Best Practices for Integrated Listening
1. Introduction
Listening to customers does not mean the same thing today as it did five years ago. Today consumers have found strength in numbers. In essence, the community is becoming the consumer. Socialization has dramatically reduced the barriers to group communication and organization. This is a huge risk for companies who can now watch missteps spread through an entire universe of critics, fans and self-appointed journalists, many of whom are not even customers.
It is not uncommon for a company to learn of a searing customer issue from flare up in a discussion forum rather than through its inbound service channels. Increased socialization also presents huge opportunities for companies who are able to successfully glean new trends and product innovation ideas from their customer communities, turning listening insights into strategic business success.
As customer comments continue to grow, companies need best practices and tools to make sense of the chatter. To listen effectively and continuously - and be successful - companies need a combination of several capabilities:
- Access to large volumes of customer comments from both direct feedback channels and indirect channels in social media.
- Technologies to automate the categorization of comments based on a company'sbrand attributes.
- Expertise to perform complex analyses and guide strategic decisions.
- The ability to execute automated engagement actions based on listening insights.
Successful integrated listening programs uncover customers' current and emerging needs, and help identify the extent to which companies are delivering on their brand promise. Structured feedback through customer surveys provides enormous opportunities for analysis. But one of the strengths of these approaches - providing data - is also a limitation. If companies overly depend on directly solicited data, they will not get a complete sense of a customer's emotional response to their brand, products or services, and they can miss early signs of future problems or opportunities. To avoid this data-only view of customer relationships, companies can implement programs and processes to continuously listen to their customers across both direct and indirect feedback channels. To succeed, companies must continuously monitor all customer conversations; identify gaps between the brand promise and customer expectations; and take immediate action to address those gaps.
This white paper examines best practices for collecting, categorizing, analyzing, and acting upon customer conversations generated across multiple interaction channels – the integrated listening process.
2. The Integrated Listening Process
There are four major steps involved in effective listening initiatives, the outputs of which continuously refine the accuracy of customer insights. Companies first need to be able to collect and gather comments from every customer touch point, each with their own idiosyncratic formats and protocols. Sources can be inside the company (emails, call centers, Web feedback forms, etc.) or they can be outside the company (review sites, discussion forums, text messages). Companies need to be able to package and send these comments to a centralized intelligent processor that categorizes them – regardless of their source. The intelligent processor uses sophisticated analytical algorithms based on categories specified by experts in the product/service, company, and industry. Next, an analysis engine evaluates the categories and sentiment (positive or negative) and discovers insights such as correlations and root causes and routes the insights to the right decision maker or application. Finally, that decision maker or application takes action. All of this happens in real time – thousands to even millions of comments should be processed in seconds, giving decision makers access to the customer experience as it is happening, not weeks or months after. Armed with this information, companies are now equipped to act and engage with customers to produce superior business results.This process is summarized in Figure 1 below.
Figure 1: Integrated Listening Process
3. Collect
When customer experiences are out of line with their expectations - either positively or negatively - they're more likely to provide feedback than when their expectations are simply met. Effective listening requires clear mechanisms for capturing customer comments from any interaction channel. This includes direct channels such as web feedback forms, emails, SMS, support forums, and call center transcripts; as well as indirect channels such as external review sites, discussion forums, blogs, and social media site postings. The following are several best practices for maximizing the volume and quality of customer comments collected by the enterprise.
3.1 Make it easy for your customers to provide direct feedback.
Companies should make it as easy as possible for customers to provide them with comments about their experience interacting with their brand, products, or services. This means making the option to share their comments readily available at every customer touch point, including web sites, feedback cards, email, call centers, and other direct forms of contact.
