Big data remains a hot topic but some SMEs lack understanding of how it can be used to improve decision-making. I spoke to Michelle Moody, Engagement Director, Insights & Data UK at Capgemini, to find out how SMEs can benefit from big data and why it’s an opportunity they cannot afford to ignore.
Here's what Michelle told me:
What steps should a SME take to determine whether big data could be useful to their business?
“Data is pervasive in this digital world, which creates a fantastic opportunity for SMEs to use big data for better business decisions. The first step is to focus on a business challenge, from opening up new revenue streams to reducing costs. Then figuring out how big data can help address the challenge. What data sets do I need? What technologies can help me? What insight from analysis might add value? What data protection mandates must be observed?
“For example, a furniture manufacturing company wanted to enter Far East markets, and using a combination of social media sentiment analyses and translations capabilities, they found a much cheaper and quicker alternative to traditional market research. Targeting new markets is just one opportunity area for big data and analytics technologies, and SMEs can learn from each others’ big data success for inspiration.”
Are big data applications affordable for small businesses - and do they require technical knowledge on the part of the user?
“Clearly scoping requirements and choosing the right technology can make big data affordable for SMEs. First, you need to be clear about what business decisions you want better answers for. Then you figure out what data sets are available, what technologies can help, and what data protection mandates must be observed. Discussions with suppliers can help you understand this.
“Pay-per-use cloud-hosted services are a good first step, and you can offset costs against value derived from data insight. Many big data applications are open source and there are pre-packaged analytical environments on offer from companies like Pentaho or Protegrity. Both of these big data specialists are SMEs themselves who understand the needs of big and small clients.
“The technology is, of course, one component of a data insights solution. Analysts run experiments across data sets, to provide your business users with aggregated data. You can use in-house analysts or buy-in this service. The high-level user who looks at reports doesn’t require technical knowledge to look at reports and dashboards which work as they always did.”
Does big data technology enable companies to target data relating to a specific business requirement?
“Use cases include optimisation of complex supply chain schedules, predicting required maintenance work or new customer behaviors, reducing customer churn or simply reducing TCO of information management. The list goes on.
“The point is that big data technologies allow you to do more than you could previously, because you can afford to hold more data, and more complex data, and do more flexible analysis faster and more efficiently than ever before.”
Are there any types of business for which big data is especially useful/relevant?
“The more data points a company has, the more likely they are to benefit. Retail, consumer products, telecoms, baking and insurance companies are currently benefiting the most from the patterns in their big data but there is increased adoption from organisations in public sector and hospitality looking to mine their data assets.
“However, even companies that do not produce huge masses of data can benefit from big data as “ancillary” information that enrich their usual sources and add information about competitors, trends, etc. Companies must comply with data protection laws and ensure appropriate levels of data protection are applied.”
How important is the quality of the data used for this type of analysis?
“Quality is key. When talking about big data, you often hear about three Vs - Volume, Variety, and Velocity. I’d add the fourth V here – Veracity.
“This does not mean all data must be perfect and certified, but it means that the level of confidence on the data the company will use must be known. Data used for statutory reporting must be extremely accurate, but data used for marketing segmentation probably does not need the same level of accuracy.
“Just as importantly, data should be complete. This might mean capturing data from external suppliers like ratings agencies or using statistical techniques to validate sensor readings.”
Given that big data cannot predict the future, should businesses be wary of becoming over-reliant on big data analysis?
“Big data should be seen as a tool for learning from the past rather than a crystal ball to the future, although insights are increasingly provided in near real-time. A combination of good data and good analytical models can make reasonable predictions so long as assumptions are well defined. It is up to the users to understand if and for how long these assumptions will be valid. In some cases, very short term prediction might be more than sufficient and new models can be created continuously.
“In the predictive maintenance case, companies can use complex statistical modelling to extract relationships from past data to predict future failure and perform servicing at low-cost times, for example when a factory is closed. They can also use that historical data to perform root cause analysis and prevent the future failures.
“It’s about being informed and understanding the level of risk. Switching off the brain and just blindly following probably isn’t the wisest idea just yet.”