Posted by Jeanne Jones
In a recent HGS webinar entitled “Rewire the Digital CX to Create a WOW Customer Experience!” participants were presented with the following scenario:
You have a (fictional) 10 percent extra budget to spend on customer experience (with no strings attached). Where do you spend it?
43 percent of respondents, the highest of any category, indicated that they would spend their dollars investing in analytics.
In reality, most of us will never get the extra 10 percent budget to use however we want. Instead, we have to rely on cold, hard ROI proposals to build our case for funding. And when it comes to justifying the investment in analytics, particularly for a customer service organization that has been traditionally seen as a cost center, this can be a huge challenge.
Voice of the consumer analytics ROI is typically centered on the ability to deliver actionable insights that will lead to positive outcomes such as new product innovations, incremental revenue, brand building, and loyalty. Standard industry practice focuses on normalizing data to arrive at contacts per million units (normalized rates) by product category and subject type, and then stratifying top contact types classified as complaints, compliments, inquiries, and suggestions.
But what if you could offer your organization insights that had never been tapped into before, and mitigate risk (avoid costs) while you’re at it? What if you could deliver, for the first time, a very different voice of the consumer that had never been heard before? Let’s call it a “new voice.”
It starts with simply doing things differently with VOC data.
- First and foremost, establish a culture of analytics. The front line should be well versed on how the information they are capturing as “voice of the consumer” is funneled through a closed loop process that leads to improvement and analysis that ultimately contributes to risk mitigation.
- Apply innovation to VOC data analytics with the same passion and tenacity as you would when developing new product offerings.
- Consider non-traditional approaches.
- Ask “what if”?
- Have no limits when exploring correlations.
Here are 10 techniques to help uncover the voice of your consumer:
- Apply a difference algorithm. Take the time to compare data sets to identify where there are measureable the differences between actual and expected ratios.
- Measure sentiment and demeanor. This is important to look at when determining how to allocate resources / capital for quality improvements. Not all complaints are created equal. This helps identify low rate issues that may be having a significant negative impact to the brand.
- Analyze key dis-satisfiers as an indicator of re-purchase intent. Similar to sentiment, this is used to look at specific subject code types at all levels (brand, product group, UPC) and provide quality teams feedback on top issues that are driving dissatisfaction. The consumer is asked, "Will you purchase the product again?" Data can be collected as yes/no/maybe and it allows for a simple way to look at top issues for which consumers indicate they will not re-purchase.
- Leverage probing questions and data collection to compare early performance trends with past product performance for new product launches. This typically measures using a comparison of contacts per million units (CPMU) against an established baseline from a similar product or past product launches.
Typically new product introductions and product changes will always demonstrate an early increase in complaints that tend to level off after 90 to 180 days depending on the product lifecycle. This is an early warning system that allows the product developers/packaging engineers/scientists to quickly respond to issues, and when needed, add probing questions to the call flow to collect the needed information to quickly identify a root cause so it can be addressed. When a response to an issue is addressed in a timely manner, a best practice is to reach out to the consumers who had called in and let them know about the changes that were made as a result of their feedback and offer an additional coupon or sample to win them back.
- Deploy early warning systems. There are many ways to leverage statistical process control (SPC) to run data daily to look for notable increases. A best practice is to run the data at the most granular level possible on an absolute basis (looking at EVERY UPC and EVERY subject code) to identify an increase or decrease week over week, month over month, and year over year.
- Empower reps to flag issues for review. Empowering the reps to flag cases for further analysis is important. There are times when serious issues may "get lost" under a subject code heading that does not adequately describe a potential risk of illness, injury, property damage, etc. Back-end analysis is needed to review verbatim comments and determine if there is a potential risk or high negative sentiment issue.
- Leverage the “multiplier” impact in the normalization process. For every one consumer who calls to tell you about an issue or share their opinion, there are many consumers who also had the same experience/opinion and did NOT share their feedback. This is referred to as the multiplier. It’s important to establish a baseline by subject type that can be used in analysis to gauge the impact of the “unheard voice.” The below illustrates how a complaint type with a high multiplier can “go under the radar” if the multiplier impact is not considered, and also how a product issue may be unnecessarily consuming valuable resources.
- Use multivariate analysis to determine product or category health. Apply weightings based on complexity factors in the product (such as storage temperature, preparation steps, and shelf-life), consumer sentiment scores, margin, and repurchase intent to arrive at a “big picture” product or category health rating.
- Discover unique patterns versus comparing traditional data sets. This requires a unique algorithm in some of the more sophisticated data analysis software products, but is a valuable technique if the tool set is available. Rather than assigning data sets and analyzing, seek out unique patterns in both structured and unstructured data to review.
- Report events at the plant level. An important technique to leverage for risk mitigation is implementing thresholds at a manufacturing plant, supplier, or restaurant level for sensitive subject code types and/or combinations of subject code types that may be an indicator of risk. If at any time the total number of complaints in a given time period is exceeded, a deep-dive analysis is performed. This is important because there may have been an "event" on a given day that caused a series of issues that may span multiple subject codes. Any one given subject code may not trigger a review, but when subject codes with a common root cause are combined, the issue becomes visible to the analyst.
For more detail on this subject, attend the next installment of our 2016 webinar series, “The Evolving Role of Customer Care Analytics, ” at 1 PM on Wednesday, February 10. Delivered by Jeanne, attendees will learn how customer care analytics can support a wide range of cross-functional departments including product quality and safety, manufacturing, risk management, product development, and even HR. Discover why a “culture of analytics” can play a critical role in translating data into insights.