Updating the definition of the voice of the customer

The term voice of the customer (VOC) has traditionally been interpreted as being analogous with or part of the domain of marketing research (MR). With the advent of customer experience (CX) and enterprise feedback management technology (EFM), many companies are rethinking VOC, often assigning it to a department other than MR. This battle was most starkly highlighted by a headline on the cover of the November 2014 Quirk’s, “Is CX out to get MR?” Now that EFM has become primarily a technology play, even more departments like the information technology department are becoming involved.

Our conclusion is that CX will most likely end up owning VOC and that MR is facing a diminished role, especially since leading companies have been wildly successful using their CX to drive customer acquisition via word of mouth. Companies like Chick-fil-A, USAA and Harley-Davidson have indicated that they obtain 70 percent or more of their new customers via word-of-mouth referrals, so there really is not much need for either MR or even traditional marketing. But there is a need to rethink the definition of VOC, which we do below.

Gerry Katz states in his December 2013 Quirk’s article that some analysts have hijacked the term voice of the customer by completely misusing it.1 McInnes and Carroll introduce enterprise feedback management and define it primarily as surveys, complaints and social media inputs.2 In an article published the April 2014 Quirk’s, Subbiah and Bosik further complicate the discussion by noting that the hype about big data and EFM is reaching a frenzy.3 The Temkin Group, in both white papers and Webinars, has suggested that text analytics and continuous insights are the key to enhanced VOC impact.4 Finally, Whipkey suggests that CX could completely take over MR.5 We now have three terms: VOC, MR and EFM. Are they the same or, if not, how do they differ? Our conclusion is that they: are different; draw on different data sources; and serve different internal customers.

Katz’s criticisms of VOC provide a good starting place. He notes that VOC often does not support radical innovation and fails to facilitate marketing conquests. However, while criticizing the overlapping use of the terms VOC and MR, he never directly states that VOC is not MR.

We believe that they are two different animals. VOC describes the experience of existing customers while MR is primarily aimed at developing the best product and strategy to tap the rest of the market by gaining new customers, usually called marketing conquests. VOC can be used to not only improve future CE but, if properly defined, can be used real-time to enhance current CE. Further, Katz’s analysis limits the sources of data that describe the experience of current customers as well the analysis and use of such data.

We briefly review his critique and then suggest, based on how technology has evolved in the past two years, how VOC must be redefined in much broader terms if it is to achieve its full potential.

Current views are flawed

Katz suggests that current views of VOC are flawed for three reasons, each of which is associated with a set of VOC users:

1. The first set of users discussed are the “Naysayers” who suggest that VOC does not highlight needed innovations. Katz suggests the basic problem here is methodological in that you should not ask customers what they want but rather what they are trying to accomplish. While we agree that this more basic analysis is very effective, a lot of VOC output has to do with enhancing the existing product or process rather than making the next great leap forward.

Customers often are relatively satisfied with the existing offering. If they are not satisfied they would not be existing customers. On the other hand, Starbucks customers, participating in mystarbucksideas.com have suggested almost 200,000 new operating and product ideas, many of which have been successful. While they may have not suggested anything on the level of, say, the Apple Watch, they have supported serious innovation. Our point is that VOC need not only look for radical enhancements.

2. Katz’s second criticism addresses the segment he terms the “Stretchers,” who suggest that VOC is helpful but often not actionable because it primarily looks at customer satisfaction. He suggests that this problem can be solved by going beyond satisfaction to look at the key drivers of satisfaction and loyalty. As he explains, we cannot just look at the fact Johnny got an F, we need to understand why Johnny cannot read. So far, we agree but the true cause of lack of impact is due to other factors.

We believe that most VOC processes do look at key drivers and still have little impact. The lack of basic driver analytical output is not the primary cause of ineffectiveness. As will be noted below, the VOC is often neither packaged to suggest a clear direction of action nor converted into revenue impact implications that motivate action. Further, there are often two sets of key drivers depending upon whether customers have encountered a recent problem.

3. The third segment of VOC detractors are the “Technoevangelists,” who say the new social media and text/speech analytics tool will replace almost all traditional data collection activities. In fact, we think technology will dramatically enhance the accuracy of the description of the customer experience, but in a completely different manner than the articles suggest, by drawing upon internal operating data that describes the customer’s current and future experiences.

