How conversational feedback increases data quality: Part 2 - Advanced Concepts7th November 2019
Here are a couple of examples of how we achieve higher data quality in comparison to traditional surveys with Hubert.
Refining uninformative responses
Many of our customers experience 'noise' in their survey results. More specifically, responses in line with 'everything was great' or 'the customer service sucked a' are not very exhaustive when it comes to information density.
When Hubert deems a response as vague, he always follows up with a probing question such as 'can you please tell me the best parts' or 'can you provide some additional details please'.
Take a look at the case below to see how we helped one of our clients reduce their noninformative data from 31% to 4% using intelligent follow-up questions.
Catching misinterpretations early
Have you ever sent out a survey just to realize five weeks later that a substantial part of the respondents misunderstood one or more of your questions?
Hubert is trained to interpret the response in relation to the question and determines whether the reply is relevant or not.
By catching misinterpretations early, Hubert can fix them while there's still time instead of finding out when browsing the results. You can be confident that your end-data is valid and closely related to the question.
Common frustrations when filling out surveys are partly idiotically phrased questions and partly idiotically phrased options to choose from.
What if I heard about your service from a friend, who is also my colleague, that sent me an online link?
Tough luck survey-maker. You'll be the last to know.
Not only is this bad from an information-perspective, but these kinds of screw-ups causes millions to exit surveys prematurely and is a clear contributing factor to worldwide survey contempt.
"If I'm to spare my precious time filling out your survey, the least you can do is make sure I can give you my feedback in an easy way."
Here's another common (and annoying) survey question:
A company that's serious about improving their customer service would never use this kind of question.
Sure it'll look good in a graph (provided you have mostly positives), but customers with something to transmit will become frustrated by this question. Which means you are losing out on highly valuable feedback.
In the 1980s, it was somewhat understandable to primarily use closed-ended questions when dealing with large crowds, but it's 2019 and there are far better options available.
Here's how a properly trained chatbot would handle the above scenarios:
Why is this so much better then?
Let's walk through the conversation in detail.
Firstly, the respondent is not bugged by having to read through a bunch of different pre-set responses and can instead respond freely. Hubert then decides to dig a bit deeper and ask a relevant follow-up question which both improves data quality and reinforces the feeling of actually being listened to.
Secondly, instead of just being able to show off a graph, now you have some quality feedback to act on in order to actually improve your customer service.
And finally, to be able to show a nice graph to support your open-ended data, Hubert asks the respondent to rate the experience. Even here, the respondent isn't limited to just the buttons but can give even more detailed information describing how to perfect the experience (Hubert would interpret the response as a 9).
Nothing about surveys inspires an exhaustive response. Filling one out doesn't come naturally and getting it over with fast is what truly counts for most.
Everyone knows exactly what to expect when agreeing to participate in a survey and most of the time, the sense of boredom is reinforced.
Messaging, on the other hand, is associated with something fun. Texting is the number one way of communicating for everyone under 45 and continues to grow year by year. The four largest messaging platforms have a combined user base of more than 4 billion people, which shows the magnitude of messaging in today's society.
Chat conversations are something you participate in out of your own free will because you have something to express. Our empirical experience shows how well Hubert regularly inspires rich responses on a whole different level than surveys. The richness in data quality can simply not be compared and is something completely new in respondent research.
One of our long-time clients, Siemens were amazed by how much better the data quality was in comparison. Read more about that in the Siemens case-study.
These and more advantages are what awaits you when you make the shift towards conversational feedback.
Don't let your respondents suffer through another endless Likert scale survey. Contact us now and let us show you the future.