The evolution of chatbots has made them valuable tools for collecting business-shaping data
Not all chatbots are created equal. While most chatbot experiences are similar on the user end of the equation – that avatar of a pretty girl or fresh-faced young dude sliding into a browser window soon after someone lands on a website – what happens on the back-end can turn bots into valuable sources of client data.
According to Vivek Sriram, chief product officer at Lucidworks, the misconception of chatbots as cheap, two-dimensional replacements for admin staff persists in the mortgage industry, and not only because of lenders’ traditional aversion to tools that further diminish the importance of face-to-face relationship building.
“Technology companies like us compound this problem because all we do is talk about technology,” Sriram says. “We don’t necessarily say, ‘If you’re a mortgage provider or a bank or a processor, here’s a way to use the technology that makes a specific workflow that’s unique to your particular business better.”
Sriram says chatbots, if programmed properly, can glean specific information from clients that allows lenders to answer broad questions about those clients’ needs or the quality of their website interactions. But they can also drill down further into the data and help mortgage providers generate far richer information.
The key, he says, is for lenders to collaborate with their tech providers to formulate targeted questions about their business and have bots and the tech infrastructure behind them find the answers. For example, a lender may want to gauge its brand awareness or client retention rates. By fine-tuning a bot, it can solicit the appropriate information, aggregate it in the proper way and create a set of reports that shed light on whatever facet of the business needs clarity.
Bots can also be used to evaluate their own performance. If a lender wants to compare the number of sessions initiated by a chatbot that led to a conversion event to the number of conversions that didn’t involve a bot, all it takes is a simple A/B test.
“You can find that out within a couple of weeks,” Sriram says.
Bots can also be used to answer consumer questions in an efficient way. Take the theoretical example of a lender who discovered through a customer survey that the information on its website is failing to answer customer questions. A bot could remedy that shortcoming, acting as a conduit between lender and client that provides the kind of direct answers that were previously lacking. These bots can also inform clients of new offers and track the response they receive, helping shape future product decisions.
Sriram says there is always the chance that vendors pedalling chatbots and other virtual assistant technology may oversell their products. Like the bots themselves, lenders should be prepared to ask the kinds of questions that will get them the answers they’re looking for.
“As a lender, you want to look at it with some degree of scepticism,” he says. “Not cynicism, but at least be sceptical about it. ‘How would I do that? What are the metrics that are actually important to my operations?’”
Because the effective use of bots involves data collection and organization, the costs involved vary based on the scale of the business and the amount of data being generated. Sriram says a massive company with a diverse product portfolio, like Intuit, would likely need to spend millions on an optimized selection of integrated bots.
“For them, the investment in a set of technologies that allow for intelligent conversations cannot be isolated to one of their business units,” he says.
For smaller operators, like a credit union with a modest selection of products, Sriram says the investment should be under $100,000. Simpler add-ons that only need to contend with a small amount of text can be ordered from companies like Drift for much less.
Whatever option a lender chooses, Sriram says it’s important that their bots engage clients in the early stage of their mortgage journeys.
“Making the early part of the customer experience good can pay enormous dividends down the road,” he says. “If I can find my answer quickly, that pushes me further along the purchase funnel and creates a more favorable income for me as a customer.”
