Data analysis: Do you need a dashboard?
This article written by one of our analytics experts will teach you how to differentiate reporting from analytics, and define the main issues related to the creation and maintenance of dashboards.
This article written by one of our analytics experts will teach you how to differentiate reporting from analytics, and define the main issues related to the creation and maintenance of dashboards.
The web is overflowing with data. A lot of data. So much so that measuring an event for a single user or group of users requires data from a multitude of different sources in order to be analyzed correctly. For example, an analysis of a marketing campaign for an online retailer should factor in:
To make the data talk, you first need to pull numbers related to performance indicators like return on investment, cost per acquisition and margins generated. But for the data to tell the whole story, it needs to be connected; that’s how you’ll know that the results of campaign X on Facebook correspond to X sessions in Google Analytics and generated X sales dollars in the internal information system. To get to that point, you need to unify the data, in other words, you need to know that the identifier John in database 1 corresponds to driver’s license number LFURKF67520.
Since the 2010s, the emergence of data visualization and unification platforms (thanks to connectors that create links between platforms) like Datorama, Tableau and Power BI are gradually democratizing the practice of creating dashboards, previously reserved for those initiated in the language of code. Of course, they are now used by a much wider pool of users than before, but it would be wrong to pretend that just anyone can use these tools properly, without any concept of data modelling. Creating effective dashboards takes forethought and planning. Here’s a summary of the major challenges in the practice of creating dashboards.
Mostly that they can save time! Campaign reporting can be tedious, because it involves a lot of manual tasks. In addition, you have to start over with each report because the data is static. Creating a single report in a visualization platform allows you to automate this task, and therefore frees up more time to actually analyze the results.
But be careful, these advantages come with a catch. The main risk is that you can end up wasting a lot of time fiddling around with the technical side instead of analyzing results. Like a car, the costs related to a dashboard aren’t restricted to its creation. You need to plan for maintenance costs because your raw material, the data, is in constant evolution, which means that periodic adjustments are needed. Adjusting campaign names, revisiting the formulas that manage data groups (channel, tactic, segment…) are all tasks that can infringe on time you’d rather spend on analysis.
How the data is structured and what processes are in place are critically important to the success of a dashboard. On one hand, they create a level of operational efficiency that make the dashboard credible: it displays exact data. Without maintenance, a dashboard simply becomes obsolete. On the other hand, the costs related to maintenance can be very high if no governance framework is defined.
The two following points are important for preserving the effectiveness of your reporting.
A major misunderstanding about data analysis and representation resides in the fact that many people confuse two tasks, reporting and analytics, that don’t have anywhere near the same goal.
The majority of organizations want dashboards so that they can extract insights. It’s therefore analytics they’re after, not reporting. Your KPIs should measure the health of your business objectives, not display your data. A dashboard that’s too detailed and missing KPIs, that contains only metrics, would lend itself more to reporting than analytics, and would consequently be useless for extracting insights because they’d be hiding behind the flood of numbers in the fore.
Did you choose your KPIs based on your business goals? Now is the time to segment them. But watch out! What are you trying to accomplish? Bear in mind that making a dashboard has two objectives:
It’s important to base your reflection on performance measurement; what do you actually need to evaluate your performance? We’re talking here about KPIs and not metrics, and you should be choosing just two or three of them. Then segment these metrics into a few main segments, and the architecture of your dashboard is done. This exercise is generally done as part of your KPI framework.
Once you’ve chosen your KPIs, it’s time to choose the metrics that will break them down and explain them. Here are a few ways to make sure you’re choosing well:
Finally, you have to choose the right type of dashboard based on who will be using it. There’s no such thing as a dashboard that answers everyone’s questions. Your dashboard will be very different depending on whether it’s used by your executives (strategic dashboard), by your managers to evaluate your campaigns (tactical dashboard) or to track your operations (operational dashboard).
The democratization of the tools that allow us to build dashboards have made it accessible to everyone. By following the guidelines above, any organization, from a small business to a large group, can efficiently evaluate their performance based on fixed goals, and take action to improve.
Happy analysis!