Why It's Too Early to Go All In on Virtual Assistants | Street Fight

Why It’s Too Early to Go All In on Virtual Assistants

Why It’s Too Early to Go All In on Virtual Assistants

As the so-called customer journey takes new twists and turns, tech companies and agencies should help local businesses and brands differentiate themselves via user experience. Catering to virtual assistants might seem to be the path towards that goal. But it’s probably too early to make huge bets on these technologies, and there are more practical ways to help out.

A new study by local digital marketing agency DAC Group aimed at mid-sized banks and credit unions accelerated my thinking on this subject. The paper does a good job of tying together existing research — although some of it is a little old — with interviews to identify core themes for delivering a seamless and consistent user experience. It shows there is a high correlation between mobile banking and overall customer satisfaction, and illustrates a demand for personalized services.

Meanwhile, an exercise senior hyperlocal company execs did at Street Fight’s recent Rocky Mountain Retreat indicated that their favorite emerging-tech bet was AI in the form of virtual assistants. That technology outpaced voice search and version 2.0 of on-demand. Perhaps virtual assistants will be the next big thing in local user experience?

Maybe. But let me make the case that this exciting new technology is more than a little raw. I’d recommend cautious experimentation, and hedging bets across platforms, for the the next 18 to 24 months at least.

The schematic above maps out where some of these platforms sit in terms of general-purpose assistants versus specific task-oriented tools, and home-based versus mobile activities. It will be easier to deliver near-term solutions that are focused on single-function or simple tasks, even if much of the industry hype is centered elsewhere.

In a smart blog post last week, Andreessen Horowitz’s Ben Evans argues convincingly that, even though voice recognition and natural language processing have made huge advances in recent years, we’re a long, long way from the Scarlett Johansson-bot in the movie “Her.” Voice input works pretty well now, but there is no magic scaling solution to hand-crafting all the answers needed to deliver every shopping or ordering experience. “Narrow domains” like music and maps work better.

Similarly, a Gartner survey indicates that the connected home isn’t catching fire. Outside of gadget nuts and early adopters, mainstream consumers aren’t excited by voice-activated home lighting or heating. While those functions are simple to script, most users don’t find them particularly necessary or fun. Amazon looks wise to emphasize entertainment and information-lite utility like news headlines and weather, which it hopes will train consumers to place replenishment orders and make simple on-demand orders.

Even Facebook, which has the scale and user engagement necessary to drive interaction automation, recently refocused its chatbot efforts towards filling out forms. Marketers have had mixed results at best in playing around with automated conversations. I am still skeptical of chatbots – partly because of messenger app fragmentation – but Q&A experiences originating from a map app seem like a natural fit. Setting expectations and teaching users how to drive these interactive experiences is critical for success. Siri over-promised and under-delivered, probably souring a lot of consumers on voice-based assistants.

Which brings us back to that banking study. Although they don’t always have the capability, most local businesses recognize that they must offer mobile access to their goods and services. Sure, they still struggle with blocking and tackling like managing store information across directories and discovery options. They need help delivering a consistent experience across devices, and order forms will feel more familiar than trying to replicate a sales associate or teller via virtual assistant.

Fairly lightweight personalization can help simplify the task or transaction. Understanding previous customer behavior and characteristics makes it easier to present a short list of options, if not the single perfect answer Google’s trying to pull off. Cameras are catching on as an input device (depositing checks, comparing prices) faster than voice, though I’m not sure whether Snapchat Spectacles fit yet.

As for placing bets on particular future platforms, take a look at the figure below, that illustrates how consumer technologies gain adoption over time.

Many technologies that are primarily substitutions – e.g., satellite TV, new storage formats, etc. – display a steady, nearly linear adoption path. Others depend on content availability: color TV, MP3 players. And several display a critical mass effect. My favorite example is the VCR. Its initial value to consumers was time-shifting broadcast TV. That was enough to get to roughly 15 to 20 percent of U.S. households. At that point, Blockbuster was born and a whole new business led to that hockey stick on the chart.

This critical mass effect doesn’t always happen: DVRs never found a new role beyond time shifting, and growth slowed and plateau’d. But keep an eye on potential platforms that reach a quintile of users or households. Amazon Echo is about halfway there.

David Card is Street Fight’s director of research.