What Is Analytics Amplifying in Your Organization?

When using analytics for your business, pump up the volume — but don’t pump up the noise.

With the current attention to analytics, it is easy to forget that business processes are the core, not the analytics about them. Analytics is about making an organization do what it usually does, only better.

Analytics amplifies. Data analysis helps organizations understand, predict and optimize by taking data — a collection of signals from underlying business processes — and magnifying it until managers can detect the business implications. Yet as promising as that may seem — and despite the considerable hype — analytical approaches are not a panacea, able to correct fundamental flaws with a sprinkle of mathematical fairy dust.

A cautionary tale can be found in the saga of British supermarket chain Tesco. Once a poster child for retail analytics, recent difficulties at Tesco show that embracing analytics does not make organizations immune to trouble. Tesco has enviable analytics capabilities, but intense competition, changing consumer preferences and potential accounting irregularities have been difficult problems to overcome.

Did Tesco’s focus on analytics contribute to its decline? Or did it stave off an even more extreme fall? The answer is likely “yes” to both questions, and underscores the challenge of applying analytics approaches well in organizations. Analytics may have exacerbated some problems such as kindling a complex loyalty system. But, conversely, data analysis may also have informed it to scale back expansion plans. The amplification effect of analytics remains difficult to manage.

Analytics may be able to take your organization up to eleven, but bear in mind these four principles of amplification.

Amplification needs signal

A Spinal Tap-style metal band goes nowhere if it doesn’t have the right equipment to perform its ear-shattering brand of music. Similarly, companies bent on producing data-driven insights are muted when they find their existing information systems don’t provide the infrastructure to generate the data signals required for analytics to yield novel insights.

The banking industry, in particular, was an early adopter of information systems, but these older systems can be inadequate for current needs. Citigroup, for example, struggled to make fragmented legacy systems support more sophisticated emerging needs. But despite limitations from existing systems, replacing core infrastructure is daunting. As a result, banks such as Deutsche Bank end up with a mishmash of legacy infrastructures and modern systems instead of a unified approach.

The takeaway: What does the necessity for data signals require of organizations? Get robust information systems in place as a foundation for analytics.

Amplification is agnostic

The music I like should be cranked up LOUD. But not the dreadful stuff the neighbors listen to. Unfortunately, amplifiers don’t know the difference and will make both louder; good becomes better, bad becomes worse.

Raw data contains both signal and noise. Analytical processes can attempt to distill meaning and insight from both, making it difficult to differentiate between the two. As a result, a decision may be made on the basis of poor data quality rather than true information.

Knowing what to amplify is particularly difficult when decisions are automated. Automated pricing, for example, offers rapid response to market conditions, lower costs, and improved transparency. But algorithmic feedback at Amazon amplified the noise, resulting in massive pricing irregularities. Noise was driving pricing, not the market.

The takeaway: How can organizations avoid amplifying noise? Understand the processes generating data to filter out misleading information.

Amplification should be invisible

Audible feedback from amplification quickly causes people to cover their ears, bringing back memories of garbled announcements over low-quality speakers in homeroom. With rare exceptions, people are interested in the underlying signal and not the amplification itself.

Having put considerable effort and expertise into complex models, it can be hard for data scientists and analysts to push analytical complexities into the background. But the background is where these details belong. The findings and results are what matters for decisions makers.

The takeaway: How do organizations resist the allure of analytical vainglory? Focus attention on the business implications of analytical results, not the processes used to create them.

Amplification can distort

It is painful to hear amplification when it was not intended. These previously hidden moments often get intensive attention and shape our impressions, perhaps too much.

Analytics within organizations can similarly distort managerial thinking. Some processes in organizations are more amenable to analytics than others because outcomes are measureable, data is accessible, or processes are clear. These signals get amplified. As a result, these processes get attention — attention that is driven by convenience not consequence.

The worst noise may be convenient signals of secondary importance. Amplification of these can drown out the important signal. In marketing, analytics is a powerful tool to help get more attention, with a focused message, for less money. But if your organization’s product won’t satisfy customers, effective marketing analytics just gets more people disappointed faster. In logistics, analytics helps get products to customers faster and cheaper. But if those products are defective, efforts are wasted.

The takeaway: What can you do to avoid conclusions driven by distorted signals? Use analytics first to ensure core products and services are excellent and squelch signal from elsewhere in the value chain.