on 15 October 20
A common challenge that many companies face in starting up their big data and business analytics programmes lies in understanding where to start and how to make it all come together. Business analytics requires the application of various skills that are normally found distributed across multiple functions in contemporary organizations. It doesn't help that the availability of new technologies and techniques is invariably accompanied by new jargon to describe them.
So if you're one of those business leaders who is wondering how to begin, here's a step by step guide for lay users to understanding analytics and how to get started.
1. Decide what the top couple of business questions that need answers are.
These are not just questions about what has already happened in the past (which your existing BI may have already answered), but questions about what can happen in the future given what you already know of the past. An example in the field of marketing would be a question about which geography or demographic group is more likely to buy what products in the future.
2. Consider what data might help answer those questions.
Is it data about what customers already bought? Could there be any clues in emails sent in by customers? Could it be in discussions that are going on in social media? Could it be in data available with other functions within your company? Is it largely in unstructured data that is not neatly stored in company databases? Is some of it with a department that's not part of IT? The answers may provide guidance for the next step.
3. Identify an owner/leader for the analytics programme.
Producing useful results from analytics requires overseeing a chain of coordinated activities. In addition to this, the data that is required may come from a number of different sources across the organisation. Getting it all together requires someone who can work with multiple functions, understands the business as well, and can be freed up enough to manage the activity as a project.
4. Get the right team together.
The ideal big data analytics team involves a number of skills that can largely be viewed in three broad perspectives - technology, data analytics and business. On the technology side, the right skills are required to work with big data capture (or extraction), storage, cleansing, analysis, and visualization. This also involves selecting, implementing, integrating and managing the right platforms, tools and technologies. This is likely to be done by the IT staff, but the drivers for technology decisions must originate from the business.
Data analytics skills are applied to statistically analyse the available data and identify trends, and create mathematical models that can be used to extrapolate the trends into the future to come up predictions. This can go further, with sensitivity analyses and probability theory being applied to answer "what if" questions. This is done using statistical, optimization and simulation tools. The results are reported using visualization tools.
Experts in the business are required to provide the leading questions, and provide functional direction and validation every step of the way.
The ideal data scientist would be one who is thorough with not only analytics but also a variety of tools that are used to implement it. In addition, he/she would have a good understanding of big data technologies as well, apart from a strong understanding of the business domain and processes. In practice, the candidates who would match this ideal may be few and far between, and for this reason the right team will probably consist of a small core team that involves and works with additional specialists in each functional area as required.
5. Execute and iterate
The above may seem simple enough, but in reality it takes time to produce results. The collection, scrubbing (cleansing) and integration of data from multiple sources is the first step and it must be done with patience and meticulousness. After all, it's the prime input to the whole exercise, and the quality of results will be dependent on the quality of input data.
Performing analytics on the data made available is also an exercise in patience, repeated reviews and testing. Predictive and prescriptive analytics is all about forecasts, and the best forecast is one that provides the most comprehensive treatment to the most appropriate data that is available. Models need to be run repeatedly, and since many external variables may change in the process over time, these changes need to continuously be tracked and factored into the models.
6. Report the results in an understandable manner
A picture speaks a thousand words. Visualization is the best way to present results, and for this, the right tools and facilities need to be selected so that the results are clearly and completely understood the first time around by people who are likely to be business leaders and business experts, but not data scientists. The lower the reliance on jargon the better.
There's a plethora of services, tools and technologies that are continuously being made available in the market even as the existing ones mature. It's easy to get lost in the maze, and so it helps if the leadership focus is not confused by what tools to use, but is maintained on how to move forward towards success.
Share your perspective
Share your achievement or new finding or bring a new tech idea to life. Your IT community is waiting!