Ticket deflection: the currency of self-service

Ticket deflection: the currency of self-service

May 22, 2017
Ticket deflection: the currency of self-service

Forrester Research predicts that self-service will be the #1 customer service trend in 2017 because of “customers’ rapidly growing preference for DIY forms of customer service”. Why is this becoming their favorite form of customer service? Because it’s often the fastest and lowest effort way to resolve their problems—there’s no need to contact an agent and everywhere 24/7 access. That everywhere anytime access is probably the biggest factor—we’re now a mobile, multi-channel, and multi-screen world. This is great news for companies, because increasing self-service leads to improved ticket deflection, or customers choosing to help themselves rather than reach out for support.

You may have already made a large investment in creating knowledge base content that is easily available to your customers on the web and through mobile devices. While you do your best to lead customers to self-help resources, you still often need your customers to make the effort to discover and use those resources.You can be doing better, and now you can with new tech and new tools that are pushing self-service into that #1 slot, vastly improving your ticket deflection ratio.

Using artificial intelligence and machine learning, you can now automate many of your high-frequency, low-touch customer interactions and bypass the need (and the effort required) for customers to discover and use that content. This frees up agents to assist customers when they need help with more complex support issues. Self-service satisfies customers and it’s also a big cost saver.

The problem with self-service, which perhaps explains why some companies have been slow to adopt it, is that it’s sort of an indirect form of customer service. It’s been difficult to actually prove that having a Help Center loaded with great self-service content is preventing your customers from requesting support, generating tickets, and needing to speak with agents. We know it’s happening, we’ve got some metrics to indicate its effectiveness, but it’s been difficult to produce the data to show its direct effect on the ticket queue and on customer satisfaction. That’s now changing.

Succeeding, but unconvincingly
For me, self-service support has always been the most important customer service trend. I was hired at Zendesk in early 2011 to create the knowledge base and help build the self-service channel for Zendesk’s customers. Over the course of several years, my team of writers and I cranked out hundreds of articles and guides and made steady progress toward our first big self-imposed success milestone: one million views per month.

We successfully hit our target about 2 years after we launched. When we did, we got a congrats from our CEO and a coffee mug imprinted with the hashtag #OMGMYFORUMWASVIEWEDLIKE1000000TIMES from the VP of our org. Total monthly views certainly wasn’t the only performance metric we tracked, but it was the big one and it was gratifying when we made it. But was it really enough to prove the effectiveness of the self-service channel? Not for me and probably not for you either.

How we measure self-service
The self-service metrics we track give us better insight into the content we need to create, the quality of our content, and our readers’ engagement with it—and they are invaluable from that perspective. However, they don’t really help us show the direct correlation between the use of the Help Center and the volume of ticket deflection. Here are the metrics we’ve been using:

Views and engagement
These are the typical metrics you track for help center performance (or for any web site): the number of views, unique users, and measures of engagement such as average session duration, bounce rate, and so on. The usual Google Analytics stuff that helps determine if customers are finding and using the content and if they find the content useful, and from that perspective these are all extremely valuable metrics. You can read more about these metrics in the four-part article series starting with Google Analytics and Help Center Part 1: Asking the right questions.

Community activity and engagement
Another important measure of success for us was the size and vibrancy of our user community. We wanted our help center to be the go-to place for customers to not only find the information we provide for them, but also where they go to engage with other customers, share their expertise, and learn from each other.

Google Analytics can be used to measure some of the activity of your community, but this is where more direct links into the support workflow are really useful. In Zendesk Support, help center reporting is segmented into Knowledge Base and Community. For each, you can track the number of posts, views, up-votes on posts, subscriptions, and comments. The targets you set for each is up to you, but needless to say you want lots of each and to track that activity over time.

Search
In the Zendesk Support reporting tab, you’ll also find data on user searches in the help center. The report includes the number of searches with no results (no articles that contained the search keywords) and searches with no clicks into articles that do exist. The first can help you determine the articles you need to create and the latter can help you troubleshoot the usability of your content (no clicks might mean that your article titles aren’t descriptive enough or don’t use the words your customers are using). You can also see the number of tickets created after a search. Finally, some insight into how self-service is affecting the ticket queue—in this case however it’s negatively because tickets were created, not deflected.

Self-service score
Whereas the metrics above give you insight into the performance and quality of your self-service content, the Self-Service Score is an attempt to measure the impact that your Help Center is having as a support channel, how it’s helping customers solve their problems and preventing them from opening support requests that then need to be handled by agents.

You determine your Self-Service Score using this formula:

Self-service score = Total users of your help center(s) / Total users in tickets

This gives you a ratio such as 4:1, meaning that for every four customers who attempt to solve their own issues using self-service, one customer submits a support request. (The Self-Service Score is also discussed in the article referred to above).

The self-service score is valuable because it allows you to create ticket deflection benchmarks, so you can compare the ticket deflection ratio from one month to the next.

At one point at Zendesk we reached an almost 40:1 ratio, which looked great on my reports to management and may have helped me get a promotion, but it still wasn’t enough to satisfy my desire to show the real impact our self-service channel had in preventing tickets from being created. There’s was a lot of benefit of the doubt going on.

So, how do we get closer to real data? That new tech, of course.

Adding artificial intelligence to self-service: people and bots power the next phase of self-service

The future of self-service is a happy alliance of people and new tech working together as DIY enablers. We’re now not only able to use artificial intelligence and automation to serve up our content to our customers, it’s now possible to directly link its use to support issue resolution. This is the data we’ve always yearned for!

We’re able to do this because of new self-service tech such as Answer Bot, which is included as part of Zendesk Guide. Let’s take a quick look at how it works.

Using deep learning and natural language processing (NLP), Answer Bot scans the text in the customer’s email and then replies to the customer suggesting help center articles that may help them solve their issues themselves.

The customer’s email request of course generated a ticket, so it needs to be solved. The Answer Bot automated reply gives the customer both the information that should help them resolve their issue and a way to then close the ticket themselves–before they’re contacted by an agent. If they don’t close the ticket themselves, they’ll be contacted by an agent who will follow up and close the ticket.

The most exciting part of this, for me, is that you can then report on these self-service ticket resolutions.

With Answer Bot, self-service is now much more integrated as a support channel. We can finally show its direct impact on the ticket queue and on customer satisfaction and present the business with the data we’ve needed to prove its effectiveness. Soon we’ll be able to do this across many of our support channels, which will give us an even fuller multi-channel view into the impact of self-service.

Customers will of course continue to use your self-service content in ways that are disconnected and untrackable as ticket deflection data, but that’s okay. It’s out there helping customers solve their problems, even if you can’t always quantify it.

Learn how to provide a great self-service experience, read 6 tips for building a thriving help center

Anton de Young is a published writer and photographer. As a long-time Zendesk employee, he built the Zendesk customer education and training teams, and then as a marketing director launched the Zendesk customer service leadership program and event series, which he then helped to expand into the Relate website and event series. Now a freelancer, Anton is busy exploring the world from his new home in Lisbon, Portugal. Find him on Twitter: @antondeyoung.

We know. It's a lot to take in.

Sign up for our newsletter and read at your own pace