Bots for Business
Gathering More and Better Data with Survey Bots
Engaging and conversational surveys encourage customers to provide feedback. Chatbot surveys and survey bots help you to collect better, higher quality data.
April 2, 2019
If you forget to listen to customers, you can expect backlash. Case in point: Netflix attempted to rebrand its DVD-by-mail service back in 2010. In an effort to separate its streaming platform from its DVD rentals (and to raise prices), Netflix renamed the legacy business Qwikster. Netflix customers took note—and bailed.
One month and 800,000 lost subscribers later, Qwikster was no more. In addition to becoming great fodder for “worst rebranding fails” lists, Netflix’s blunder is a cautionary tale against making significant decisions without soliciting customer input.
Surveys are a great way to listen to your customers. But a great way isn’t always a foolproof way. For you to extract meaningful insights from a survey, your customers have to complete one. And getting customers to complete a survey can feel like pulling teeth.
Enter chatbots. Using a chatbot to survey customers can be an effective strategy for collecting more and higher quality responses, as compared to standard, form-based surveys. You get the information you need to make smart business decisions that further endear you to customers.
But first, you’ve got to create a survey bot that truly engages them.
Creating a customer survey that works
Imagine you just ordered a slice of key lime pie. You’re sure it’s going to be delicious. Except that it won’t, if the limes are rotten, the crust is dry, and the chef didn’t add enough sugar. Then it’s not even edible. After all, the success of anything—even something as universally appreciated as pie—depends on the quality of its execution.
Ditto for customer surveys. Poorly worded questions, nebulous answer choices, or a thoughtlessly generic approach can make a survey close to worthless. Online survey provider Questback points to five types of survey questions that may turn customers off and result in useless information:
- Yes/no questions that lack context, such as, “Did you enjoy your experience today?”
- Personal, prying questions, like, “How much money do you make?”
- Two-in-one or multipurpose questions that customers can’t easily answer in a text box or numerical rating, for example, “Did you enjoy your coffee, and was it priced reasonably?”
- Required questions, which can encourage users to abandon the survey or to choose random answers to finish it quickly.
- Confusing questions that can’t be answered accurately or that include loaded assumptions, such as, “What do you think about our excellent products?”
So how can you avoid these issues and craft a customer survey that actually yields fruit? By seriously considering the kind of information you want and making it easy for customers to provide that information. Oh, and by keeping the survey short. That, too.
Take Hilton Hotels, where customer surveys are taken so seriously that they’re handled by a full-time director of Syndicated Customer Research. According to the individual who holds that post, use of short surveys and a straightforward 10-point rating scale can optimize response rates and can help ensure that data is usable.
Slack is another company that places heavy emphasis on user feedback. Per Slack Founder Stewart Butterfield, the platform “take[s] feedback any way [it] can get it. Whether through a help ticket, a tweet, or a command created just for feedback (Slack’s got one!), the company goes out of its way to solicit input from users. In some form or fashion, these insights influence every new feature or product enhancement.
Is Slack responding to customer surveys? Not exactly. However, the organization’s approach demonstrates the value of going out of your way to find out what customers think. And of being a little bit unconventional.
Capturing valuable data with survey bots
Speaking of unconventional surveying tactics, gathering feedback via chatbot can be an effective way to obtain the information you need to build a better customer experience. Here are a few reasons to create a chatbot that can survey your customers.
Higher survey response rates
To enact changes based on customer feedback, you need a data set that’s large enough to be useful. And that means you need the highest possible response rate.
Response rates are tricky. According to online survey provider SurveyGizmo, everything from brand loyalty and recognition, incentives (like raffles), customer demographics, and distribution tactics impact engagement and response rates. That doesn’t mean your response rate needs to be in the high double digits—surveys of large groups can yield statistically significant data, even when response rates are low. Instead, it means you need to shoot for the highest response rate possible given the attributes of the group you’re surveying.
Using a chatbot for surveys doesn’t always guarantee a higher response rate, but it can help you assemble a usable data set. In a study conducted by Michigan State University (MSU) and YouGov, researchers found that surveying customers with Facebook Messenger bots improved response rates among recipients with high levels of brand recognition.
The lesson? Consider your audience. If they know your brand and they’re already comfortable using chatbots, higher response rates are likely.
Simple and engaging chatbots
As noted earlier, distribution tactics and incentives can impact customer engagement with your survey. Luckily, chatbots exist to engage! Chatbot dialogue that’s concise and well-written can go a long way toward helping you collect the data you need. And so can the speed with which your bot delivers the survey.
Compare a survey distributed via email to one conducted by a Facebook Messenger bot. For the emailed survey, a mobile user would have to find the email in their inbox, click a link, wait for a new page to load, complete the survey, and submit it.
The Facebook Messenger bot makes answering a survey easier. Messenger’s Quick Reply feature prompts users to complete a survey. They tap once, and they’re in. Simple chatbot dialogue, along with easy, single-tap numerical or yes/no responses, makes survey completion a cinch. It all unfolds in an instant, and there’s minimal commitment.
Higher survey quality
Another thing you can get from a survey bot is a conversation. And a conversational experience can boost the overall quality of your surveys.
As Liana Epstein explains on the SurveyMonkey blog, the order and flow of survey questions should always feel like a conversation. When they do, your customers are more likely to complete the survey. She uses the following example, which involves fictitious characters, with Lucy (questioner) and Charlie (respondent), who are on a blind date:
Lucy: “Do you have a good boss at work?”
Charlie, unsure how to answer: “In what sense? In terms of his work performance? In terms of his personality? Overall?”
Lucy: “I mean, how well does he manage you and your team?”
