In the world of data, as data storage costs decrease faster than data processing costs. This creates pools of data that can be stored, but not efficiently analysed. Oftentimes these pools of under-analysed data are called Big Data.
Many companies are now trying to put their Big Data to work to uncover actionable insights, but it’s not as simple as it seems. The challenge is shifting from disparate data sets to leveraging data in a way that produces better insights and ultimately better decisions. To solve this challenge, companies need to build processes that effectively analyse a vast array of ‘Connected Data’ – a system of rich data sources (on sales, margin, transaction composition, weather, demographics, loyalty, guest satisfaction, share of wallet, etc.) that are effectively linked together to generate actionable business insights.
In order to balance the immensity of scale with the need for actionable insights, it’s crucial to remember that in data analytics, richer data sets beat larger data sets (e.g., building data sets about competition and customer spend across your industry will serve you better than even more lengthy social media feeds). Collecting relevant data points, from multiple sources that are impactful to your business, will enrich your analysis and your insights. From there, it’s crucial to employ smart data structures and nimble software solutions to leverage these assets to the fullest. The goal is to build the right data and make it accessible through the right software platform.
First, in order for an organisation to get the best insights from their Connected Data, it’s key to record the right information. Typically, in-house data include metrics such as sales and margin, labour, transaction composition, customer satisfaction and loyalty. To further enrich analyses and develop competitive advantage, organisations can combine in-house data with additional data sets on characteristics such as weather, competitors and demographics. Some leading organisations are even incorporating anonymised and aggregated insights about customer spending outside of their four walls to understand whether share of wallet is growing, which new markets present opportunities, which customer segments have the largest headroom to grow and more.
This data can then be cross referenced to create insights, but the analysis will only show part of the picture and is not statistically reliable. For example, suppose a retailer notices that some of its stores saw a sales uplift. Cross-referencing the data reveals that stores with sales uplift ran more promotions. Promotions may have accelerated sales, but the uplift may have also been a coincidence. In reality, the increase could have been driven by a combination of different factors, such as competitor closings nearby, good weather, changing management, more staff on the sales floor, etc. The only way for any company to determine the true incremental impact of an initiative is to conduct statistically rigorous test vs. control business experiments.
It’s important for modern businesses to be aware that data needs to drive decisions rather than drive them crazy. To make data driven decisions, companies can leverage their Connected Data to conduct business experiments using a Test & Learn approach. Test & Learn – the process of scientifically testing a new business initiative in a subset of the network before broader rollout – is a concept that has been applied in the field of medicine for many years. In the business arena, an initiative will be trialled with some markets, stores, or customers – say stores in this case. The performance of each “test” store is measured against a well-matched group of control stores, allowing companies to understand the true incremental impact of the new initiative on economic outcomes. The richer the data sets, the more granular and multifaceted the insights and derived recommendations from Test & Learn.
When conducting a statistically rigorous business experiment, the insights executives can achieve are plentiful. For example, returning to the retailer who ran a promotion, if the retailer records store- and transaction-level data, analysts can use Test & Learn to measure the impact of the promotion on sales and margin. The retailer can also investigate if the lift was driven by greater transaction count or spend per transaction, as well as analyse the impact of the promotion on its different categories and sub-categories. By incorporating initiative- and store-level characteristics, the retailer can go one step further to identify which types of offers (e.g., BOGO vs. 50% off) were most successful and which types of stores (e.g., smaller vs. larger, rural vs. city, etc.) respond best to the promotion.
New ideas are risky and can come with high costs. In fact, companies across industries report that over 40% of ideas do not break even. As such, one of the tangible benefits of conducting test vs. control analysis is that it prevents companies from introducing ideas that, as they currently are, will fail. Test vs. control analysis allows companies to understand the true incremental impact of an idea, which components of the idea work best and where it should be implemented—to improve profitability.
While organisations are spending more time and money in this area, it’s important to do it right if you want to succeed. First, you should start with the data you have. Then leverage the data to conduct business experiments using a Test & Learn approach. Lastly, build a funnel of new ideas to test to continually improve profits.
By Rupert Naylor, senior vice president of Europe at Applied Predictive Technologies (APT)
Want to keep up with the latest ideas in digital marketing? Free conference and exhibition Integrated Live is the place to be.
PrivSec Conferences will bring together leading speakers and experts from privacy and security to deliver compelling content via solo presentations, panel discussions, debates, roundtables and workshops.
For more information on upcoming events, visit the website.
comments powered by Disqus