Automation may be the shiny new object that gets lots of fawning attention from the analyst community, but the reality is that inside the enterprise – where customer experience is crucial to business success – automation still has a long way to go.
That is the view of Steve di Tirro, a respected vendor operations leader, with experience working at large, well-know technology brands that do a significant amount of work with third-party IT/BPO providers. “I don’t think a lot of companies are doing well with their automation strategy,” Di Tirro told us during a recent interview.
A Matter of Language
For starters, Di Tirro pointed to the flaws in the use of automation in language support. Getting the right balance between live support and automation in any kind of customer experience can be challenging. There is a great deal to think about in terms of the mechanics of grammar: diction, sentence structure, both spoken and written.
“When you think about the classic channels of phone, email, chat and asynchronous chat or SMS, each has their own style of interaction or flow,” he says Then, one must consider specific language differences. “In Spain there are two dialects of Spanish, and then as it has been propagated through Latin America, each country has added its own flavor to that. Being able to try to centralize that type of language support requires cultural understanding of the origin of the caller,” he says.
He cites the example of an agent in Mexico taking a contact, but the person calling is from Buenos Aires. “The way I speak to my friends in Mexico is very different from the sorts of words I use or the way I speak to someone from Argentina,” he explains.
It’s the same for an English support agent in Kingston, Jamaica serving a customer in the US or the UK. The agent would need to know and understand the kinds of things that would allow them to personalize the interaction. “Those things are often overlooked. People tend to think that English is English, Spanish is Spanish, French is French,” he adds.
Di Tirro is seeing change at big tech companies, though. “They have done a lot to understand how to operate comfortably for their customers or users in their marketplaces. So there is an acknowledgement that we are a globally recognized brand, but here we’re your version of that brand,” he says.
“I have seen that done in a digital marketplace, in support, in logistics – in the way that it’s very different how you deliver a package in Mumbai or Chennai or Delhi, than you do in New York or London.”
Where the Bot Fits
Di Tirro believes that automation can play a critical role in this, particularly if you have known issues that have known resolutions. He doesn’t see value in using a human being in the initial screening of information, either, and suggests that it is better to have a bot do that part and then do a soft hand off to a human to handle from there. “If you have known issue types, that 90% of the time are resolved the same way, why are you letting a human do that?”
But di Tirro acknowledges that this automation and use of AI can be, in his words, “creepy”. “With GDPR and recent laws in the US, in California, your digital identity is recognized as an asset. But the majority of consumers are still not aware of what people can find out about you and what can be used in a malicious way. The protection has preceded the general awareness of the risk.”
Companies wanting to use data in automated ways need to make their customers aware of it and ensure that it is clear exactly how the information will be used. “It needs to be about automation that enhances your customer experience versus the creepy side,” he says.
Di Tirro recommends two main must-dos for the use of automation:
- Full disclosure: customers need to know they are not talking to a human being, otherwise they will feel tricked. There are many ways that customers can be fooled by a bot; even the fact that we name bots adds to this. “The bot’s name is Sophie. Why are we naming a piece of software?” di Tirro asks, adding wryly that it seems to have worked for Siri and Alexa.
- Learning: companies using automation need to look at the small interactions, collect data, and then make decisions based on those interactions. “Any interaction with any customer is a form of friction and you can choose to make that friction worse or less,” he says. Automation should create less friction, not more.
Di Tirro believes that the BPO industry still has work to do to get automation right for the B2B customer. He says that the types of workers in the classic BPO space are evolving – and it is much needed. “If you take the easy survey yeses or easy tens aways, then you require skilled agents to take what’s left. There needs to be a shift in hiring profiles, a shift in the type of employee that the BPO companies need to get for a company that’s there,” he says. But, by his estimation, only 10% of the industry is close to that or working towards that.
He adds that this approach to artificial intelligence is spawning new types of work for humans that is very low risk. This is the human judgement work that is seeding AI and machine learning, feeding image recognition into software that ingests it and “learns: from it to become more and more intelligent through the validation of information that is fed into the system.
For di Tirro, automation is not about replacing jobs but about creating new opportunities, a kind of augmented intelligence that brings together the best of both human being and AI.