You can have 90% customer satisfaction scores on individual interactions but have an overall low NPS. Why? They loved the effort you made in supporting them but not if it took 5 support requests. Customer experience is an art and a science and the art bit is done by adding vision to the voice of the customer. Below we'll cover key problems with data analysis and then show you how to add vision.
What's VOC and how do companies measure it?
Voice of the customer (VOC) is the "process of capturing customer's expectations, preferences and aversions. Specifically, the Voice of the Customer is a market research technique that produces a detailed set of customer wants and needs" via Wikipedia.
Here's how Hubspot describe it "This process captures everything that customers are saying about a business, product, or service and packages those ideas into an overall perspective of the brand. Companies use VoC to visualize the gap between customer expectations and their actual experience with the business."
How do you go about doing this?
Customer interviews, surveys, NPS, live chat, customer reviews, feedback forms, social media and more.
The underlying problem with data is people post-rationalise
This article contains a powerful list of key points on the dangers of surveys including:
- "When people answer direct questions they attempt to rationalise their decisions and behaviour"
- "We are prone to making sub-optimal and often irrational decisions. This is contrary to how we like to perceive ourselves and naturally we don’t articulate this when answering survey questions"
- "Relative importance of an item is heavily influenced by the ease with which we can retrieve it from our memory".
At Upscope we realised they copied and pasted their answer from our example
Upscope had a form asking customers:
"Where did you first hear about Upscope? e.g. "Googled co-browsing services".
Around 5% of the answers said "Googled co-browsing services".
This made the marketing team think "Ok, we should double down on that and create more ads for that search term as it works".
It turns out that new sign ups were simply copying and pasting "Googled co-browsing services" or typing that in directly with various spelling mistakes.
We discovered something both obvious and funny. In some countries people are not familiar with the abbreviation "e.g." and of course some people just wanted to get it done and the above was a close approximation of their search term or not at all.
5 star ratings all the way through but overall a 3 star? Yes
Here's an example by Mckinsey showing the importance of end to end customer experience. The individual steps were great but the overall score was not.
This is partly why the Intercom co-founder Des Traynor suggested someone in your company should be signing up to your own service every 2 weeks and going through the whole process to check on the overall process.
Another example is that, at Upscope, we realised were were sending too many emails from across several functions but no-one had complained until someone finally said "the emails were a little intense".
New technology distorts the data further
Google now have automated replies built into gmail.
Let's say you send a very angry customer an email confirming that you've completed a task for them and rather than typing out a fully reply they simply select Google's preset "Awesome, thanks!" reply.
Over time you've got 1000s of replies that suggest everything is awesome.
It's definitely not.
"Because if you know customers are having problems, it should be your goal to get to the root of it. On the flip side, if you know where customers are finding enjoyment in your product, you should want to find out why so you can expand on it."
"VoC best practices stress that you ask probing questions and not just set yourself up to receive positive remarks"
How to add vision on VOC to give context to data
On top of all the visitor analytics, text analytics, social media mood measuring, surveys and NPS scores you can add screen recordings and one to one co-browsing sessions.
The visual side is hard.
It's not hard to implement, it's hard to watch the truth.
When you've spent months on redesigning your website flow and then you watch customers fumble their way through and struggle on parts you never imagined would cause problems, that's tough.
When your data suggests everything is OK but it turns out they're simply not bothering to complain then you end up questioning your own strategy.
Add screen recording which has in-built struggle detection
Some companies use screen recording tools like SessionCam which can be integrated directly into customer experience tools such a Qualtrics and Medallia.
They can include features like "Struggle detection" which measures when customers struggle with an interface and records that session for you to view.
This can then be combined with data to give you a new set of assumptions to test.
Looking at a recording, seeing a problem and making changes is just as bad as going off just the data. The recording gives us a clue as to what might be missing in the data. That gives us assumptions to test. It is still data driven.
Co-browsing and one to one sessions
Talking to someone while watching them navigate through your website is painful but rewarding.
Co-browsing is a form of interactive screen sharing built for onboarding one to one because there no downloads or installs and you can click and scroll with the customer on their screen.
The difference between this and screen recordings is that you get to ask the customer what they don't understand and take notes as you go along.
This is often where you'll discover the real reason why they struggled with part of a form, navigation, what they understood from a set of instructions given and what's unclear.
Some companies are taking one to one onboarding to an extreme
Imagine Google onboarding every single person, one to one, when they signed up for a gmail account.
Superhuman are doing it for their email product.
Scroll down to the first point in this article to read more about both positioning your product in someone's mind and how it fits with onboarding them, along with Superhuman's super crazy plan.
Imagine the feedback they get from doing this. Many companies can't do it but Superhuman are scaling one-to-one sessions and likely have rapid accurate product iterations.