A review of Fortune 500 web sites revealed that feedback forms and contact emails are often buried deep within a site's navigation structure, requiring 2-3 clicks to get to a contact link, making it difficult for the public to communicate directly with the enterprise. The option to provide direct feedback should be easily accessible from the moment a customer enters a company's website. Doing otherwise is akin to placing a supermarket's customer service counter by the back loading dock instead of by the entrance. In addition to feedback forms, email addresses, phone numbers, and other contact details should be easily accessible on a company's home page. A good example of this practice in action is Yahoo, Inc., who has placed links to direct feedback functionality at the top and bottom of all their most trafficked properties - a practice which has yielded in excess of 500,000 feedback records for analysis per month.
Moreover, direct feedback channels should allow customers to share their thoughts and comments in their own words. Companies that resort solely structured surveys and ranking mechanisms have a difficult time understanding the root cause and emotional sentiment of how customers are interacting with their brand. Additionally, customer feedback requests that lead with free form questions achieve a significantly higher response rate than those which do not.
Companies can also increase inbound comment volume by developing a direct "feedback imprimatur" culture, whereby all collateral, web sites, receipts, and other customer-facing materials include clear printed directions to direct feedback mechanisms (e.g., web site, email address, and phone numbers printed on a restaurant check).
3.2 Provide feedback opportunities at all major customer touch points / transactions.
Companies that solicit feedback at every touch point will be able to collect valuable customer insights at the point of customer passion. These touch points include any transaction or interaction a customer has with the enterprise, and varies across industries.
Table 1: Soliciting Feedback at Customer Touch Points by Vertical Industry
| High Tech | Software | Etail | Telecom | Travel |
|
Registration |
Registration
|
Home Page |
Operator Assistance |
Reservations |
3.3. Develop and participate in on-line communities to obtain customer opinions and sentiment.
The development of online customer communities provides another way in which companies can generate high quality feedback. In such communities, groups of like-minded individuals focused on a company’s brand, products, or services, come together, and interact online. Successful examples of such communities include XMFan around XM Radio, MyCokeRewards, for Coca Cola, and Dell IdeaStorm for Dell Computers. These communities encourage discussion, self-service, and offer participants product previews, special promotions, and other benefits. Customer conversations in these contexts can be leveraged to improve the customer experience in these communities in real-time – from proactively managing customer complaints to leveraging cross- and up-selling opportunities and improving the relevance of product development campaigns.
3.4. Mine external social media sites for customer opinions and sentiment.
Indirect sources of brand conversations such as review sites, forums, and blogs are another rich source of feedback. Customer communities often sprout up on the initiative of passionate customers, and their conversations contain unfiltered sentiment and valuable insights. Companies that can glean such insights, and participate in conversations outside their “four walls,” can be perceived as being truly customer-driven above the competition. A great example of social media listening success is the case of BMW of America’s monitoring of the Mini brand, which revealed a great community affinity by owners that led to one of the most effectively designed marketing campaigns in automotive history. In order to collect information from social media sites, companies need to invest in data extraction technology that is able to scrape user posts from forums, review sites, blogs and other social media properties. These technologies are able to extract selected posts from a site and handle basic pre-parsing to create relevant streams of data for analysis. These data can then be submitted for categorization and analysis using the tools described in Section 4.
3.5. Encourage comments through the use of rewards and promotions.
Direct feedback can be actively solicited via rewards, sweepstakes, and promotions. An immediate benefit in return for providing feedback motivates customers to take time to share their comments, and also builds confidence that a company is listening. Similarly, consumers can be encouraged to participate in conversations on community sites by providing posters with value in the form of exclusive content, pre-release products, special deals, and other brand-enhancing offers. These tactics set the stage for developing ongoing relationships with frequent contributors, and winning them over as spokespeople for your company. A great example of this strategy in action is the Microsoft MVP program that recognized technical community leaders for sharing their high quality, real world expertise in offline and online technical communities.
