Justine Brown | October 13, 2016

Trust, but verify: Why CIOs need a role in bot rollouts

Given the hundreds of billions of dollars invested in technology over the years, one might expect a dramatic improvement in worker productivity. But arguably, that hasn’t happened yet. Employees at many companies still spend a significant amount of time on low-level tasks and struggle to connect disparate data to provide a big picture view.

“We all know the frustration of calling a help line and being asked for the same information over and over,” said Neil Kinson, chief of staff at Redwood Software. “Even in today's age people are still re-keying and transferring data manually between multiple systems.”

Manual processes are not only labor-intensive; they also hold potential for introducing errors.

Recently, one organization Redwood worked with accidently manually overrode about 30% of their data, Kinson said. "Those manual activities and processes create the kind of frustrations and issues that manifest themselves in terms of data quality."

But the rise of bots has the potential to change all that.

Already, bots and corresponding robot process automation (RPA) are automating many front office functions and helping businesses connect the dots, particularly when it comes to customer data. But such applications often don’t involve the IT department.

"A lot of the RPA that has been deployed so far has been deployed in spite of the CIO," said Kinson. "In fact, a lot of the RPA vendors position their technology around 'you don't need to engage IT because if you do that it will be expensive, it will be time consuming, it probably won't deliver. The white blood cells in IT will all conspire to kill the initiative.'"

But there is significant cause for concern around this approach, according to Kinson, because applying things like RPA, AI or analytics to faulty data can’t help a company improve its operations.

The CIO’s role in RPA

The CIO can not only ensure better data quality — because bots are only as good as the data they are fed — but he or she can also help rework faulty processes from the beginning so companies can use bots more strategically.

In doing so, the CIO can play a significant role in driving process improvement. Instead of just running sub-optimal processes faster, the organization improves the processes themselves and ultimately the quality of the customer outcome.

"CIOs can look at how to drive process improvement, how to improve the end-to-end process [and] the business outcomes that those processes are aiming to deliver," said Kinson. "That’s kind of what we're seeing in the marketplace. In my view, you cannot get a cohesive and consistent process that is scalable moving forward. You can maybe put a Band-Aid on today's problem but you're not going to fundamentally improve your process and data quality moving forward.”

CIOs should also be involved in how and where RPA is deployed, said Kinson, so they know how to fix something if it does break.

"You won't know as you deploy something what's going to break in those underlying processes," he said. "If you’re taking a resource away without considering that, then you're actually bringing significant risk into your business. The CIO needs to be focused on what that business outcome is, and they certainly shouldn't be excluded from these [RPA] initiatives. I think that's dangerous and inappropriate."

Performance and security caveats

Of course bots also have to work correctly to be effective, otherwise the customer experience gets worse, not better. CIOs and the IT team can play a role there, too.

"As we've seen with interactive voice recognition systems, [not all bots are] yet perfect," said Trevor Hawthorn, CTO of Wombat Security. "When 'intelligent' automation goes slightly wrong, it makes for a bad customer experience."

Security is another important aspect of RPA CIOs can ensure is considered.

"From a security standpoint, my hacker mindset already wonders what type of input the chatbot is able to handle," said Hawthorn. "Can I ask a question that causes it to throw an error? What if I paste in content it doesn't expect? Can I send it a URL that will trigger a back-end system to fetch that URL? What if I send it a malicious URL? ... Beyond functional security flaws, if the chatbot uses every customer interaction to learn or train itself, can I automate interactions that will pollute or mistrain the bot?"

In the same way, completely removing humans entirely from the equation could also be a mistake, at least from a security perspective.

"There are a lot of Western Union operators catching wire fraudsters when the conversation feels off," said Hawthorn. "Even over text, misspellings, the cadence of the conversation, different spellings or odd messages might make a human suspicious."

Bots don’t have that same level of automatic skepticism.

Enterprise potential

Despite concerns, bots and RPA still hold a lot of potential for the enterprise. For example, they could allow workers to focus on more interesting and stimulating activities.

"We should liberate people from the drudgery of some of these manual activities that no one really wants to be doing, and free them up to where they can add value to the business," said Kinson.

And despite reports that RPA could take away jobs, Kinson says it’s more likely they will help solve labor shortages.

"Most of the western economies are going to suffer from significant labor shortages over the next several years," he said. "As companies approach full levels of employment, bots may help shift the nature of jobs so that the less interesting tasks are handled by automation and the 'human' jobs are more interesting and strategic."