Go Beyond the Survey to Learn More About Your Customers

Surveying your customer base is an essential best practice for customer service and support teams alike. But the survey is a medieval construct. Here's how to get a 21st century view of customer satisfaction.

This article is the third in a series of excerpts from the second edition of David Taber's Salesforce.com Secrets of Success, which is scheduled to be released later this year.

Previously I've covered the importance of surveying and scoring your prospect base and social listening.

Those articles, though, focus on the prospect who's still in the funnel, not the customer who's made a purchase. For customer service, the issue has to be customers—and the metrics of "customer satisfaction" are usually relatively primitive. Think about it for a second: Marketing scores its work product (lead "hotness"), sales scores their work product (the forecasted probability of close), but most support organizations aren't scoring their work in process, aside from a happiness survey once a year. It's time to raise the bar.

Fix Your Customer Survey Flaws

There's nothing wrong with an overall annual survey of customer satisfaction. Attach that data to the account record. Business-to-business and business-to-consumer companies alike need to add brief surveys at the end of every customer service or support transaction, collecting info on the timeliness and effectiveness of the problem-solving as well as the general "willingness to recommend" your product or company. (Full disclosure: I'm a card-carrying marketing guy who believes that willingness to recommend is the only question that matters in marketing.)

How-to: Use BPM to Improve Customer Experience

You can attach that data to the case and contact record. But after a few months, start analyzing them to determine data quality (percentage of wild points), distribution (average is nice, but quintiles are better) and clustering (divide them by product, geography, customer type and look for modalities). You may find some interesting trouble spots in your sales, product and service groups just with that analysis.

But you'll need to dig a bit further. Surveys are almost inevitably flawed in four big ways:

  1. Surveyor bias. Questions focus on things the surveyor cares about, inevitably omitting things that are relevant to the customer. Provide optional text forms on every question, and end the survey with the blanket question, "Is there anything else you'd like to tell us?"
  2. Semantic bias. Questions make sense to the bureaucratic needs and language of the surveyor, not the language and context of the customer. Pre-test surveys with neutral outsiders to sidestep this problem.
  3. Sample bias. The only answers you get come from people with the time and inclination to answer surveys. This omits key members of the audience—namely, those who are busiest and most important. To improve response rates, offer tchotchkes and other incentives for people to answer.
  4. PITA skew. There's a wonderful book whose title says it all: Don't Make Me Think. In the Short Attention Span Theater that is modern business, you have to keep your survey to five multiple-choice questions answerable from an email optimized for a smart phone. Provide a link to your website for those who want to wax poetic. Thank your lucky stars if more than 5 percent of respondents go there, even if you offer a prize for doing so.

Related: 3 Lessons Learned From a Failed Customer Feedback Software Test

Tap CRM System to Identify At-Risk Customers

Surveys are essentially medieval. Don't believe me? Check out the Domesday Book of 1086. You'd think with cloud computing, we could do better.

For the purposes of assaying customer happiness, what we really want is a propensity to re-up, renew support and recommend our services to others. While there are lots of components, the aggregate score should best be thought of as an "at-risk index," with a lower score being better. (For those screaming, "But higher always has to be better for my executives…," too bad. The scientific method says you can't prove the positive here; only the absence of the negative.)

The at-risk index should be collected at the contact level in the CRM, but it must be aggregated to the opportunity record, as this is typically your closest proxy to a project or purchasing entity.

What are the measurement points for this at-risk score? Fortunately, you already have a lot of them in your CRM system:

  • Direct indicators such as number of SLA violations, long-open cases, unsuccessfully closed cases, complaint emails, non-renewed contracts, non-upgraded products, number of negative course evaluations and so on
  • Implicit indicators such as case notes suggesting that a customer isn't a happy camper, posts in your online document commentary area ("How can we make this page better?") and in your ideas area ("How can we make the product better?")
  • Occluded Indicators, or a negative-sentiment index as measured by a social media monitoring system that evaluates posts in your online community and, if you know the person's handles, in social networks

How-to: Use Big Data to Stop Customer Churn

If you're lucky enough to have a product with usability monitoring built into the UI, or if yours is a SaaS offering, you can collect click-by-click metrics on where a customer's getting stuck. While this data should really be collected and used to improve the product—that's how Salesforce itself uses this data—it can also be aggregated to discover a frustrated customer. This data is unlikely to be in your CRM system now, but best-in-class companies are all going that direction.

Set Realistic Scoring Expectations

Scoring systems may be a pain to set up, but it's even harder to get really good results from them. That said, it's worth it.

Technology isn't the problem here. Scoring must be based upon a model and a set of assumptions, and nobody has a really solid model for how individuals decide whether they are happy about a purchase or a company. The modeling errors from a single bad assumption or weak relationship easily overwhelm the predictive power of a dozen good elements.

Expect to tune your scoring models for many months, if not quarters, before the results are credible. Because of this break-in period, it's critical to set expectations low, particularly with executives and sales people who'll really only give you one chance.

As always, there's a key caveat: Analytics will inherently magnify any problems you have with data quality. Be sure about your semantics and data filtering before you make critical decisions based on either scoring or analytics.

David Taber is the author of the Prentice Hall book, "Salesforce.com Secrets of Success" and is the CEO of SalesLogistix, a certified Salesforce.com consultancy focused on business process improvement through use of CRM systems. SalesLogistix clients are in North America, Europe, Israel and India. Taber has more than 25 years of experience in high tech, including 10 years at the VP level or above.

Follow everything from CIO.com on Twitter @CIOonline, Facebook, Google + and LinkedIn.

To comment on this article and other CIO content, visit us on Facebook, LinkedIn or Twitter.
Download the CIO October 2016 Digital Magazine
Notice to our Readers
We're now using social media to take your comments and feedback. Learn more about this here.