Solve specific problems around targeted business issues with minimal supporting data.
While placing my order for a $2,000 fish-finding tool, I envisioned walleyes jumping into my boat faster than I could toss them in my cooler. After a few frustrating afternoons on the water, I realized I would have been just as well off with a simple $149 model. Shelling out two grand for a tool loaded with bells and whistles would make sense if I were deep-sea fishing for tarpon. But deploying it for walleye was like using a bazooka to kill a mosquito.
Depending on what information they’re looking to reel in, contact centers investing in “big data” may also experience buyer’s remorse. Big data is all the rage these days, but for many organizations, big data is overkill. What it often produces is big expenses and big headaches.
Indeed, big data, which is engineered to optimize predictive analytics and other sophisticated methods for extracting value from unstructured information, requires a big, big effort. Collecting a critical mass of data and gaining clarity about how to massage it to achieve predetermined objectives is a big job in itself. The processing piece that comes next demands highly skilled analysts who know how to discern operational subtleties and find the proverbial needles in an endless supply of data-stuffed haystacks. That’s a tall order for organizations of any size, much less contact centers whose size and budget don’t allow them to even consider jumping on the big data bandwagon.
A better option? Skinny data. More specifically, skinny speech data. Unlike big data, which is essentially a collection of data sets so large and complex that they become awkward to work with using traditional database management tools, we can define skinny data as solving specific problems around targeted business issues with minimal supporting data.
A data source that’s particularly well suited for skinny data applications is speech analytics, which enables you to move from data creation to easy-to-interpret results to business decision faster than you can say, “I can’t believe how inexpensive this is.” In fact, valuable “skinny” bits of information are probably in the speech tool you’re already using.
Leveraging skinny speech data can help contact centers improve agent effectiveness, minimize compliance risk, capitalize on selling opportunities, identify complaint trends, reduce customer churn, decrease operational costs and mine rich new veins of business intelligence.
Conversations I had with data science executives from two large insurance companies demonstrate the folly of pursuing big data when skinny data can get the job done quickly and efficiently. The first executive told me he wanted to process large customer data sets to predict churn using voice, text, email, demographics, tenure, surveys and other forms of data.
The second executive was confident he could leverage speech analytics to generate valuable insights about churn and other business issues. He told me, “We already have so much big data that we don’t know what levers to pull right now. I just need some basic information that’s aligned with our KPIs (key performance indicators).”
Bingo. Skinny data can help you cross the finish line while big data is still generating a multitude of reports that are just as likely to confuse as impress. The contrast is stark: why sink $1 million or more into an enterprise CRM when actionable targeted analytics can be achieved by simply combining customer interactions and sales data?
Granted, if you want your analysis to uncover issues that are currently unknown and unanticipated, that’s a job for big data. But skinny speech data is ideally suited to boost contact center profitability by applying basic but effective methodologies to issues like first-call resolution (FCR), root cause of calls, uninformed or unprofessional agents and customer frustration.
Skinny Speech Studies
A speech analytics pilot study built around an important business issue can help you discern the “why” behind the “what” and serve as a catalyst for change. For instance, if your rate of repeat calls is trending upward, an exploratory study can help identify the root cause of the higher rate and provide the necessary insights to take corrective measures.
Start your study by identifying an issue that relates directly to your KPIs, then spend two or three weeks searching through agent-customer conversations to find calls that support the study’s objectives.
Next, fill a bucket with 75 to 100 relevant recordings to share with your company’s decision makers. Invite them to sit around a table and listen to selected calls. Actually hearing the voice of the customer is far more impactful than reading reports or interpreting a set of dashboards.
Presenting the study with a compelling narrative interspersed with skinny pieces of data—snippets of audio and nuggets of analysis—can build awareness, initiate discussion and move your team to action.
While many organizations apply speech analytics to ongoing, long-term initiatives, many quick wins have been gained along the way, many of which have made an immediate impact on the bottom line.
Here are seven industry examples that may spark ideas about how you can use skinny speech data in your own organization to solve key business issues.
- Higher Education. Business analysts exposed agents who were making potentially litigious “guarantee” statements. Those agents were taken off the floor the next day.
- National Retailer (consumables). Analysts quickly uncovered a recurring issue related to packaging/shipping failure.
- National Retailer (apparel). Analysts discovered an opportunity to reduce callbacks with a simple change: the retailer’s website offered a next-day shipping option when no such option existed.
- National Retailer (home improvement). Analysts recognized an opportunity to identify customers who were threatening lawsuits. After shifting focus of the initiative to “small claims court” phrases, the client was able to rectify the marketing issue that was causing the problem.
- National Retailer (wellness). Analysts saw an opportunity to curtail the ongoing problem of agents making medical misstatements that left the client vulnerable to significant legal consequences.
- Collections. A chief compliance and privacy officer was aghast when told by his analysts that a number of his collections associates were routinely using phrases that could be considered false threats, such as “Take back your car.” According to this executive, the dollar cost of an enforcement action could run as much as $10 million.
- Telecommunications. During the detection auditing process, an analyst discovered agents who were using unprofessional language and inappropriately commenting on customers’ bad credit ratings.
Unlike long, complex initiatives that take nine to 12 months and produce results that don’t always justify the expense of producing them, skinny speech projects can generate fast, low-cost wins in four weeks or less and demonstrate to key stakeholders that your speech software is generating a healthy ROI.
Two Skinny Stories:
All speech analytics software tools are designed to search recorded conversations for keywords and phrases that determine customer intent and predict outcomes. Most tools also allow the user to build language patterns and algorithms to provide deeper insights into each business issue. That can take time but the payoff can be significant.
For example, a large health care company attempted to predict churn using nothing but skinny data. Within three months, they successfully identified patterns that could predict a customer who was about to churn with 38% accuracy. Would other forms of data augment these results? You bet. Accuracy could unquestionably be improved by spending considerable amounts of time and money to tediously mine large repositories of relevant unstructured data. But why go that deep? The inexpensive insights produced with skinny data had the potential to save millions in revenue.
If an end user’s speech tool and servers have enough capacity, they can track an almost unlimited number of phrases. I’m aware of a “no boundaries” user who was new to speech analytics enthusiastically kicked off our relationship by brainstorming a list of more than 400 phrases related to a wide range of issues.
They decided that novice runners shouldn’t try running a marathon on their first day in sneakers. Searching for 400 phrases would bury them in a virtual avalanche of data. It was decided to temporarily set aside entire categories they considered nonessential, and hold off from including certain phrases that occurred infrequently. They ranked their categories by business priority.
Net result: the user was able to produce an abundance of high-value actionable intelligence from a manageable amount of search results.
In the months that followed, this company gained valuable hands-on experience, learned both the strengths and limits of their new technology, recognized which categories where truly more relevant to their business goals, and determined the best way to integrate their new software with their current processes.
The final tally: Skinny Data two, Big Data zero.