3.6. Capture structured data in addition to free-form comments.
Companies soliciting direct feedback and scraping community sites for post should capture relevant attributes such as a company’s referring URL, navigation history, post date, demographic information, structured field attributes, and any other data that can be used to develop multivariate analyses and deliver an appropriate response. In addition to better analytics, such data capture enables companies to respond directly to customers who leave their contact details as one of the structured attributes. This collection of additional data should be done without encumbering the consumer with privacy concerns or lengthy surveys. A good rule of thumb for structured data collection is to ask for no more than 10 explicit attributes in feedback forms (e.g., name, email address, date of incident, store location, product line, Net Promoter Score, etc.). Social media posts like forums or review sites can similarly be scraped to include several key attributes, including site name, crawl date, author, location, post date and timed, etc.
3.7. Acknowledge direct feedback immediately upon submission.
Immediate and personalized acknowledgement of direct feedback after it is received reinforces a company’s commitment to the listening process and is a critical component of an effective listening program. Such acknowledgments should re-state the topic of feedback received and address any overt sentiment. By immediately generating a response via web or email, a company initiates a two-way exchange with a customer. No longer is feedback a one-way process, instead companies are able to kick off a dynamic conversation with customers based on their specific interests and concerns (this will be discussed in more detail in Section 5 below).
3.8. Publish findings to customer communities to encourage virtuous feedback cycle.
One of the best ways to encourage customers to share their comments is by reporting a summary of the most discussed issues and topics back to the customer community. Whether it is via direct acknowledgement, posts to community sites, or periodic newsletters, letting customers know that the company is listening to them and sharing details on their follow-on actions builds trust and confidence in the brand. Publishing metrics regarding volume of feedback activity, hot topics of discussion, and sentiment leads helps build community size and satisfaction by enabling participants to better understand the current status of the brand and context for their individual participation.
4. Categorize
Once a company has generated a healthy and ongoing stream of customer comments from direct and indirect channels, it must categorize all incoming data into company and industry-specific topic areas that can be used to develop analysis, correlations, and other insight metrics to improve business performance. Customer comments, which include both structured and free-form data, must be categorized quickly and accurately in order to deliver actionable value.
4.1. Define categories and category groups that reflect your company’s brand attributes.
In order to gain insight from customer comments, companies must develop categories that reflect the brand attributes that they would like to see reflected in their customers’ voices. Each piece of verbatim feedback can then be “scored” against one or more of these attributes or categories. The resulting categorization lends structure to the free-form comments being received. For example, a computer manufacturer that prides itself on reliable and innovative products delivered at value prices, the top brand attributes they would want to track might include hardware design, performance, customer service, price, and software compatibility. Similarly, a luxury hotel chain would want to track their brand promise in terms of customer perceptions of their facilities, service, room quality, and other related attributes reflective of its premium brand.
Table 2: Listening Categories / Brand Attributes by Vertical Industry
|
High Tech |
Software |
Etail |
Telecom |
Travel |
|
Price |
Price
|
Navigation |
Installation |
Reservations |
4.2. Implement the right tools and technology to automate categorization.
The deployment of tools and technologies that help automate categorization increase the effectiveness of listening programs by replacing inaccurate manual processes. The sheer volume of customer records collected by companies cannot be accurately and cost-effectively categorized via manual process. Moreover, manual human efforts are prone to subjective interpretation and do not easily accommodate scoring across more than one topic area. Finally, the sheer cost of manual classification quickly becomes prohibitive when a company achieves meaningful volumes of feedback. One study revealed that manual categorization costs about $1 per record processed, before any analysis is actually performed (e.g., it takes about 4-5 minutes for a customer service representative to read a paragraph of unstructured text an categorize it into the appropriate topic areas using existing CRM systems; at a cost of $15 per hour, processing costs can rise rapidly).
Many listening initiatives rely on ad-hoc analysis via the use of surveys, keywords, reading of sample of website posts, and using Excel spreadsheets and statistics pulled from a web analytics tool. These approaches similarly do not scale to the needs of customer-focused companies. They take too long and fail to deliver on the hard data required to justify rapid interventions and investment in an increasingly constrained economy. The use of automated categorization tools that leverage text analytics and natural language processing (NLP) technologies makes it possible to examine large volumes of data with increased accuracy.
