What Self-Service Rate Can You Expect from Artificial Intelligence?

We often get asked by prospects or even by our customers what the standard or benchmark for self-service is. This is a very legitimate question as organizations need to justify expenses and ensure all investments will bring a positive return for the business.

Even though there is no “real” benchmark for automated self-service solutions, we’re going to try to give you some indications of the kind of results you can realistically expect from such solutions.

But first, what is customer self-service?

Customer self-service is a solution or a set of solutions that enables web users to access information or even perform some simple tasks autonomously, without requiring the assistance of a customer service representative.

So what are the kind of queries or tasks that can be handled or performed by customer self-service?

Tracking a package, requesting a quote, or paying a bill online without getting in touch with a person at the company for help, are all examples of customer self-service tasks that we perform on a regular basis.

In terms of customer inquiries, not all queries can be handled by self-service since some complex issues will still require human intervention. However, customer self-service solutions are very efficient at solving tier-1 repetitive queries. These are the perfect type of requests as they represent approximately 80% of the questions received by customer service agents and take up a large amount of their time, so they can easily be automated.

Which metrics to measure self-service KPIs?

Each organization has to figure out the best metrics to use when trying to measure the performance of self-service tools. However, there are a few key metrics that a business should probably be monitoring on a regular basis.

Call deflection rate

“Call deflection” refers to routing a customer’s inquiry to an alternate service channel, such as chatbots, FAQs, community forums, or knowledge center databases. The goal of call deflection is both to ensure that customers receive the answers they are seeking in the most efficient manner, and to reduce the number of calls routed to human agents. Even though this metric refers to “calls”, it also includes any contacts requiring the attention of a human agent, such as live chat and emails.

Measuring call deflection rate can be complicated as we’re trying to measure what didn’t happen! According to DB Kay & Associates, one method is to estimate both the percentage of users who are successful with self-service and the percentage of users who would have contacted a live agent. The difference between those two percentages represents the deflection rate.

Customer satisfaction

Implementing customer self-service channels is an exciting project for any organization as it aims at improving the customer experience. But if customers are not satisfied with the tools you put at their disposal, if they find them too difficult to use or inefficient, then the self-service channel cannot be considered a success. Customer satisfaction must be tracked for each self-service channel via surveys, direct feedback, and Net Promoter Score (NPS) in order to clearly understand which channels are most successful as well as which ones need improvement.

Self-service success rate

An easy way to determine the success of self-service can be to track how many customer inquiries are handled by self-service channels without being escalated to a human agent. This can be, for example, the percentage of times a “how to order” FAQ leads to an order rather than a customer-initiated chat session or the percentage of times a knowledge base search leads to a useful article, indicated by the user rating the article as “useful” or indicated that “this solved my problem.”

When deploying one of Inbenta’s Customer Interaction Management Platform modules, that self-service rate is automatically tracked, calculated, and available for our customers in the back-end, along with plenty of other useful metrics.

How to calculate self-service ratio

Let’s start by defining the percentage of issues that can be solved by customers themselves using self-service channels. As stated before, not all inquiries can be handled by self-service tools and the more complex ones will require human intervention. We’ve observed over the years that this percentage largely depends on the use case, the organization, and even the application, but usually, 50% of queries can be self-resolved by users.

Out of this 50%, we need to quantify how many are redundant or repetitive. As mentioned, approximately 80% of queries received by customer service agents fall into that category. These are the ones that are suitable for self-service.

The maximum aspiration for self-service would be the product of these two percentages, ie 0.5 x 0.8 = 0.4, so 40% would be the maximum self-service rate that can be expected.

Finally, you need to take into account the efficiency of the Artificial Intelligence powering your tool. With the right AI, the right content, and a powerful Lexicon such as Inbenta’s Lexicon, your self-service solution could reach as much as an 80% answer rate for these repetitive queries.

As a consequence, 32% (0.4 x 0.8 = 0.32) is a good target for self-service ratio.

Of course, these are indications only, and results can greatly vary depending on the use case, the industry, or the type of technology powering your self-service solution(s), but that gives you a good basis of comparison.

If you’d like to go further and calculate the ROI of a Chatbot or Knowledge Management system, our article will be of interest to you.

Or, schedule a demo with one of our solutions specialists for a free consultation and impact assessment. 

Check out our similar articles