Are you finding money-making data insights fast enough?
- Posted on October 30, 2018
- Estimated reading time 3 minutes
There is a company that supplies soap and cleaners to restaurants. It had a hunch that there was a connection between how a restaurant purchased its products and how likely the restaurant was to pass its next health inspection. Could the connection be found in a way that helped both the restaurants and the supplier?
As it turns out, it could. Avanade worked with the supplier and that found that a restaurant that reduced its typical two-year soap order by 50% or more in the month prior to an inspection had a 50% greater chance of failing that inspection. A restaurant that maintained the typical order level had a 94% chance of passing inspection. The supplier had exactly the information it needed to target the restaurants most likely to increase their purchases and improve food safety in the process.
“You just made us two million dollars,” the supplier told us.
If you think analytics projects need to be massive, months-long affairs, think again: This one took just five weeks and five data sources. The returns to the company were high—and typical. When it’s done right, a company should generally be able to see a 50X financial return on this type of analytics project: a fast, focused approach to find value hidden in a company’s data. We call this approach the Insight Discovery Process (IDP). How fast is IDP? Its combination of design-led analytics workshops, value targeting and storytelling generally proves its value in about three to six weeks.
The soap company used IDP to gain actionable customer insight, but IDP’s application is broader and can include subjects such as marketing performance, operational efficiency, and just about any other organizational issue for which big data can be mined. And this shouldn’t be of interest to just the CDO or CIO. IDP is about making analytics part of everyone’s job and encouraging experimentation throughout the company.
The IDP works best when you keep these five guiding principles in mind:
- Treat the integrated “data exhaust” from every customer interaction as a cash equivalent. This is an idea that used to generate eye rolls; no more. Every customer interaction leads to others, forming a sort of data trail, like a car’s exhaust. Think of a call center contact that leads to a payment, a shipping transaction, post-sales service, etc.—all logged and stored in a database. Integrate it with other data (like integrating soap sales with inspection results) to find the patterns that will generate financial returns.
- Focus on the small but important information hidden within big data. Big data—more data than you have time to analyze—is nothing new. What’s new is focusing on the signal-to-noise ratio of your data—the relevant to the irrelevant—to generate insight. How do you determine what’s signal and what’s noise? By focusing on succinct, use case-based investigations. For example, you might start with a simple hypothesis such as “clients that purchase more cleaning products than their two-year monthly average are more likely to pass a subsequent health department inspection than those that purchase less than their two-year monthly average.” Of course, keep in mind that the signal that helps uncover millions of dollars of opportunity today can, and likely will, be tomorrow’s noise.
- Don’t build an Olympic monument. We recommend you don’t make massive, time- and money-consuming-investments—in analytics specifically or IT more generally—without explicit, ongoing value measurements and data-driven business cases to prioritize them. Otherwise, your ROI is likely to disappoint you. Think of Olympics facilities built for millions, used for a month, and then largely forgotten. Worse than wasting money, the IT-as-monument approach can lead you to miss opportunities to do things faster and cheaper. It also forces you to give up on the idea of generating returns in weeks rather than months or years.
- Put experimentation before industrialization. A fast-fail mentality is a key to success. Experiment, iterate, find what works and what doesn’t, and don’t attach negative consequences to approaches that don’t pan out. If you’re working fast and light, you can afford the false starts that can end up teaching you the most. Most false starts tend to be great teachers.
- Start now. Don’t get caught in analysis paralysis. If you’re like the vast majority of companies, you have nothing to gain by putting off the initial, low-cost aspects of data exploration. The cloud is ideal for the “think fast and small” approach and cloud providers are doing a better job every day of addressing the diminishing number of challenges—such as regulations around data sovereignty—that used to keep companies off the cloud. If your corporate culture is infused with an overabundance of caution, try a process such as Data Value Assessment, championed by the MIT Center for Information Systems Research, which helps companies to move ahead with business analytics and big data in impactful ways.
If IDP was worth two million dollars to the soap supplier, how much could it be worth to you?
Learn more at https://www.avanade.com/en/technologies/data-analytics.