With the availability of consumer data, now more than ever, data drives marketing decisions. But how do you make sense of all the Excel spreadsheets, the CSV files, or even Facebook Insights? If reporting is a novel, data collection is only the beginning. The real story develops in the later chapters, bringing life to the plot and all of its complications.
Data Collection: The Exposition
Data can only be as meaningful as the way it is collected, and data collection is a topic big enough for its own book. However, there are some takeaways that we can mention in this article.
It’s always better to over-collect than under-collect.
Try to anticipate your data needs before they happen. Be as granular as possible because you can always roll-up data, but the same cannot be said for data separation. For example, if you are trying to understand how your website is performing in different metros across the United States, include a date dimension so that you can analyze trends and exclude or highlight any irregular patters without removing an entire city or making assumptions off of simple aggregations.
Know what perspective you want to view your data from, but be open to change.
It’s always good to have an idea of what you want the end result to look like before you collect your data. This allows you to make a shortlist of data needs, but can also allow you to think through your reporting in a way that you might not have otherwise. For instance, if you knew that you wanted to provide a post-level analysis of your Facebook page, you might export that data from Facebook Insights. While doing so, you notice that one of the dimensions Facebook provides is “post type”. This would be useful to incorporate when analyzing post messaging to see if a particular theme is being better consumed as a photo, link, or status on your Facebook page.
Trend Analysis: The Rising Action
Trend analysis is one of the most common ways to look at data because it tells you directionally how a particular project is doing. However, adding in different dimensions or layers to trend analysis can reveal valuable insights to your data you might not have seen otherwise.
Take for instance a Facebook collect application that is supposed to drive signups for email communication. Your company has approved paid advertising to promote the application and has chosen to reach out to magazine publications and bloggers to help spread the word. Standard trend analysis would be able to take traffic volume and signups over time and say overall whether or not the promotion was a success.
Let’s say in this instance it was successful, but now your company wants to do another round of promotions to encourage more signups. Adding a traffic-source dimension to daily analysis would allow you to see specifically how traffic from each magazine publication and blogger performed, and whether or not users from that source were likely to convert. After all, at the end of the day, your conversion metric is a signup, not just traffic to your application.
Conversion Funnels: The Climax
The first step in understanding your data is to also understand the ways it connects. For example, if you are looking at your website’s activity and need to include how many transactions your e-commerce section generated, remember to include all the interactions that lead to a transaction. Understanding how users are interacting with your website can not only help drive more transactions, but also increase site engagement. Take the following process for a website trying to drive users to share social content:
In this particular instance, there is a large barrier to website conversions. The goal here is to get users to create original content on the website and then share it through various social channels. As depicted in the graphic above, once the user engages in the process, conversion rates remain very high. Getting the user to complete the first step, however, is a huge challenge. If we only included the final share counts, we would be unable to see the opportunity to improve the website’s design to drive more conversions.
Data Visualization: The Falling Action
Data visualization is where you start to make sense of the trends in the data. This is where context and data finally come together to become something meaningful. In each of the scenarios outlined above, there was context provided to help explain why these numerical anomalies were happening. That same context needs to be present in the story articulated in a report.
Context can be lost however, if the visuals are not complimentary to the point that needs to be made. The general rule of thumb here is the simpler, the better. A common misconception in building strong visuals to support data is that they need to be elaborate or visually stunning; the true mark of a good data visual is if your reader can walk away from the report remembering what they just saw. The purpose of a data-driven report is to take disparate points of interaction and turn them into something meaningful and actionable. If your visual does not lend itself to either of these, then perhaps you should reconsider how to summarize the insight you are trying to communicate.
Take the visual below. In a compact space we are given a written explanation of context (text box on the left), year over year monthly trends (histogram at the bottom), user role breakdown (pie chart on the right), and the YTD total with percent growth (text box within the pie chart).
Data-Driven Reporting: The Resolution
This is where analysts really earn their paycheck. By taking our entire process, beginning with data collection, analyzing trends, understanding the conversion funnel, and building data visuals, we finally arrive at reporting. This is where we can take our learnings and outline the next steps for future projects. Though there are several ways to structure reports, perhaps one of the simplest is to mimic the user experience.
For example, if you are reporting on microsite performance, take the readers through the process and highlight performance in each step. Following the social sharing website outlined in the “Conversion Funnels: The Climax” project, guide the reader in the way that the user would interact with the site, from content generation to publication via social networks. This sets the stage for recommendations.