Sophisticated tools can automate the categorization process and generate actionable insights from massive amounts of textual-based data using statistical algorithms and linguistic techniques. The best systems for automating classification are optimized for real time, and automatically handle generic text, variable grammar, poorly formed sentences, misspelling, and other characteristics of unstructured customer feedback.
Moreover, robust categorization engines are able to indentify relevant themes in text via clustering of words and concepts that frequently appear together in body of documents. This real-time clustering can provide valuable insights on topics that a company might not be explicitly monitoring, and can also facilitate the detection of early warnings. In one example, upon the launch of its LiveMail product, Microsoft detected a large percentage of direct feedback that characterized by clusters of text containing words like “preview pane” and “sizing.” By looking into these verbatim records, the developer was able to detect a display error in the window sizing functionality and was able to immediately publish a bug fix before more customers were affected.
4.3. Categorize comments in real time to identify emerging issues and sentiment.
The success of integrated listening depends upon the speed and accuracy at which a company is able to handle all incoming streams of customer comments. Continuous and immediate processing of every verbatim record helps company identify emerging issues and sentiment, and provides valuable insights within an actionable time horizon. For example, a restaurant company that is able to immediately identify a majority of its incoming feedback as being food quality-related may be able to identify a favorite dish to include in a promotion, or a problematic supplier or sub-standard health practices before the problem spreads and becomes pervasive. In this way, real-time categorization provides considerable value by enabling companies to act on issues that may significantly impact their brand in a time horizon that improves business performance.
5. Analyze
The effective and rapid analysis of customer comments helps companies assess the gap between customer expectations and their company’s brand promise. Examining customer comments in this way helps answer a multitude of questions, such as "Is the issue we've uncovered isolated or systemic?" And "Where in our organization can we best deal with this situation?" Key strategies for effective analysis include:
5.1. Provide systematic structure for analysis efforts.
Analysis of community activity is an intelligence task whose results are generally unpredictable. However, once a train of analysis has proven successful in creating valuable insights, this process can be automated. This includes the definition of key reports and metrics that examine category volume, changes, trends, correlations, etc., and will help inform business decisions. An analytic application can then enables repeatability and predictability via automation (see next section).
Table 3: Analytics Frameworks for Customer Comments
|
Area of Inquiry |
Sample Questions for Analysis |
|
For a particular category or group of categories: |
|
|
For all categories or groups of categories under scrutiny |
|
|
Are there anomalies in the feedback data for a category? |
|
|
How are customers feeling about a product or service? |
|
|
How many comments are being categorized? |
|
|
Who are the influencers? |
|
5.2. Utilize business intelligence capabilities to automate analysis.
Complex data analysis of multiple topics, communities, influential voices and moods in a dynamic environment may not be possible to perform manually, even with large numbers of resources thrown at the problem. Leveraging technologies designed to extract, sort, summarize, and present selected data can help automate the creation of insight reports based on categorization data marts. These reports can include valuable category volume metrics, hot topics, conversation buzz, community sentiment, influencer reporting , and other data views that provide actionable business insights that humans cannot attempt unassisted. The departments who take their cues from the feedback need to have a quantifiable structure to evaluate the feedback and give them the mandate to act – use BI tools to deliver those insights.
5.3. Create a central repository of all customer verbatim records for every channel.
Ready access to customer verbatim comments provides the necessary color and detail for assessing brand performance and analyzing the root cause of any issues identified. Maintaining a central store of all customer conversations facilitates queries and analyses, giving companies the ability to gather sample verbatim across any topic area at any time. In order to maximize system performance and time to insight, lengthy and unstructured data records should be kept separate from the meta-data characterizing these entries. On a minimum, a central repository containing 25 months of data should be maintained to enable year-on-year analysis. For selected industries with more significant compliance requirements (e.g., banking and health care), longer storage terms may be required.
