The data world is growing and the data analytics spectrum is evolving with it. Traditional BI methods are no longer sufficient to get the most value out of your data. More and more companies are embracing Advanced Analytics, Data Science and AI, but most are still trying to get grip on how to add value with it. What are the differences between these data analytic methods and which one should you focus on?
BI is the traditional method of how a business uses its data: the business raises questions, the BI teams searches through the data and reports back. This method only looks at historical data, looking backwards.
For AA the team consists of analyst and/or data scientists. They not only create dashboards and reports based on the business questions, but also analyse why (diagnostic analysis). With these analytical methods they can create predictive models, being able to predict what is going to happen based on the data; looking forward.
With using AI, the business is able to answer simple questions with self-service tools and interactive dashboards(which is also somewhat possible with AA). Complex questions in the form of use cases get asked to the data scientists, who then use their skills and tools to solve the use cases. This can result in prescriptive models, which can specifythe next action to take to reach a required outcome, or can directly communicate with and impact the business systems themselves.
AI and Data Science are often interchanged. It should be noted that Data Science is a part of AI, but AI includes many more subfields (machine learning, computer vision, etc.). For the purpose of this blog post, I will keep referring to AI as the prescriptive data analytics form.
So, in summary:
1. What happened? (Descriptive Analytics) - Mainly BI
2. Why did it happen? (Diagnostic Analytics) - AA
3. What will happen? (Predictive Analytics) - AA and AI
4. How can we make it happen? (Prescriptive Analytics) - AI
All companies need analytics in some form and maturity. If you have no insight in your data, how do you know if you are doing well? Research shows that companies that inject big data and analytics into their operations have productivity rates and profitability that are 5% to 6% higher than their peers. But doing Data Science and AI does not mean you no longer need BI and descriptive analytics. And while I see many companies taking the next step in their data maturity, they often don't know how to convert this into value. Jumping into AI just because your peers are doing it might not be effective if your company isn't ready yet. To quote some statistics:
Luckily, these different methods of using your data can co-exist and work in parallel. You can invest some in data science and AI projects, and invest some more in your BI landscape.
The BI world itself is also rapidly changing and has shifted its focus from a market based on reporting to one that is business-centric and user friendly. Gartner defines modern BI & analytics platforms as: “easy-to-use tools that support a full range of analytic workflow capabilities and do not require significant involvement from IT to predefine data models upfront as a prerequisite to analysis.”
Self-service analytic tools are allowing users from all over the organization (so not just the BI team!) to get insights in their data. The tools are so simple to use that anyone can answer their own data questions, relieving the BI team from the ad-hoc requests and shortening waiting times for all data requests. When this is implemented within an organization, we are talking of Self Service Analytics, or Self Service BI.
I can do everything myself? This sounds too good to be true!
No more waiting on the BI team, and I can do everything myself? This sounds too good to be true! And as it turns out: it is. What is often overlooked when talking about self-service BI are the following three points:
For correct data analytics:
If incomplete data is used or incorrect analytics is applied, wrong conclusions can be drawn from the analytical results and this will lead to wrong business decisions. Also, these tools sell themselves as easy to use, business user friendly and intuitive, but in reality you always need training to get started.
Another risk with implementing SSA is data governance and creating a dashboard jungle. When everyone is able to make their own analysis, data sources and reports, not only will this lead to data redundancy, it can also cause multiple truths. It is not uncommen to have multiple dashboards showing different numbers for the same period, simply because different departments report them differently (e.g. we detract the returns, we only look at actual shipments, we include the cancellations etc.).
So should you forget about self-service BI altogether?
A common phenomenon with companies that don’t provide any self-service BI, is that people get impatient when they have to wait for their data questions to be answered. They will download and use these tools anyways, creating shadow IT, or they download the report in excel and continue from there. Having all this data stored locally on your colleagues laptops is a security breach waiting to happen...
So in short, it is wise to facilitate in some form of self service BI, be it in the form of an interactive flexible insights dashboard, or a fully functional Modern BI Self-Service tool. But remember to train your staff on how to use these tools, get them familiar with the data (provide a data dictionary), and ensure data governance processes are in place and adhered to.
And, after you've invested in all of the above, don't forget to keep your business engaged, inspired and show them how fun data analytics can be!
Thank you for reading! In my next blog post I will write about my experiences with implementing self service analytics and setting up ánd maintaining the self service analytics community.
