Here Is How and Why A Virtual Assistant Will Kill The BPO Industry...A Real Life Example

A lot has been said and written as to the impact Artificial Intelligence (AI) and the eventual integration with Natural Language Processing (NLP) will have on the BPO industry.

If some of the more aggressive forecasts are to be believed up to 70 % of the more clerical support based BPO activities are on the chopping block. Further integration into work processes should extend the reach of AI powered applications all the way into the more complicated tasks currently being outsourced.

Of course...all of this happens in the abstract and it remains to be seen how fast (or if) the new era will approach.

In my opinion it is already here...perhaps not visible to most people yet...but the beginning has been made.

I would like to tell you about a real life example which I was involved in...and which has led to a solution which is liked by the management and the customers.

About 4 months ago we were approached by a company which distributes a SaaS product which is being used by a lot of smaller to midsize companies. The product has rather universal appeal, very limited customization and is a low revenue product with high margins.

Now...the low revenue characteristic of the product is the reason for the rapid adoption...but it also is what has been eating into the profit margins of the company. Simply stated...the product support costs were rather  high vis a vis the revenue per customer.

Continued growth of the company has been forcing the company to scale up support staff to deal with inquiries as to the best use of the product. Despite the company limiting support to chat and email for the lower priced plans the support costs were chewing up margins.

Management decided that rather then hiring yet another customer support agent they were going to develop an AI based solution. They consulted with us and we were tasked with he development of a Virtual Agent for chat and email support to handle. to the extent possible, their level 1 support.

One advantage the company has was that the existing support history was rather extensive. They were tens of thousands of emails and chat logs with the Q's  and A's needed to develop the Virtual Agent.

As is undoubtedly the case for most organizations...the same questions do repeat themselves. A question might be asked in a variety of different ways...but the answer is often times the exact same one.

With the help of the company we reduced the more than 10,000 support cases to about 200 actual support scenarios.

It had been decided that we were going to use Watson as the backbone of our support agent. We developed a procedure and the required technical tools and went to work.

It took about 2 months to get a Minimum Viable Product (MVP) out of the door. Another months was spent on refining and improving the MVP at a steady clip.

The interesting aspect of the exercise is the result for the company. A couple of things materialized after the deployment:

The company experienced very little negative feedback. Even in a situation where the Virtual Agent was not able to help the customer a built in escalation feature to a live operator resulted in high customer satisfaction ratings.

Key Performance Indicators (KPI's) the company had been using to measure customer satisfaction for their support rose across the board. Due to the immediate availability of the Virtual Agent (no need to wait for a live operator) KPI's rose rather dramatically in the areas of expediency and quickness of response.

The company was able to migrate level 1 support staff to level 2 support and avoid the hiring of additional personnel

The costs associated with running the agent (IBM charges for each API query) resulted in net reduction of approximately 75% in total support costs.

Perhaps most beneficial for the customers...the company went to a 24/7 level 1 support after about 6 weeks of usage for the Virtual Agent.

The Virtual Agent has been getting better. It goes without saying...but the "intelligence" aspect of the agent results in better answers over time.

All in all the company views the implementation of a Virtual Agents as a huge success. Customers are satisfied and the company estimates that it will reduce hiring, given current growth rates, by about 5 bodies.

Of course..the real life example cited above cannot necessarily be the blueprint for every company. Not every support scenario can be successfully resolved through the use of a VA. Never the less though...AI is here to stay and slowly but surely it will work its way into real life support situations.

The consequences will be felt...for some providers of BPO companies  rather soon...for other companies
it might be years before they will feel the impact.

There is little doubt though that once corporations go the AI VA route the consequences for BPO companies can be dramatic.