metrics – what numbers do I need to see to complete this analysis?
- Dimensions – what kind of segmentation do we need?
- Structure – do we need to reshape the data for any reason? Is it structured in a way that plays nice with other sources that we need to merge?
In addition, you should also have a general plan of how you’re attacking the problem. What tools are you using to manipulate the data? Excel, sheets, R? Have a rough idea of the steps you’re going to take. It’s worth it to spend some time in the front end preparing!
Have an objective from the beginning
This one is closely related to the previous point of sorting out the details before rolling up your sleeves and getting into the nitty-gritty. With most analysis projects, you’ll have more direction (and an ultimately easier time) if you have a specific question you’re setting out to answer.
- How has Google’s close variant matching affected our account’s performance?
- Where are we wasting money? Which platforms, devices, ad types, etc.
- How has the account restructure affected CPL and conversion volume?
Use graphs and visuals
It’s easy to get bogged down in the weeds with data. When looking for insights, graphs can be incredibly useful in both finding and displaying learnings. Humans are visual creatures, we are designed to look for patterns. As you might guess, patterns can be hard to find when you’re looking at an Excel sheet or data table.
I don’t know about you, but I nerd out over a great graph or other data visualization. Here are some examples of some spruced up visuals.
Nothing is inherently wrong with tables… – Read more