Our view is that none of these characterizations are appropriate and are rooted in the incorrect assumption that VOC and MR are the same thing. They are not.

Totally different environments

Current customers and marketing targets (potential customers) reside in totally different data environments. Because of this, how companies see them must be constructed totally differently. VOC studies and reports the experience of existing customers. MR examines “the market” and how to gain new conquests. The dynamics of these two targets, existing customers and future conquests, are totally different. Existing customers have an experience with the product and have already committed to the brand. The key objectives of VOC for this audience are enhancing loyalty by preventing problems, upselling and cross-selling and increasing engagement to foster positive word of mouth. Potential customers most likely are committed to another brand and must be lured away from it. Even the willingness to provide feedback is completely different between the two groups.

EFM is introduced by McInnes and Carroll and further described by Subbiah and Bosik. Both articles describe the key change as creation of an IT platform for developing surveys administered via multiple channels, building online communities and analyzing and reporting data from social media and surveys. However, they use the term EFM almost interchangeably with the term big data and fail to describe how much broader EFM can be.

Our June 2013 article in Quirk’s pointed out that big data, which is not at all related to surveys and complaints, can often more accurately describe the customer experience than can service interactions and surveys.6 This is because the company has internal operating data describing the customer’s transactions, transaction failures, purchases and shipments, often in more detail than the customer could ever provide via surveys and complaints.

Often, based on internal operating data, you know what the customer experience has been or will be even before the customer does. For example, a delivery company knows from internal data that 500 customers did not receive their shipments on time. Its traditional VOC system only received 200 complaint calls and 100 satisfaction surveys. The internal operational data can be more accurate and timely than the contact data or surveys in estimating the size of the problem.

This data can also be used for immediate, proactive service that dazzles the customer. For instance, ServiceMaster notes an upcoming work order for replacement of a home water heater and proactively apologizes to the customer for their two days of cold showers (raising satisfaction 20 percent just by warning and apologizing in advance).

We do not suggest ignoring traditional data sources. The surveys can best provide a short-term estimate of the damage the problem caused. Long term, a big-data analysis can combine experience and actual sales data to calculate an accurate, verifiable impact of various experiences on total sales, loyalty and revenue. Further, the operational data can actually be used to proactively affect the customer experience by warning the customer in advance that the package missed its connection, for example.

Primary differences

MR, VOC and EFM have different objectives. The following outlines their primary differences.

MR is about winning new customers. MR is aimed mostly at gaining conquests and producing specifications for products and services and an estimate of what customers will pay for them. Its data sources consist of surveys, focus groups, interviews with non-customers and ethnographic research. The objective is to identify the attributes of a product the customer most desires or would consider buying. There is significant overlap with VOC in that existing customers can also be interviewed and participate in online communities. This research can also include generic ethnographic research looking at how customers in general use the overall product category, e.g., how customers use a clothes washer and detergent to do laundry. One challenge in much market research is the level of customer commitment and thought – studies of generic products are often hypothetical. The customer really has very little at stake and therefore is not providing as thoughtful an answer to the questions. Market researchers seldom, in our experience, focus on complaints and service transactions because of the difficulty in extrapolating them to the marketplace as a whole.

VOC is about keeping your existing customers. VOC includes several of the same data sets as MR but the sources are limited to feedback from existing customers. However, the customer has more at stake because they have actually invested in purchasing and using the product. We have always encountered higher response rates and more thoughtful open-ended answers to questions from existing customers than from members of panels who are dontdontevaluating hypothetical products. (We know we’ve now angered all the panel companies who read and advertise in this journal.)

Further, VOC can include internally-generated operational data on customer transactions as well as employee input on those same transactions. The output should be customer satisfaction with existing products and the revenue damage of less-than-perfect experiences.

EFM should be about integrating and acting on data describing the experience of existing customers. EFM is reasonably well-defined by McInnes and Carroll. It primarily consists of gathering and reporting information on all the customer feedback sources, e.g., surveys, complaints, social media and online communities. It also should include information on customer transactions as well as data which allows for segmentation of both existing and potential customers. The output of this broader analysis can include suggested new offers to make to existing customers (based on their history) and some of the other outputs of what is called big data. What is not included is operations data and employee input on causes of customer issues. Also, there often is no way of integrating the data into a single picture of the CE. We have seen that multiple realities from multiple data sources can create conflict and paralysis.


