Small data, big insights
By DANIEL RUSSELL
We have all heard the claims: Big Data is the next frontier for innovation, competition, and productivity. Big Data is a management revolution. Big Data is the new currency. Clearly, the possibilities presented by Big Data are numerous.
Today, there are many examples of how large companies are tapping into Big Data to bring amazing new products and services to market.
Google Translate, for example, exploits word associations in massive databases of free-form text to yield a tool that can instantly translate Icelandic to Indonesian.
The 2012 US presidential campaign has been called the first “Big Data” election.
Organisations such as Netflix and Amazon tap into Big Data to identify the tastes and preferences of their respective customer bases and use this information to provide helpful and relevant real-time offers.
While large, global companies are relying more and more on Big Data, entrepreneurs tend to prefer using instinct to make decisions versus taking a data-oriented approach to decision making.
However, the 2011 movie Moneyball demonstrated that reliance on data analytics provided a clear advantage over human judgment (or instinct).
As George Bernard Shaw said:
“Beware of false knowledge; it is more dangerous than ignorance.”
Business decisions informed by analytics yield more value than decisions informed by instincts.
What about SMEs?
Despite the hype surrounding Big Data, most companies only need “small data”. Small and medium enterprises (SMEs) can also benefit from Big Data-style analytics in the same way as large, high-tech companies.
Access to tools like BigQuery by Google, Hadoop, and NoSQL has made business analytics accessible to tech-savvy entrepreneurs.
Data visualisation tools such as iCharts, Raw, Google Charts, Bonsai, and many others make it easy for budding data scientists to make insights come to life.
While the data and tools are plentiful, many are still intimidated by Big Data.
Deloitte Analytics has worked with SMEs and large organisations who are just getting started with Big Data and provides five key steps to building sustainable analytics capabilities:
1. Identify the desired insights to drive business objectives
Perhaps the data scientist’s most important skill is the ability to understand the organisation’s questions, problems and strategic challenges, and then translating them into data analysis projects.
While some have argued that the scientific method is no longer needed in working with Big Data, we recommend that organisations continue to use specific domain knowledge, creativity, and critical thinking to develop hypotheses which are tested using analytics.
Famous mathematician John Tukey once said:
“Better to have an approximate answer to the right question than a precise answer to the wrong question.”
2. Identify existing data assets and future data requirements
Once one clearly understands the key insights needed to drive organisational performance, the next step is to identify what data exists to help inform the analysis.
Here too, data scientists will need to have good knowledge of the organisation’s operations, systems, and staff to learn what data is collected, how it is stored and accessed, as well as how to determine the accuracy of the data.
Often data scientists will have to use existing data as “proxy variables” to substitute for desired data that is not available.
Examples of “proxy variables” include gross domestic product as a proxy for standard of living, salary as a proxy for job seniority, company tenure as a proxy for expertise, and number of customer visits as a proxy for loyalty.
Clearly, one can think of many examples when the proxy variable would not substitute for the target variable, but generally the relationships hold.
More data may be needed to be able to conduct the necessary analysis and the organisation may choose to access data from other sources.
Many companies today “buy” data when their target variables are not available. Services like Amazon’s “mechanical turk” can help companies find participants online for focus groups and surveys.
Consumer businesses may offer discounts or giveaways for customers who try new products or complete preference surveys.
Big Data projects that combine internal data with external macroeconomic or global trends are more powerful than insights based on internal data alone.
Government agencies or associations are good sources for this type of benchmark data.
3. Implement data management processes
Big Data is not a “once and done” endeavour. The promise of Big Data is to help companies see around the corner of the next disruptive innovation.
Continually collecting data to assess trends is vital to predictive analytics. Data collection must be built into organisations day-to-day processes to ensure accuracy and timeliness.
One of the key hallmarks of Big Data is “velocity”. Velocity in this sense is the need for organisations to continually warehouse, integrate and analyse new data in real time.
With a clear understanding of the key insights needed to run the business today and into the future, as well as the data needed to generate these insights, companies will need to create a roadmap for data management.
Data management is the process of continually collecting the needed data elements, ensuring accuracy, storing them, and updating analyses with new data in real time.
4. Connect analytics, marketing campaign tools, and social channels to engage customers
Using the web to generate interest and provide services to customers has helped propel many startup SMEs to greatness.
A wide variety of tools is available to help organisations better understand their customers’ online behaviour.
Ad preference testing, referral source, page views, time on page, website pathways, conversion tracking, and other analytics provide valuable information to online marketers.
Connecting data from social media channels to website analytics adds further insights.
Adding customer segmentation, loyalty and purchase history can truly drive differentiated performance.
The most sophisticated companies today are leveraging behavioural science and “choice architecture” to provide customers with tailored options that guide them into the most relevant products and services for their unique needs.
“Choice architecture” is simply designing interfaces, policies, and options so that they are presented in ways which help make it easier to choose the right action.
Examples of this include, making healthier food options standard on menus, presenting bundled value pricing first, and renewals for ongoing services automatic.
5. Learn, iterate, expand, and stay nimble
Just as markets and the competitive landscape changes frequently, business and analytics strategies must continually evolve to stay ahead.
Just as it’s better to get an approximate answer to the right question, it’s better to get approximate insights to the future than perfect reporting on the past.
Through practice and iteration, predictions will become more accurate and, more importantly, questions will get better.
Businesses can try new things like experimenting with new marketing channels, new products, and new promotions and are able to get quicker and more actionable feedback to evaluate effectiveness.
Running effective real-time experiments such as these rely on working quickly to evaluate and refine the initiative to capitalise on successes – or stop failures before they do too much damage.
By continually working to improve data quality and expanding data collection and access, organisations will be better equipped to respond quickly and generate the insights needed to stay ahead in the future.
For more Thought Of The Week articles, click here.
First appeared on Leaderonomics.com. Published in English daily The Star, Malaysia, 30 May 2015
Daniel Russell is a partner at the management psychology firm RHR International. He has been researching destructive leadership since 1991 and has published research articles and professional papers on leadership assessment and development, talent acquisition, and workforce analytics.