6. Act
Given the ability to collect and categorize customer comments in real time, companies can deliver an appropriate response based on the topic area, sentiment, and any other attribute in order to bridge expectation gap between their brand promise and what their customers actually experience. Learnings from customer insights can similarly initiate processes for making changes throughout the organization. The following are some best practices for initiating actions based on customer listening activities:
6.1. Dynamically respond to direct feedback.
With the use of business rule automation, companies can instantaneously acknowledge comments, or provide dynamic offers to sell additional services, products based upon the categorization value of verbatim records. Moreover, attrition risks and major customer service issues can be identified in real time and remediated with retention offers or by routing the comment to a customer advocate for immediate follow-up. For example, a business traveler submitting direct feedback to an airline about the poor quality of service he received on his trip can be instantly presented with an acknowledgment and travel discount voucher. For a traveler identified as a premium customer, the comments could be routed directly to a service manager who can follow up directly with the customer to address issues directly.
Furthermore, companies can leverage automated categorization and business rules to obtain more detail on a specific customer issue or concern. For example, if a customer shares their dissatisfaction with a company’s shipping service, the company can immediately respond with a browser window or email, asking for additional information and offering free expedited shipping on their next order, turning a potentially lost customer into a net promoter. Dynamic surveys of this kind produce a better understanding of root cause and engage the customer in a two-way exchange. Dynamic surveys enabled through the use of automated categorization and business rules change the approach from asking then listening, to listening, then asking – a true customer-oriented approach.
6.2. Participate in social media conversations.
Armed with the major topics of conversation and community sentiment expressed in social media conversations, companies are able to participate in social media exchanges and share the actions they are taking to address customer concerns and/or present unique promotions and offers. Exposing this information to the community in an abbreviated fashion helps build the confidence and trust in a brand that listens to its customers.
6.3. Engage influencers.
Once companies understand who the key influencers are in a community and what topics they write about, they can approach them to be part of the topic creation or brand values that you hope to spread among the community. Encourage influencers to start conversations and, with the enterprise pointing to them as the experts, let them lead the topic — thereby engaging other members. Anoint key community evangelists to take part in planning, decision-making, and spreading the word to others. For example, Constant Contact is a private "embassy" for 30 of its influential members in a community of 14,000. Rather than directly pay them, the company rewarded them with exclusive access to new information, publicly recognized them, and solicited their feedback.
7. Monitor and Improve
As with any well-run corporate program, each phase of the listening process needs automatic feedback loops that track effectiveness and results. The listening team should monitor all feedback in order to: 1) ensure the volume and quality of customer comments collected remains consistently good; 2) categorization criteria are relevant and accurate; 3) analysis are delivering actionable business insights, ad 4) engagement actions are producing the desired business results.
8. Conclusion
Customers are talking about brands, products, and services in a multitude of places. These customer conversations are unstructured and idiosyncratic, and often outside the control of the enterprise. Strangely, few companies are even listening to this valuable source of feedback – fewer still have strategies and programs in place for collecting, categorizing, analyzing and acting on this rich source of insight about their brand performance. This paper has outlined listening best practices enabled by sophisticated software and organizational expertise. Companies that continue to ignore the new best practices of integrated listening, who continue to ignore the customer conversation, will be left behind and miss the power of truly engaging with their customers’ brand experience.
End Notes
Forrester Research, 2008
IDC Product Marketing Research, 2008
Database Marketing Institute, 2009
Quirks Market Research Review, 2008
IDC Product Marketing Research
www.dellideastorm.com, www.xmfan.com, www.mycokerewards.com
Li and Bernoff, Groundswell 2008
www.mvp.support.microsoft.com
Island Data Client Services
IDC Product Marketing Research, 2008
Microsoft IRT Case Study, 2008
IDC Product Marketing Research, 2008
“Customer Feedback, Are You Putting to Use?,” MyCustomer.com 2008
“Distributed Influence: Quantifying the Impact of Social Media” Edelman 2008 Forrester Research, 2008
