Early this year, Bank of America introduced a customer-facing AI chatbot, nicknamed Erica, which generated 1 million registered users in three months – clear evidence that faith in chatbots is building.
As these examples of chatbot popularity emerge, Nearshore Americas conducted a survey to find out where Nearshore buyers and vendors stand on both adoption and development of chatbots.
The survey was open between May and June 2018, and generated responses from 179 vendors and 142 clients, 91% at the executive or senior management level, and all of which currently using or selling chatbots.
Customer-Centric Focus for Chatbot Clients
These automated helpers are infiltrating myriad industries to serve multiple functions, and are an especially important technology for improving customer experience.
We asked chatbot buyers what their number one reason was for investing in chatbots, as well as the results they found after implementing them. Perhaps unsurprisingly, improved customer service was at the top of both lists at 64%.
This customer-centric focus was followed by improved efficiency, with 14% of respondents citing it as their number-one adoption driver and 36% finding it as a result of adoption.
Only 7% wanted to adopt chatbots to cut the size of their workforce, with 14% of them actually succeeding after adoption. But are chatbots really well enough developed to offer that incentive?
Tae Moon, Senior Manager of Digital Banking at TD Bank, told us in a recent interview that “chatbots are essentially AI infants”. He mentioned that they need training and they are unlikely to be optimized for specific processes. This means they require patience, commitment, and a significant time investment if they can truly be considered as a replacement for human agents.
In the study, we asked clients what their number one concern with chatbots was, before and after adoption. Pre-adoption, negative customer feedback is high on the list at 29%, and still a concern for 25% after adoption, which is an interesting contrast to their number one attraction to chatbots – improved customer service.
If the goal is to improve customer service, but negative feedback is a concern, then it is worth asking if chatbots are really right for the company. Most clients in the study reported to have invested somewhere between US$50,000 and US$100,000 in the technology, so chatbots are an investment that companies need to be sure about.
“While it’s possible to build a chatbot in one day, even without an IT background, there are basic Q&A chatbots and next generation AI-powered applications, which takes much more time, effort, and knowledge to implement. An intelligent chatbot must be able to combine information from whatever multimedia a customer has been using, responding to whatever they want to know, or, indeed, providing information on what they didn’t realize they wanted to know about. That is the intelligence component of AI chatbot solutions.”
After chatbot adoption, 32% of clients expressed that they had security and data protection concerns. Chatbots are being used to share salary information, health insurance details, and similar data, so clients are right to be concerned. Security is always a hot topic with developing technologies, so clients and vendors must be clear with each other on what is possible to protect sensitive data being sent through chatbots.
Reasons Not to Adopt Chatbots
In total, 79 buy-side survey respondents said that they are not currently using chatbots, so we asked them why they hadn’t adopted them yet. Most of them said that the “technology is not well-enough developed”, showing that potential clients need to be educated about what is possible with chatbots.
There is also the end-user factor, as noted by Ian Jacobs, Principal Analyst at Forrester Research, during his chatbot panel at Nexus 2018.
“We surveyed around 100,000 consumers and end users about chatbots, and 80% of them expected to be using them by 2020,” he said. “Of that 80%, 54% of respondents said they expect the experience to be negative, so there is quite a hill to overcome in convincing them otherwise.”
Vendor-Client Relationship Insights
Banking and Finance was found to be the main industry in which vendors’ clients operate (61%), lending weight to the fact that data and security is a top concern for respondents.
BPO and customer service is very low (6%), perhaps correlating with the fact that more higher-value work is spreading throughout the industry, such as technical support, knowledge-process outsourcing, and telemedicine.
Vendors also expressed what they thought are the main concerns for chatbot clients. While negative customer feedback is up top (40%), the concern that most clients actually have, security and data protection, is right at the bottom (6%).
This shows misalignment between vendors and their clients, which has perhaps come about through the overarching focus on what chatbots can do for customer experience, instead of what it takes to implement them.
Vendor Chatbot Teams & Revenues
Over half (56%) of vendors are developing or deploying chatbots with pretty small teams of less than 10 people. This shows that the technology can be developed with relatively little investment from the vendor side, providing the talent can be sourced.
On the salary side, almost three quarters (70%) of vendor respondents are paying US$20,000 or above to chatbot developers, with a third (33%) paying an attractive US$50,000 or above. Again, this shows that chatbot teams are relatively cheap to form. As an example, a small team of 4 people could set you back around 100,000 a year, depending on the location of the teams.
In terms of revenues, over one third (39%) of vendor respondents are pulling in more than US$200,000 a year from the sale of chatbots, while a quarter (26%) are only making US$5,000 to US$10,000.
Depending on company resources and its willingness to invest in a team, this shows that chatbots are still kind of a gamble to get into. Take the example from before, paying for a team of 4 developers for US$100,000 a year. If you doubled that in a year, the benefits are clear, but only making US$10,000 of that back would be a poor result.
Why Are Vendors Not Developing Chatbots?
The study was also open to vendors who were not yet selling chatbots, with 61 responding. When asked why they hadn’t jumped on the chatbot train, most of them said there was “not enough client interest”.
Looking back at previous findings in the study, this is because non-adopting clients “don’t believe the technology is well enough developed”. Perhaps vendors should get together with clients to explain what is possible, and actively generate more client interest.
Overall, the popularity of chatbots in the Nearshore region is clearly strong, but there is plenty of room for vendors to enter the market, if they can show clients that the technology can be trusted.
With huge commercial rollouts like Bank of America’s Erica, that trust should be much easier to achieve.