Businesses should focus on small data for better insights

Atanu Biswas

Professor, Indian Statistical Institute, Kolkata

THE toys manufactured by Danish company Lego Group are mostly made of interlocking plastic pieces. The firm was on the verge of bankruptcy in 2003. It decided to visit the homes of consumers across Europe. Company officials met an 11-year-old German boy and asked him what he was most proud of. The boy showed them an old, worn-down pair of sneakers and said: “It shows I’m the best skater in town. If I slide down the skateboard, I am number one, and this is my evidence.” That discussion with the youngster served as an inspiration for Lego to ‘shrink’ its creations to the size of little bricks and produce The Lego Movie, which eventually achieved a turnaround for it. Is that the magic of ‘small data’ in this age of big data?

Over the past 20 years or so, big data has become a catchphrase. In order to formulate successful strategies in every domain of life, people want to analyse a lot of data. However, it becomes unmanageable because, regrettably, both our computational capabilities and statistical expertise are insufficient to handle massive amounts of data.

In contrast, the insights obtained from ‘small data’ are becoming increasingly popular for framing strategies in certain businesses. Small data is actually the data that is ‘small’ enough for human comprehension, and is easily accessible, understandable and actionable for everyday tasks.

In a 2013 piece in The Guardian, British economist Rufus Pollock perceived that small data represented a real revolution. “This next decade belongs to distributed models, not centralised ones, to collaboration, not control, and to small data, not big data,” he wrote. Remember that by then, the much-hyped big data project Google Flu Trends (GFT), launched in 2008, had proved to be a failure. By aggregating Google Search queries, GFT attempted to accurately predict flu activity. It failed because people often search for symptoms of illnesses that resemble the flu but may not actually be so. And people realised that big data might not be the holy grail of data science.

Big data is all about finding correlations, but small data aims to find the causations that are rational enough to be comprehended in the context of a particular business and may be analysed for insights that lead to improved decision-making.

A 2016 book, Small Data: The Tiny Clues That Uncover Huge Trends, by branding guru and best-selling author Martin Lindstrom, may have provided the best explanation for collecting and analysing small data. Time magazine described him as a “modern-day Sherlock Holmes… an original and inquisitive mind harnesses the power of ‘small data’ in his quest to discover the next big thing.” At least conceptually, Lindstrom’s small data technique is straightforward. He suggests that spending time with people in their natural settings can provide valuable marketing insights when paired with attentive observation. Lindstrom spends 300 nights a year in strangers’ houses, meticulously examining every detail to unearth their secret needs. He is hired by the world’s top brands to discover what clicks with their customers.

Small data is about humans, in contrast to big data, which is about machines. Many businesses can benefit from a more straightforward strategy, even though big data has proven beneficial for large multinationals. Small data can really be found in very large datasets if we try to learn more about a small subsample that has a unique trait. Reducing the data into small, visually appealing objects that reflect different features of massive datasets is an excellent method for understanding big data.

According to Lindstrom, traditional market research techniques, such as focus groups and surveys, frequently fall short of capturing the psychological and emotional elements that influence consumer behaviour. Organisations can enhance comprehension of their consumers’ requirements, aspirations and motivations by gathering and scrutinising small data, including social media posts, customer reviews and other online communication channels.

Lindstrom uses personal anecdotes to highlight the significance of small data. In the case of a large chain of hotels, he employed small data on the items in minibars and looked at how guests utilised the amenities in their rooms to learn about their preferences and behaviour. The hotel experience was then improved by utilising this data to add additional in-room entertainment options and a larger assortment of minibar items.

The book includes examples such as how a stuffed toy in a girl’s bedroom turned the tide for a fashion retailer’s 1,000 stores in 20 countries; how a noise reduction headset led to the creation of Pepsi’s trademark signature sound; and how a magnet found on a fridge in Siberia led to a revolution in US supermarkets. Lindstrom describes small data “as seemingly insignificant observations you identify in consumers’ homes... is everything from how you place your shoes on how you hang your paintings”.

According to Lindstrom: “If you take the top 100 biggest innovations of our time, perhaps 60-65 per cent are really based on small data.” But while acknowledging the captivating allure of small data, it’s crucial to keep in mind that the primary issue is that the sample may not accurately reflect the true distribution of data within the population.

However, as small-data techniques develop, more and more industries and business processes will use them because of their greater efficiency, accuracy and transparency. Consider the design of new consumer products, the recovery of industrial images, the identification of defective factory machine parts and many more applications.

What’s ahead for us? “Mastering the human dimensions of marrying small data and AI could help make the competitive difference for many organisations, especially those finding themselves in a big-data arms race they’re unlikely to win,” says a 2020 article in Harvard Business Review. Yes, in the age of artificial intelligence, human-centred small data approaches could be very beneficial. Merely 10 or 20 well-crafted samples may serve better than millions of noisy data points. One wonders whether that represents the triumph of human intelligence as well.

Join Whatsapp Channel of The Tribune for latest updates.

#Europe