Ah, the future…
I marvel at how “the future” occupies so much of our focus, and yet it always looms just out of reach, never really “here” and always “out there.” It’s never quite real, the future; at least, not really like the present is perpetually real. In fact, the present is the only real thing we consistently have. From the dawn of human thinking, we have worried about the future. The ancients walked miles to see the Oracle at Delphi. Seers and prophets were some of the most revered figures in antiquity. Even today, people pay good money to have fortune-tellers describe their tomorrows.
It seems almost everyone wants to know their future. And the same is true for businesses. Knowing the future means being able to plan, change course, and get ahead of the competition.
But unfortunately (or maybe fortunately), predicting the future is not something reliably achievable by oracles, seers, prophets, and fortune-tellers. It is also notoriously hard for CEOs, CFOs, investors and consumers. When it comes to knowing the future, we are all at the same disadvantage.
Fortunately, However, by Utilizing Lots of Data, Computers Can do Better
“Data” is fashionable these days, especially “big data” (which just means mountains of it), but it is not magic. Data can be thought of as “traces” or “records” of what came before. When the data reflect your customers’ past needs and behaviors, computers can discover patterns and predict likely future needs and behaviors. In fact, data science, and machine learning in particular, are largely devoted to extracting patterns in data so that those patterns can predict the future.
Here is a simple fictitious example. Let’s say you have a scatter plot of (x, y) points on a graph where “x” represents “time on site” (in minutes) and “y” represents “dollars spent.” A computer can recognize a pattern in this data. (For simplicity, let’s say it’s a line, but it could be a curve or other shape.) Then, having this pattern, or “model,” in the form of our equation for a line (y = 16.706x – 6.5922), we can predict the future dollars spent based on how long a new visitor has remained on our website. Let’s say that a new visitor has been on the website for 12 minutes. Our model predicts that they will spend about (16.706)(12)-(6.5922) = $193.88. We can then use this prediction, perhaps by offering a shipping discount, related items, or flash sale. We can also update our model based on this new data point after this customer completes his or her purchase.
This simple example used two continuous variables, “x” and “y”, which stood for “time” and “dollars,” respectively. But more sophisticated predictive methods can handle data of almost any type, including unordered categories, ordered categories, counts, proportions, and ratios. Also, beyond just “x”, we can have numerous inputs to inform our predicted outcome, or even predict multiple outcomes simultaneously.
So where can modeling your past customer needs and behaviors make a big difference to your bottom line? A fantastic application is for website and mobile self-service help.
If your business is growing, you already know that you cannot scale your customer support team to manually handle every inquiry that your website or mobile users may have. And why would you want to? Forrester reports that self-service has increased from 67% in 2012 to 76% in 2014. A 2013 report entitled “The Real Self-Service Economy,” which surveyed 2750 consumers, found that 70% of consumers expect a company to include self-service on their website. These numbers have undoubtedly only grown. High-touch personalized help is appropriate if all else fails, but as a first line of defense, your customers must be empowered to help themselves—saving both you and them time, frustration, and expense.
Unfortunately, many of today’s self-service help solutions are not predictive at all. They exist as stand-alone knowledge bases where less than 1% of your customers will bother entering a search. Every customer starts in the same place—at the topmost-level of the knowledge base, entering search terms, hoping that they can adequately describe their help need. The search results returned often badly miss the mark, and require more effort to wade through. Even if customers get lucky and find their answer, this is a dissatisfying experience.
Now Consider the Power of Predictive Self-Service
With a context-aware predictive help solution like AnswerDash that resides on-page or in-app, both customers’ past help needs and answer-seeking behaviors and their current context can be leveraged to predict which questions they are most likely to have in that place and time. AnswerDash serves up the most common questions—along with their answers—so that customers can access them in just a few clicks, without typing a single character to describe their help need. AnswerDash constantly observes your customers’ on-site or in-app behaviors, updating its predictions based on what questions and answers are used, and what behaviors take place before and after.
The result of all this observation, modeling, and prediction is an incredibly fast, user-friendly, context-aware system for getting your customers the answers they need, right when and where they need them. Customers get their answers without ever leaving the page they’re on—at their point of action—becoming far more likely to complete successful purchases. This is why AnswerDash has been able to lift revenues from 5%-20% on websites that have deployed it. (For example, see our Tire Buyer case study.)
Although no person or machine can be the Oracle at Delphi, predicting your customers’ self-service needs can make your customers feel like they have met their personal fortune-teller, showing them that their online future is indeed happy and bright.
Let AnswerDash Help You Harness The Power Of Predictive Self-Service
AnswerDash predictive self-service support is proven to create a frictionless way for customers to help themselves on the web and in mobile Apps. With AnswerDash, your customers get smart self-service that is relevant and personalized right at the point of action, without having to leave the page or type to search. By improving the “Help UX,” AnswerDash customers are reducing customer friction and achieving a 30% - 50% reduction in support tickets and cost. Learn more at www.answerdash.com.