The Biggest Problem With Customer Surveys?
I am feeling a bit disturbed and disappointed at the moment. Disappointed with myself, really. I have spent too much of my life managing survey processes that don't make enough of a difference for customers. With a reasonable amount of hindsight, I now feel there is one problem with customer survey and improvement processes that overwhelms all others. And that problem is essentially the failure to address issues for the 90% of customers that have not responded to your / our surveys.
Let's put aside the issue of companies that perform surveys and do nothing at all with the results, at least for now.
The (unacceptable) norm
The standard survey process involves asking questions, determining a score and trend, and taking action. The bare minimum action is to contact angry customers and fix their problems. Often called "Detractor recovery" or something similar.
Let's suppose yours is a product company and you have a 10% response rate to your surveys. It turns out that your Detractors / complaints are often about late deliveries. You contact those who have complained and do your best to empathise and recover the situation. You pat yourself on the back and feel great about what you have done.
But what about the 90% of customers who did not respond to the survey? Many of them had late deliveries too. Are you simply assuming that since they said nothing, everything must be OK? (Try that in your personal relationships. Does absence of communication about a significant topic prove that everything is going well?)
Surveys are largely a waste of time, compared to the alternative
But, and this is the main point I am trying to make here, you already know which customers have been delivered late. You know it for 100% of the cases from operational data. You don't need surveys and you must not use them to find out who is unhappy with the basics. Surveys should mainly be used to periodically calibrate your system; to help you to understand which of your operational KPIs matter most to your customers. Then use those KPIs to detect unhappiness and to trigger improvement actions.
AI-based tools can help... a lot
Yes, that is difficult to do well. That's where modern AI-based tools come in. They are particularly useful for detecting patterns of KPIs that accurately predict customer retention. In the world of customer experience, only your customers have experiences with you, by definition. Retention is what it is all about, including upselling and cross-selling. And it has to be about retention of all of your customers, not just those who have responded to surveys.
In short...
In short, the biggest problem I see with surveys is that we depend on them to detect customer feelings about us and the actions they are likely to take. We depend on something that provides information about 10% of our customers, while we are sitting on unused data that has the capacity to provide insights about 100% of them.
Feel free to agree or disagree below...
CEO & Founder @Thematic | 20yrs+ Machine Learning & Natural Language Processing experience | AI Phd | Science Communicator
3 年Maurice, sure, AI models can be trained to predict "Which customers are likely to leave?", but will it tell you why or what you can do about it? Only partially. You really need to dig deep into feedback across many data sources (Intercom, contact center, complaints, in-product surveys, reviews, community, social media etc.) to build a picture of how to reduce churn. I also would note that depending on the company & the feedback collection methodology, the distribution of issues among those who did not respond might be similar or different. For example, I was buying an exercise hula-hoop recently and noticed that its Trustpilot reviews tend to focus on delivery issues, but Facebook review ratings were much higher and talked about the quality of the product. Finally, when it comes to delivery, I hope companies with these issues address it both for specific customers and also as a whole. After all, it's a Hygiene Factor (or a must-have). :) Thanks for initiating this discussion!
Executive Educator, Inspiring Leadership and Driving Exceptional Customer Experience for ambitious Enterprises | Founder: Commercial Excellence Partners | Speaker | Travel-Tech ?
3 年There 2 issues at play here: service recovery for those who have “complained” - via their response to the survey - and gauging customer “feelings”. I’m not sure I’ve seen companies lean on the survey methodology for the former. Usually, it’s just a reaction to a complaint made by a customer. Granted many people don’t complain, and therefore receive no resolution; but this is another story. As far as trying to gauge customer sentiment, loyalty, preference, feelings etc etc. the 10% who respond may be statistically relevant enough that you get close to the “truth”. The other 90% are technically irrelevant- assuming the sample size is large enough. My personal issue with surveys has nothing to do with this. I hate surveys because they are usually badly constructed, badly implemented, barely understood, and you’ll likely never know if customers are telling the truth. That last bit is made worse if you’re asking customers to answer the “wrong” questions.
C-Suite Strategist | Thinkers 50 Top 10 | Best-selling author | Columbia University Business School Professor
3 年I would add 2 big problems. Not using relative measures for your independent variables and not having a dependent variable. Oh and terrifying your staff so they TELL your customers that anything less than a score of 10 will have serious repurcussions for them!
Helping organizations create a better future through Space that supports invention, discovery and innovation
3 年Well said. Charity Wilson, MBA, MA in Economics was able to predict NPS scores based on metrics. There's a great little video about a pub being run like a bank. My favorite part is right at the end.