5 Powerful Ways That Data Analytics Can Help Your Business, and How to Capitalize On ThemAug 02, 2022
When I tell people what I do for a living, they usually smile, nod their heads, and change the subject.
As soon as they hear a word like analytics, they tune out.
Even my mother gets frustrated (“but what can I actually tell people about what you do?”), mumbling that it was easier when she could say I was a college professor.
All kidding aside, while the specifics of business analytics are foreign to most people, business leaders should be acutely aware of how data analytics can help their business.
Make no mistake, data analytics are becoming increasingly vital to every industry.
What data analytics can do
Put simply, analytics are tools to better understand the world around us.
In some ways, data analytics are like detective work; it helps us figure out what happened, or “who done it.”
In other ways, data analytics are like fortune-telling, predicting what might happen next.
I like to think of analytics as business intelligence.
Sometimes, analytic results confirm what we already suspected.
Other times, what we learn comes as a complete surprise.
Either way, data analytics allow us to focus our efforts more strategically and make better decisions.
Let’s look at some powerful categories of analytic value, and also what companies need to do to capitalize on that value.
What Analytics Can Do
1. Illuminate. (Example: Five percent of people incur 50% of medical costs.)
Data analytics help shed new light on a topic simply by describing the data and how they are distributed. Even simple discoveries can provide important insights.
I recall a meeting with the CEO of a large pharmaceutical firm, where we described the number of people employed by their organization each year. He had been told they had about 25,000 FTEs. However, if we included all employees who were employed for at least one day, with transitions in and out, the number was 15% higher. Those unsuccessful workers---some coming in briefly and leaving---had never been brought to his attention, and he became interested in their onboarding and retention processes.
Another important role is providing new ways of thinking about a topic.
Take the usual practice of describing things with their average, even if the average is not typical.
For example, on average, an employed person will spend about $12,000 on health care in a year. However, spending is extremely lopsided. The five percent most-expensive people will spend 50% of all the dollars. And the bottom 50% percent least-expensive people will only spend 3% of all the dollars. The “average” is higher than 80% of people—which is not typical.
When we understand these facts, it makes us think differently about how different people need different types of support (e.g., a focus on the top 5%).
By looking more closely at our data, we discover important patterns that give us new ideas.
Maybe a portion of the data are missing.
Maybe outliers are making a metric look especially good.
Maybe we see a high number of dropouts in our programs.
Digging into data helps us get smarter.
2. Characterize. (Example: Who is using our service?)
Who does what? Who are your best employees? Who are your loyal customers?
Understanding more about people and behaviors helps businesses think strategically.
For example, to market a service effectively, we need to understand who uses it, how often, in what way. With that knowledge, we can make decisions about targeting, messaging, or repositioning. We might also learn more about why people aren’t using the product.
In one example early in my career, I was analyzing data about use of a low-cost onsite fitness facility at a large insurance company.
It was 1987 and the concept of a company-sponsored facility was new. Despite state-of-the-art equipment and a centralized location, enrollment was not as good as we expected.
Our hunch was that people unaccustomed to visiting a gym were reluctant to try it, which would suggest that our users would be younger men.
However, a quick comparison of users to the overall employee population showed that indeed more males were enrolling, but they were not young.
Something else was going on.
Digging deeper, using a comprehensive database about employees, our analysis found three predictors were most important in characterizing who joined: salary, tenure, and location.
Highly paid, long-tenured employees in nearby buildings were our frequent users. Location made sense.
But we wondered if the price was still too high, discouraging low-wage workers.
This finding led us to run focus groups with low-paid non-users to understand the barriers they faced.
Turns out, the issue was not money. It was time.
Senior managers (mostly older males) felt free to spend more than an hour at lunch time doing a workout.
But they were not as tolerant of their subordinates taking extra time.
Eventually, this led to a policy change about flexibility of lunch duration in return for starting or ending work at a different time.
Because we had data, we were able to identify an unknown barrier and develop a solution.
Businesses need comprehensive information about their employees and customers to find accurate insights about their preferences.
3. Differentiate (Example: Do bonuses really improve sales performance?)
Companies need to understand differences. Is one location performing better? Do employees at the home office stay with the organization longer? If so, are those differences attributable to other things?
Analytics allow us to test differences to see if they are real:
1) more than small random differences;
2) not caused by other factors.
Statistically, an analysis can determine if a difference is significant, which means that we would not expect to see this type of difference by chance.
