Knowledge Manager Roundtable: How do you measure self-service activity and knowledge base usefulness?

11 Comentarios

  • Sherri Anderson

    Great article, lots of VERY useful information. 

  • Brandon, Rob

    Great info in here.  I'm interested in the overall success of a help centre and the measurement of reduced help tickets.  We can see reduced webforms submitted and relate that to orders but a customer could easily pick up the phone or start a chat.  How is this overcome in your areas?  We've considered additional reporting so if a customer clicks on live channel support they would not be included as a deflected contact.  Thanks, Rob

  • Zac


    I would focus on categorizing your inbound requests (including via phone and chat). Can you tag those interactions in some way? I would then correlate them to the content available in your Help Center- does the Help Center a) contain content that would deflect that type ticket, b) is it understandable to the user, and c) is it served it in such a way that your audience can discover it when they need to?

    I would then try to understand the user journey of the people who have called or chatted you. Chat clients often give you visibility into clicks on your site prior to a visitor initiating a chat, and you may be able to get some ad-hoc Q & A done by your agents during a phone support interaction to find out if the requester tried to self serve. Did they have trouble finding what they needed? Did they find an article but deem it unhelpful? What elements do they have in common, and how can you learn from those journeys to better position your help content?

    By doing this, you can focus on deflection by contact type. This helps you piecemeal your strategy and keep you from taking on too much at once, and also avoids the pitfall of looking for direct conversions from Help Center -> Tickets (a pitfall you described in your post).

    In general it's hard to measure contacts that are deflected away from a non-integrated channel (like phone and potentially chat) because it's hard to tell from your Web audience who was a potential contact in the first place. You can measure with certainty which contacts were deflected from the Web form by Help Center suggestions, and also which contacts were deflected by Answer Bot, because you know for sure they were in the process of contacting you for help. In general, when looking beyond that, I think it's helpful to look at the trends described above, and also correlate your inbound phone and chat contacts with Help Center engagement overall to look for an inverse relationship.

    As you notice reduced webforms submitted, you can also check if your top-line calls and chats are increasing - that'll give you an idea of the contacts are deflected or just driving to another channel.

    I hope this is helpful!

  • Sherri Anderson

    That's easier for us here because we don't really take phone calls and any voicemails that do come in are converted to tickets. If you record the number of incoming calls and incoming chats, which you must somehow.  Why not just take those numbers, add those together and use the self-service score like this?

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

    I wouldn't include people who call in or submit a chat as people who tried to self-serve though.

  • Anton Maslov

    Hi Rob, in our case, we automatically create tickets for chats and calls, what is kind of similar to your approach. Also "start chat" button is on Zendesk HC page, so basically in Google Analytics, we can see if a customer tried to search anything before starting the chat.

    To evaluate help center success we use "call deflection". Basically, it is even possible to measure success in saved money to show the result to stakeholders: you calculate "deflection" to understand how much tickets avoided and multiple to ticket cost. Some links with general algorithm description: - an official guide from KCS - check "CALCULATING CALL DEFLECTION" here. - a guide from KCS certified company. 


  • Andrei Kamarouski
    Community Moderator

    Hi Rob. I just want to add several points (thinking out loud) to the great experts' pieces of advice. 

    Old well-known call deflection idea (coming from the era of the assisted support when nearly 100% of support demand were satisfied via phone channel and full deflection was valuable) is a bit outdated in the world of the multichannel support and of the rising importance of holistic customer experience and engagement (deflection is a kind of 'serving'). Several important questions arise:

    • How does customer support journeys look like? What are most used multichannel support funnels? 
    • What is deflected for each given case - calls, chats, tickets, bot interactions etc? 
    • What is the experience behind deflection? Does it contribute to the overall support CX positively? 

    Collecting and combining data of WHAT is happening and WHY is this happening is challenging but very valuable to get the complete picture of support CX. 

  • Ryan S Campbell

    @Jennifer Rowe. How can I join the Knowledge Roundtable? I'm starting in a new role as Director Knowledge Management with a company who uses Zendesk and I would love to start connecting with my fellow Knowledge Management peers who use Zendesk.

  • Jennifer Rowe
    Zendesk Documentation Team

    Hi Ryan,

    Thanks for your interest! I'll reach out to you via email today.


  • Jessie Schutz
    Zendesk Team Member

    Thanks for sharing your insight on this, Andrei!

  • Kourtney Stuthard

    Hi everyone,

    This roundtable has been incredibly useful to us, there's so much great information! We have started measuring our own Help Center metrics. Has anyone here used Data Studio? We are in the process of building our own and want to know if anyone can share examples of how they configured theirs.


  • Andrei Kamarouski
    Community Moderator

    Hi Kourtney, 
    You probably mean Google Data Studio? Yes, I have used it several times for reporting on HC stats, collected via Google Analytics. You can have a lot of default metrics from GA + create specific metrics and events. Pls, refer to this article for more info. 
    P.S. Sorry if my reply is too late for you. 


Iniciar sesión para dejar un comentario.

Tecnología de Zendesk