Make Powerful Data Visualization Work For You
Published: October 29, 2014
Author: Leslie To
Now that we’ve covered what data you should avoid and how to curate your data, we’ll get into data visualization, which is an equally crucial piece of the presentation pie.
Data visualization is by no means rocket science, but it can easily be done incorrectly. Anyone can build graphs, but it takes a little more effort to build graphs that tell a story and lead both you and the client toward better insights.
I know everyone has heard the mantra “pictures are worth a thousand words.” Graphs (and data visualization in general) are probably not worth a thousand words, but I think they’re worth at least a couple hundred.
When we find meaningful data, especially information that drives or changes strategy, it’s important that it’s represented in a way that is not only clean and polished, but also communicates our points clearly and concisely. Analysis helps explain the information to clients, but graphs should strive to tell a story. When graphs are allowed to speak for themselves, they’ll make for a much better presentation.
For example, in the pictured graph, you can tell that users visiting the Bedroom section of the site are strong in both volume and conversions. Visitors to the Process page, while not particularly a strong traffic segment, convert so well that the page represents a key part of the site. Note that:
– When not pictured together, it’s easy to forget how visits and conversions are related, especially at the subcategory level.
– Traffic/click volume is equally as important as traffic/conversion quality.
I personally think that polished reporting is so important for both your own professional development and also for your agency’s. You’ll never be quite certain who the final recipient of your report or document is going to be. Your report (and anything you turn over to the client) can be easily passed onto someone else higher in the chain. Putting an agency’s name (and/or your name) on it should bear some importance. Polished reporting and deliverables ensures that we and you always look good to a third party, whoever that may be.
I can’t tell you how you should build graphs, because like curating data, it really is client-specific. Instead, we’ll cover common mistakes to avoid and cut out of the process, so that the storytelling aspect of reporting is a lot easier.
Dimensions vs. Metrics
First, understanding the difference between dimensions and metrics will help you build better custom reports in Google Analytics.
Dimension – attribute of the visitor to your website (source, landing page, medium, campaign, ad group, etc.)
Metric – a number, whether ratio or count (sessions, pageviews, conversion rate, conversions, etc.)
Building your own custom report allows you to dictate what you want to review, rather than letting the platform tell you what you should report on. This really goes for everyone regardless if you are paid search or SEO. Google Analytics custom reporting is really robust and should be used when applicable. For paid search folks, this requires Google Analytics integration.
Don’t Compare Apples to Watermelons
Always calibrate your segments correctly, so that you’re always comparing apples to apples. (Mobile is a device, not a source!)
In graphs, each individual metric should be independent of its neighbors. If there is cross-pollution, then the graph is useless and will lead you towards conclusions that aren’t valid.
Time Frame is Key
Comparing against historical performance is always important and can usually lead to insights. But in order to establish growth or changes, we need to make sure we pick a fair comparison window. Picking the wrong time frame to compare your data can lead to bad analyses.
Contextualize Your Graphs
When you graph metrics, make sure the numbers tell the full story.
If there are inherent relationships between two metrics, we need to display both metrics. For example: you can’t just say conversions grew 10% if spend was also increased 100%.
Let Your Graphs Do The Talking
While I’m not a huge fan of pie charts in general, I think this actually demonstrates my point quite clearly: mobile conversions are not growing at the rate that mobile traffic is. We need to investigate why mobile is a poor converter or optimize mobile towards traffic, rather than conversions.