Figure 1 suggests a view of the three types of data collection activities, MR, VOC and EFM. The chart in Table 1, drawn from Goodman’s book,7 outlines the full range of data that should be part of both VOC and EFM outputs as well as support MR.


Deserve to be reviewed

Now that two years have passed, Katz’s three criticisms from VOC naysayers deserve to be reviewed.

1. VOC does not create significant innovation. When combined with customer experience journey-mapping and ethnographic studies, VOC can be a catalyst for significant innovation. One could argue that Uber was created by identifying and responding to the three main weaknesses of existing metropolitan taxi systems: unreliable, unclean/unsavory and expensive. As noted earlier, Starbucks had done well using VOC delivered via online communities. Part of the analysis must be to look at the causes of issues beyond basic product performance including incorrectly-set customer expectations and customer usage errors.

2. VOC has limited impact because it is not actionable and does not create change. We agree that most VOC reports have not been prescriptive. They tend to say “X percent of customers would recommend and these product attributes get the best score; do what you want with it.” This lack of impact is a function of three issues.

The first is what we call score-chasing. This problem arose with the advent of executive dashboards in the 1990s. Executives wanted a single metric that summarized a sector of business performance. As long as that score was within acceptable range, no further attention was paid to that aspect of the business. This was the genius of the focus of VOC on satisfaction scores. The latest versions of this single score focus are Net Promoter and Customer Effort Scores.

The other two causes of lack of impact lie in basic weaknesses in the research output: failure to create an economic imperative for action; and reluctance to suggest a plan of action and organizational accountability.

If the analysts interpreting the VOC data have a broad enough understanding of the business to suggest what should be done as well as best practices in the arena of the customers’ points of pain, actionable recommendations can be easily generated. The usual barrier is justifying the resources to execute the recommendations.

A recent study by our firm found that in companies where the CFO has bought into the business case of the VOC, there was double the probability of sustained satisfaction improvement and five times the probability that most issues raised by the VOC were fixed in a timely manner.8 We have also found that when the revenue and word-of-mouth implications of not acting are credibly quantified, that action usually is taken. Quantifying the cost of inaction precipitates action. Therefore, creating the business case and suggesting specific plans of action are both critical to VOC impact.

3. Technology will dramatically enhance the accuracy of VOC because it often can report on CE better than traditional complaints and surveys. This point is partially true for multiple reasons. Social media is more immediate but much harder to project to the marketplace. Our firm’s December 2013 National Rage Study found that only 6 percent of consumers complain on the Internet and social media and less than 30 percent use review sites.9 Therefore, social media, while growing, is still not anywhere near projectable on the population.

Text analytics can provide more detailed analysis of customer expectations and motivation than traditional survey analysis. As the Temkin Group notes, text analytics provides much richer detail than closed-ended surveys or manual analysis of social media. Speech analytics can also discern types of emotion. Complaints, tied to text analytics can explain much of the why and how the customer ended up in the current situation in terms of incorrect expectation, customer error due to not reading directions and factors external to the product. On the other hand, surveys can provide a more reliable estimate of satisfaction and intention to repurchase than complaints. Closed-ended, scaled questions can be linked to actual future purchases and operational data (an application of big data).

Technology is already going well beyond text analytics and even beyond what Subbiah and Bosik allude to and even the broad automated reporting formats that McInnis and Carroll describe. Operational data, as noted in the delivery failure example above, can more accurately describe the CE than the customer can via complaints and even surveys. Further, operational data and online communities allow continuous inputs which do not just contribute to strategic reports on what needs to change in the company offering but also can drive immediate action on the ground. This is what we called the delivery of “psychic pizza” in our June 2013 Quirk’s article – using intelligence (also known as big data) to proactively intervene in the customers’ situation on-the-fly because you know what the problem is going to be before the customer knows it is happening. Finally, as noted above, online communities can provide ideas and indications of levels of interest. For example, a recent Frito-Lay contest engaged millions of consumers with over 7 million likes and 1 million votes.10 However, rigorous extrapolation to the marketplace as a whole is still difficult.