In real life, this exchange could get a bit awkward. It would be comparably uncomfortable in a traditional, form-based survey. You’d suffer from at least two of the survey no-nos covered earlier—two-in-one and confusing questions.
A survey bot, on the other hand, can do this well. The dialogue might unfold like this:
Bot: “On a scale of 1–10, how highly would you rate your boss’s performance at work?”
User: [Indicates 7]
Bot: “Thanks! Now, please tell me how well your boss manages your team. You can use the same 1–10 scale.”
User: [Indicates 8]
And so on. By putting your bot in control of the conversation and employing familiar idiomatic nuances, you’re in a unique position to build a high-quality survey that results in high-quality customer data.
Helping your business with a survey bot
So we’ve established that a survey chatbot can improve response rates, boost survey quality, and keep your customers engaged. But how can it help you meet business objectives?
For one thing, an effective survey bot can help you make more informed, impactful decisions than you otherwise would have. Consider a survey effort by Syllable, a California company that uses AI to improve health communication. Accustomed to a 10 percent open rate for email campaigns and a survey completion rate of less than 10 percent, Syllable decided to take a different tack—a survey chatbot they promoted with Facebook ads. The result? A 35 percent completion rate for all participants.
Compared to the below 10 percent completion rate from legacy tactics, the bot-centered campaign was a major success. Most importantly, Syllable’s client, a community healthcare provider, is now able to expand upon the limited (and relatively inaccessible) data available from patient records. They can develop health programs that target specific needs of the greater community, and they can do it with more precision. According to the campaign’s administrator:
We have consistently seen chatbots outperform traditional methods while conducting surveys for healthcare providers, small medical clinics, contract research organizations, and public health organizations.
Maybe the bot’s performance has something to do with the feel of quality chatbot-human interaction. Whenever you’re communicating in a chat window (as opposed to, say, filling out a form), there’s a narrative backdrop to the experience. When it’s designed well, it can be as if you’re engaged in a conversation with an actual human. At its best, a chatbot survey can give users room to tell their story, even if it’s a story tapped out in 1–10 ratings and yes/no responses.
Stories, after all, can yield greater insights than run-of-the-mill surveys can. That’s what Stephen Shapiro thinks. A self-styled innovation consultant and a frequent speaker at corporate events, Stephen is accustomed to receiving feedback about his talks—most of it positive. Congratulations are nice, but it was only after speaking with an audience member that Stephen realized how much more valuable a story can be. He explains:
Every day I receive surveys from companies wanting to know my opinion...The answers to these questions often don't get to the true value...Stories are a better way to learn a customer's true experience and the value they received. Of course, you can ask for examples on your surveys. But nothing replaces a conversation where you can go deep. Probe through dialogue. Gather real insights rather than platitudes.
So how can a chatbot encourage customers to report on these broader issues? That depends on your audience and the nature of your survey. But, as the experiences of Syllable and MSU/YouGov demonstrate, the greater engagement that users experience with a bot-driven survey is real.
Use it to your advantage, and you can unlock more and more nuanced customer data.
Exploring survey bot platforms
Ready to use a bot to conduct your next customer survey? Or wondering what the survey bot landscape looks like so you can begin to build your own bot? The following survey-focused chatbot platforms are excellent places to start:
This aptly named platform emphasizes ease of use. First, you use the Surveybot builder tool to create a survey. Then you deploy the bot over Facebook Messenger. Surveybot offers 18 different question types that you can choose from. Surveybot comes with a smorgasbord of analytics so you can analyze your survey responses in depth—straight from the platform interface.
Cons? Early on, some users lamented that Surveybot couldn’t survey respondents in all languages, but Surveybot says that they’ve cleared that up, at least for the most widely spoken languages. Other than that, the only downside is that Surveybot is exclusive to Facebook Messenger and Facebook Workplace. If you’re not deploying your bot over those platforms, you need to look elsewhere.
Calling itself a conversational survey, Wizu cuts right to the chase:
Higher completion rates and better quality responses. Sounds like the sort of things you’d be looking for in a chatbot survey. But how well does Wizu work in the wild?
Pretty well, if its feature set is any indicator. It also works on various chat platforms, so you’re not bound to something like Facebook Messenger. The only notable customer grumbles have to do with ease of use—Wizu has a lot of features, and it can take a while to figure out how to use them all.
Come to think of it, that might not actually be a problem.
Personalization is the AceBot.ai mantra. This platform emphasizes the ability to personalize the user experience using survey logic, thereby improving response rates. Other platforms also do this, but AceBot stresses it more than most. It’s also got robust analytics and team collaboration tools.
Subscriptions are billed monthly, which may or may not be ideal. Some platforms, by contrast, bill by the number of survey responses. If you send just a few surveys now and again, you may not want a monthly bill.
A longtime player in the online survey space, SurveyMonkey lets you distribute surveys via a Facebook Messenger chatbot. If you’re thinking about administering surveys via chatbot and other means, SurveyMonkey might be the way to go. You’ll be able to compare response rates across various survey formats and communication channels—all from a single platform.
But, if you’re just looking for a dedicated chatbot survey tool, you might want to look elsewhere, especially if you want customers to chat with your survey bot on your site instead of on Facebook Messenger.
Using bots to listen to your customers
Ultimately, the more you listen to customers, the more you learn from them. Plus, the easier and more enjoyable you make the survey experience, the more data you’ll have to learn from.
Nobody wants to make a Netflix-grade mistake. By engaging your customers in nuanced conversations with your survey bot, you’ll avoid the peril of making major decisions based on gut instinct. Instead, you’ll make those decisions based on what you know will work for customers.
And rest assured that your customers—current and future—are sure to notice.