We are looking to track Customer Sentiment on a more regular, ongoing basis and use this as part of our health score.
Currently: CSMs will choose red/amber/green manually upon gauging customers' likelihood to renew on each interaction (e.g. EBR, Escalation etc). The score will go stale after 30 days as CSMs should be interacting with each customer at least once monthly.
Ideally: Every time a customer interacts with support, we can also track the sentiment of the call. This would aggregate and we could show rolling averages which could be used to update the health score and create risk/expansion CTAs.
We send out a CSAT after email/chat/phone support cases but this has a very low response rate so isn't particularly reliable. Is there a more effective way of tracking customer sentiment and do you ask the customer to record this or the support agent?
Already have an account? Login
Login to the community
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.
We give it a low weighting. It's just another insight to help round out the health picture. It measures overall ticket activity over the previous six months, runs once a week.
0-2 tickets in last six months: Red
36+ tickets: Orange
19-35 tickets: Yellow
3-11 tickets: Lime
12-18 tickets: Green
We intend to pull data from the tickets themselves in a future version. We want to know if a ticket is an issue, usage question, or bug. That will give us more insight into the nature of the interaction.
What support tool does your support team use? One thought would be to add a field (if the tool supports custom fields) in the support ticket interface to log sentiment, and then upload on a daily basis the support cases from the past 24 hours to Gainsight, including that custom field that contains the sentiment. Then you could use a rule to set the sentiment score based on the most recent sentiment record (either by your CSM or support team).
That sounds like a great system. We also have a similar Ticket to Tenancy ratio (our main unit of use on our platform is tenancies and, as our customers range in size, we find a ratio better than an absolute value) that feeds into the support measure of our scorecard.
Likewise, we assume that a high ticket to tenancy ratio is unhealthy as a customer is requiring a lot of support. However, we haven't yet considered that a customer not filing *any* support tickets is also not healthy - I will have to look to incorporate this.
Similar to you, I am now looking to pull more data from the tickets themselves.
Our support team uses Desk.com (at least for now, given its retirement in 2020). We currently have a number of custom fields setup and the support cases are automatically synced with our Salesforce.
I was thinking the same re asking support agents to log sentiment on every case. I was wondering if any other user had implemented a similar system and, if so, how they found:
1) Accuracy of sentiment
2) Reaction of support agents to the extra field to complete
I see two different methodologies applied to sentiment after a case. 1) Add a field to the case object for sentiment and make it mandatory to fill it out before closing a case. Then fetch that data and use it to set your support sentiment score. This is low effort for customers but requires you to train support agents to ask customers how they feel after the support interaction (or interpret the customers feeling themselves).
2) Survey the customer after the case is completed. This gives you a CSAT score you can use to update the Healthscore in Gainsight. It's a model that scales well but you need to implement rules to keep from creating survey fatigue in your customers.
Which if you required this, would work. You could then average them out by account and apply a Score.
A side note on having the customer give you the information:
The one thing I like about CSAT or a customer survey is that the customer will reveal trends in support engineer effectiveness. If you have a support engineer that is urking your customers it will show in those scores.