Gaining conquests

In summary, we believe that VOC differs from traditional MR in that it focuses only on existing customers while MR primarily looks at gaining conquests. EFM, to be successful, must go beyond an aggregator and reporter of surveys and customer complaints to mapping all the data into a unified picture of the customer experience. Our conclusion is that technology has not yet fulfilled its potential. In the near future, by drawing on operations and employee input, EFM will expand and supplement VOC sources, not replace the traditional ones like surveys and complaints.

Our suggested definition of VOC is as follows:

  • VOC draws on all data that describes the end-to-end CE of existing customers including surveys, complaints, operational data derived from internal systems, employee input and social and online community data.
  • It measures causality of CE, drawing on a broader set of causes than they traditionally have looked at.
  • It quantifies the economic implications in terms of damage or opportunity to revenue and margin by issue.
  • Finally, it provides prescriptive reports to multiple audiences that humanize the data and creates the economic imperative for action and suggest plans of action.

Five actions

Want to build a top-flight VOC process? There are five actions you must take.

1. Separate your analysis of current customers’ experience from market research. Existing customers have been exposed to your education as well as your products. Understand their experience and product utilization as well as problems and questions. Ideally conduct key-driver analysis of those customers with problems vs. those who have not had problems. You will find the drivers are quite different. Whether MR and VOC are located separately or together is purely a function of your company’s internal politics.

2. Supplement your survey and complaint data with operational data and employee input. You most likely have internal data that describes the customer experience with your products and services, whether orders, transaction interactions, service requests or billing inquiries. You will need to partner with the IT department to gain access to this data. A useful medium of communication with IT is the CE journey map.

3. Quantify the revenue implications of the current CE. Create a credible economic model of the revenue, word-of-mouth, risk and cost impact of key customer problems and opportunities as described by the VOC. This analysis must be made credible to the CFO and CMO. If they accept it your VOC output will have dramatically greater impact. Quantifying the cost of inaction precipitates action!

4. Report VOC at multiple levels for multiple purposes. VOC can provide guidance on individual service rep performance, on team performance, can identify weaknesses in response guidance and processes as well as needed improvements in product offering and marketing and sales processes. It can occasionally identify breakthrough opportunities for dramatic enhancements (based on both employee and customer suggestions). Each level requires a separate analysis and separate reporting processes. It is more labor-intensive but has much greater impact.

5. Develop suggested action plans specifying what should be done as well as accountability for results. Two of the three weaknesses we’ve seen in VOC processes are that the insights manager does not have the courage (or operational knowledge) to suggest what should be done and who, across the multiple silos, should take the lead. You can get input from the functions and then suggest what should be done and who is best situated to lead.

1 Katz, Gerry. “Hijacked again!” Quirk’s Marketing Research Review, December 2013.
2 McInnes, Andrew and Carroll III, John. “Charting a new path to the same destination.” Quirk’s Marketing Research Review, October 2013.
3 Subbiah, Suresh and Bosik, Darren. “Instead of big data try value data.” Quirk’s Marketing Research Review, April 2014.
4 Bahil, Denise and Temkin, Bruce. “Text analytics reshapes VOC programs.” Temkin Group, Boston, May 2014.
5 Whipkey, Eric. “Insights in their sights.” Quirk’s Marketing Research Review, November 2014.
6 Goodman, John; North, Peter and Beinhacker, David. “Taking the good with the bad.” Quirk’s Marketing Research Review, June 2013.
7 Goodman, John. Customer Experience 3.0, AMACOM, 2014.
8 Goodman, John. Factors for Customer Experience – VOC Success, Callcenter Pipeline, January 2012.
9 Broetzmann, Scott, Grainer, Marc and Goodman, John. 2013 National Rage Study, Customer Care Measurement and Consulting.
10 “Wasabi ginger wins Lays potato chip flavor contest.” NY Daily News, October 21, 2014.

Authors: John Goodman (vice chairman), David Beinhacker (director of research) and Scott Broetzmann (president and CEO), from Customer Care Measurement and Consulting, an Alexandria, Va., firm
Article was first published on Quirk’s Marketing Research Media