Also, the analysis can examine whether the difference would exist if other conditions were the same (called “controlling” for other things).
As an example, a national retailer allowed locations to have different compensation practices.
At a few of their stores, employees were given a small percentage of sales as a regular bonus.
Executives wondered if those bonuses resulted in higher sales because employees had an incentive to sell more.
First, we needed to understand which employee and team characteristics were associated with higher sales overall.
Using the variety of datasets that were available, we found that sales revenue was dependent on the department (what products were being sold), employee tenure (how experienced they were), team size (there was an optimal number, so they weren’t over- or under-staffed), and a few other team characteristics.
Next, we controlled for those important characteristics (department, tenure, team size, etc.) and tested whether a bonus had an independent effect on revenue.
The answer was that, yes, bonuses increased overall revenue. These results altered how the company designed compensation, and also how they structured teams within the department.
Again, these informative findings were the result of having comprehensive, integrated data combining human resources, payroll, and business outcomes, by department.
4. Quantify. (Example: Ideally, how large a bonus should employees receive?)
In the previous example, our analytics showed that offering bonuses increased revenue.
However, leaders wanted more information to guide decisions about how such bonuses should be designed. How much bonus is the right amount?
Using the same models from that analysis, we could quantify the expected increase in revenue per 1% of sales allocated to bonuses for the team.
Further, we were able to model the distribution of those bonuses across team members of varying seniority.
As a result, we could propose the level of revenue to be allocated as bonuses (up to, but not more than, 3%), such that there would be a net gain to the company.
Also, we were able to suggest a distribution of the bonuses that favored longer-term employees.
This is one example among thousands that happen every day.
I’ve been part of analytic projects that quantified everything from the excess medical costs associated with cigarette smoking, to the productivity impact of ineffective managers, to the lost workdays from COVID-19, to the reduced number of accidents expected from better safety training.
To make good decisions, we need clear information about the magnitude of differences.
Every choice a company makes is a trade-off of time, resources, or money.
When we use analytics to understand how much of an impact a choice will make, it makes us smarter and more effective.
5. Predict (Example: Who is going to quit in the next 90 days?)
Forecasts help us plan. The more accurate the forecast, the more prepared we can be.
During the recent ‘great resignation,’ where a large portion of the workforce quit, companies have struggled to retain and hire enough talent. In customer-facing industries such as retail, hospitality, call centers, and health care, the effects have been disruptive to normal operations.
In one large call center organization, analysts examined patterns of quit rates to identify and quantify important factors that foretold near-term choices to quit. A combination of many (30) factors contributed to the likelihood of quitting, including employee engagement, job satisfaction, tenure, recent absence, whether they had received a recent raise, employee characteristics, and other team factors.
The model was 90% accurate in finding people who would quit in 90-180 days. With only a small number of false positives (someone who was predicted to quit but did not), this model was among the most effective I have seen in practice.
Using these predictions, human resource professionals were able to devise outreach plans and manager training to understand, and perhaps mitigate, upcoming turnover.
I have seen many successful examples of prediction using data analytics: who was likely to be injured, who would be successful in their job, who might develop an illness. If we can measure it, we can probably predict it.
Essentials for successful analytics
To capitalize on analytics in business, companies should consider three essential steps.
- Integrate data company-wide.
Readers likely noticed that most of the sample insights mentioned here used data from a variety of sources. We couldn’t understand how bonuses and workers characteristics influence revenue without linking HR, payroll, and business data. We couldn’t understand how job satisfaction, health, worker engagement, performance, and absence predicted turnover without having all those sources in the same database. Data integration is becoming widely recognized as a business imperative.
You get the picture.
Otherwise, you will remain in the dark.
- Integrate your people.
Too often, analysts and business professionals operate in separate divisions, interacting via transactional ask-and-answer requests. Analytic teams need a better appreciation of business needs. Business professionals would benefit from an understanding of challenges faced by analysts. There is a pervasive us-and-them attitude in many organizations.
Things work better when we see ourselves as one team.
- Hire or train an analytic translator.
Having a person who is dedicated to understanding both worlds and speaking both languages can facilitate better use of data and analytic results. Rather than hoping your teams have sufficient data literacy, having an analytic translator guarantees that someone is bridging the gap. Analytic translators work hard to discover what’s needed and decipher results into usable business strategy.
Make sure there is at least one person dedicated to making analytics work for your business.
For more information about what analytic translators do, download the first section of my book for free here.