Good Bot, Bad Bot
The impact of adding CS automation to your business
Adding AI to customer service sometimes seems like the pinnacle of the support experience.
A robot solving all the problems that our customers have. Sounds like a dream come true, right?
However, when looking at the details we find that there are hidden costs to customer satisfaction and thus to the rate of repeat purchases.
In this blog we’ll show some real-life examples of the good and bad of adding AI to customer service and how to navigate the AI-bot waters in order to get only the benefits and avoid the drawbacks.
Reducing SLAs is one of the best features of bots. What can be better than a cutting edge AI burning the midnight computer oil just to provide your customers with the fastest response only computers can? Let’s look at the good and the bad -
Here’s a chart of how SLAs were decimated as soon as one of brands we surveyed started using an AI-agent-email-bot for the CS operation -
They definitely pulled the long straw.
Another brand we spoke with however paid for their improved SLA with a huge dent in their CSAT as their bot was so eager to answer questions that it answered even those that they shouldn’t, eventually leading to longer handling times because a human agent would need to step in and resolve the original request as well as the confusion.
Overall we see that adding bots normally means that companies are paying with CSAT to improve SLA. We’ll discuss this more later in this post.
Reducing Operational load
Another key benefit that operations, marketing, and CX leaders want to tap into when adding bots is reducing the size and complexity of the CS operation -
- Bots work 24x7
- Handle surges easily throughout the day, week, season, and year
- Bots don’t require shift management and they’ll never be a no-show
- Bots have predictable productivity and efficiency outcomes
We spoke with companies that managed to save up to 20% of their CS workforce and are quite happy with the financial upsides and the reduced operational headaches.
On the flip side, we see examples of automatically generated answers that create additional work on the CS team. In one of the companies we’ve surveyed, a bot that automatically replies with instructions on how to return, got the customer’s intention wrong about 20% of the time. This led the CS team-lead to decide that every bot response needs to be double checked within 24hrs by a CS member creating even more work for them and increasing frustration with the customers.
Understanding the customer or misunderstanding the customer?
An advantage of bots is that they don’t forget, they don’t need to be reminded of company policies, and they are (relatively) consistent.
One can configure hundreds of different service flows and expect the bot to get them right every time.
Want to return a jacket? You’ve got 15 days to do so. Pants? 30 days. Shoes? Got to send images first.
An agent might confuse them, a bot won’t.
However, AI and bots in general often misunderstand the nuanced intentions customers have.
Here’s an example of an automated lengthy response that’s benign but silly -
Here’s another example where Priceline’s bot completely misunderstands what the tweet is about -
Courtesy of twitter.com (x.com)
Ultimately, what every eCommerce business wants is repeat customers, and one of the best ways to get that is by making sure customers are satisfied.
From the companies we’ve surveyed a clear image emerges in which bots reduce CSAT over agents or other solutions.
Where’s the silver lining?
The trifecta of low SLA, high CSAT, and low ticket cost is sought out by every company when they explore, integrate, or change their AI/bot solution. It’s hard to find, but it’s possible.
With the advancements of chatGPT and similar AI systems a new breed of AI agents is now being offered, that can achieve customer service goals at much higher rates, as well as to perform pre and post sale activities to help increase revenues.
These are definitely exciting times for customer service automation.
Overall CS Capacity Calculator
Monthly Inquiry Capacity: