3 Reasons Why Medallia and Clarabridge aren’t enough in 2022
‘Selective hearing’ and rigid text analytics models cause Medallia and Clarabridge users to miss up to 98% of their customers’ issues.
Here’s a fascinating statistic for you: Spiral internal research shows that 70-98% of customer issues can be found across just three (3) channels: support emails, live chats and phone calls. Customers take the time to call or write about their problems with your product or service and their grievances fall on deaf ears. That's because companies like Medallia, Clarabridge, and their peer text analytics organizations aren’t able to systematically, nor thoroughly analyze these channels. While they’re popular Voice of the Customer tools, let’s take a look as to why ‘Medallia vs. Clarabridge’ is the wrong question to be asking when doing your research.
Ideally, you should be aware of 100% of the issues affecting your customers – that way you know what to prioritize and fix. Here are 3 reasons why Medallia and Clarabridge simply cannot help you realize that goal:
1. They’re looking in the wrong places
To start, they’re not analyzing the right channels. Companies like Medallia and Clarabridge have built large businesses on the back of surface-level insights from supplemental channels like NPS and relationship surveys – but…these channels only capture about 2-5% of all customer issues! 😳 I’ll say it again, louder for the people in the back:
NPS & ad hoc surveys only capture 2-5% of all customer issues!
As you can imagine, this leaves you with a very incomplete view of the world and blind to the other 95-98% of the specific issues your customers are trying to tell you about. And this is because 1. the total volume for NPS and ad hoc surveys is low due to spotty response rates, and 2. the data quality is low and bi-modal, due to selection bias among responders, and low-effort responses that are vague (the popular responses are: A. good NPS score + no comment, or B. bad NPS score + generally angry at the brand itself, with little to no description of the problem they’re having). It takes time to tell you about their problems, and most customers are unwilling to do so in the 21st century – and NPS emails and ad hoc surveys certainly isn’t the forum for it.
Figure 1: Average percentage of customer issues by feedback channel type
It seems logical that support emails and live chat/phone calls contain the lion’s share of customer issues since the reason customers are contacting you is to explain their problems! It’s literally a cheat code to capture and analyze this data.
“Today, most companies only scan <5% of their customer feedback. Unfortunately, they miss an absolute gold mine of specific issues, which are trapped in the remaining 95% of unanalyzed customer support data. Truly customer-obsessed teams in Product, Engineering, and Customer Service require these insights to calibrate their efforts to the customer.”
- Elena Zhizhimontova, co-founder & CEO of Spiral
You’ll be left disappointed if you try to assume that NPS and ad hoc surveys will give you a representative view of the issues happening in the field. To be fair, it’s hard to blame Medallia and Clarabridge for not being able to meaningfully analyze these channels because…
2. Their text analytics models have limited capabilities
Even if they wanted to venture into support emails or live chat and attempt speech analytics on phone calls, Medallia and Clarabridge’s text analytics models would break down when trying to analyze messy customer conversations. Because their pre-built, topic modeling, classifiers, and keyword clustering are too rigid and they listen for limited categories, creating lots of blind spots.
An example is a ‘financial services categorization model’ that includes pre-selected topics based on popular keywords like “loans,” “deposits,” or “branch location.” These may be words of interest to a bank, but the models would only be able to display general metrics like: number of mentions, general sentiment (positive or negative), and then perhaps direct you to begin manually reading through survey responses that include these words, which defeats the purpose of automation – no human is capable of reading & digesting conversations at the speed which feedback comes in.
These text analytics models were originally intended to carry out simple tasks, like assigning tags to news articles to categorize them, but they’re not going to be able to pull descriptive insights from support conversations. As far as NLP applications go, their ranges are very limited.
Their respective failure points include, (but are not limited to):
- Vulnerability to bad grammar
- Ignorance of semantics
- Inability to find new topics, or “unknown unknowns,”
- Uninterpretable results (eg. lists of random keywords with no context).
And if you want to find issues buried in customer feedback, these are baseline requirements you must solve for. And since Clarabridge and Medallia haven’t yet satisfied these requirements…
3. Their “insights” are vague
The Clarabridge vs. Medallia comparison really is a pick your poison scenario. All you’re going to get from either one is high-level categorization and perhaps some basic sentiment analysis, forcing you to do a bunch of manual legwork to discover anything usable. This, again, defeats the purpose of buying an automated tool.
Since the point of analyzing customer feedback is to identify issues and determine the improvements that will delight your customers, wouldn’t you ideally want a real-time list of specific issues? Unfortunately, basic “insight reports” like the below (source: Medallia.com; Qualtrics.com) aren’t actionable. You just wouldn’t be able to look at something like this and figure out how to satisfy an NPS detractor, how to influence first call resolution (FCR) or determine any sort of starting place when trying to conduct customer churn analysis.
Figure 2: Graphic from Medallia’s insight dashboard: recent topics + sentiment
Figure 3: Graphic from a Qualtrics XM insight dashboard: recent topics + sentiment
What is Spiral?
Specific customer issue detection using the latest machine learning (ML) research, and neural networks.
How specific are we talking?
Let me compare our level of detail against the former standard:
‘Status quo’ detail
Topic: checkout experience | sentiment: negative
Issue title: “Clearance items disappear from my cart; full-priced items are added correctly”
Impact: 13.5% of customers (68,322 reports) started experiencing this yesterday, first reported at 10:37am PST
- Happening only on Android 12.0
- Happening only on app version 7.3.8
- Only for customers completing ‘Guest Checkout’ mode
- 87% of customers who experienced this are NPS detractors, with an average score of 2 out of 10
As you can see from the above, we strive to go much deeper than traditional conversation analytics you’d get, even from large and prevalent companies like Qualtrics, LivePerson – which are only able to display coarse, quantitative metrics as seen below (source: LivePerson.com).
Figure 4: LivePerson’s Conversational AI analytics dashboard, which provides performance overview data, and coarse topic insights labeled as ‘Intents.’ Note in the top left quadrant - 61% of their conversations are unclassified (due to topic modeling limitations), which is problematic when trying to construct an accurate understanding of what your customers are experiencing in the field.
What’s the Spiral experience like?
Because we thoroughly analyze customer support channels, each time you log into Spiral, we’ll present you with the full list of current customer issues, starting with the biggest issue at the top, so you can decide what to work on and in what order.
You can also subscribe to alerts for emerging issues or to receive ongoing reports of any number of issues or categories.
If your customers talk about it, Spiral ensures you know about it.
What if I already subscribe to Medallia or Clarabridge?
No worries, many of our customers use Spiral in tandem with other subscriptions. We can integrate with Medallia or Clarabridge (and any other CRM, NPS, or survey provider like Qualtrics etc.) to import and analyze customer data for specific issues on a daily basis.
How can I try Spiral?
If you’re curious what Spiral’s turnkey insight solution can find in your customer feedback, let’s chat. Schedule a demo and before our first call, we’ll scan your company’s customer reviews and show you a personalized demo dashboard based on your public data.
We can get started with as little as one feedback channel, or process all channels in your omnichannel support program. We work with live chat, support emails, phone calls, NPS, DSAT, CSAT, social media, reviews, SMS, surveys – or any channel you consider to be feedback.
Our customers come from spaces such as: financial services / banking / credit unions, fintech, connected hardware and software, insurance, retail, B2C SaaS, and managed marketplaces – and we’d love to hear from you.
Thanks for reading!