VoC Data Sources and How to Use Them
With customers being more vocal than ever about their lifestyle and choices, the sources for Voice of the Customer (VoC) data collection have evolved. Let’s dig deeper into these sources to develop a better strategy for capturing the Voice of the Customer. Knowing each source well is imperative as only then can you choose the right type of data source to harness the insights that you are looking for.
Types Of Data Sources
There are three types of data sources - Direct, Indirect, and inferred. Within these three main branches lie eight top-tier sources of customer data, which we will discuss in detail shortly.
- Direct Sources for Capturing Voice of the Customer
Direct sources where the customer directly interacts with you, the provider, either on the phone, by email or text chat, or by responding to surveys. We will examine these direct Voice of the Customer data collection sources - Surveys; Call Centers; Emails and text.
- Indirect Sources for Capturing Voice of the Customer
Indirect sources are still viewpoints expressed by individual customers, but they appear on third-party forums, review sites, and consumer pages. These will tend to suffer from the same sorts of biases as direct sources but may prove a little more trustworthy since users are less likely to pull their punches when they aren’t talking directly to a service provider. These sources are - Social Media; Review Platforms; and News.
- Inferred Sources for Capturing Voice of the Customer
Lastly, and more ambiguously, you can infer certain sentiments from customer behavior, based on the data extractable from your own sites and databases, as well as those of your competitors. These are - Website use; and Purchase History.
Discover more: Voice of The Customer Process.
Which Are the Top Voice of Customer Data Collection Sources?
Under the umbrella of the three types of data sources - direct, indirect, and inferred - lie eight subdivisions of data sources that are considered the best for collecting VoC. These are surveys, social media comments, reviews, customer care emails and call logs, news monitoring, website use, and purchase history.
You’ll have seen the pop-up boxes that ask you one or two key questions about a product. They can be irritating, but sometimes it’s just as easy to complete them as to look for the cross hidden in the top right-hand corner. As well as interactive pop-ups, there are surveys sent by email, surveys that appear after a purchase and optional questionnaires that release a discount code.
Regardless of how brands motivate people to complete online surveys, it’s still true that these tend to be completed by a self-selecting group of respondents – those with a grievance, and those who love a product or service. It’s worth bearing this in mind when you consider this data source for collecting the voice of the customer.
Harvard Business School recommends conducting face-to-face interviews and focus groups where possible, to counter this self-selection bias. Although this is more time-consuming, it helps counter self-selection bias, they argue.
Learn more about the voice of customer surveys
- Call Centers
“This call may be recorded for training or service improvement purposes”. The mandatory warning is there for a reason. You’ll get heartfelt insight, particularly into service gaps and complaints, by performing semantic search and sentiment analysis on the transcripts of these calls.
Although it is true that calls can also be self-selecting (few customers phone support to say the service is acceptable) they do offer unfiltered insights straight from the horse’s mouth. However, it’s a little more difficult to transpose this into actionable data than survey information, which lends itself to metric conversion.
See how it works with our call center sentiment analysis case study
- Emails or Text Chat
However, text analytics APIs still must deal with abbreviations, grammatical errors, slang, typos and more when searching these sources for actionable information. These interactions also suffer from the same self-selection bias as other direct interactions.
- Social Media
Social media listening is fast becoming a go-to choice for garnering real insight. On social media, people feel free to express their honest opinions, albeit in the heat of the moment. You may have to correct for an impulsive response a little, given how delighted some online influencers can be at an unboxing, or infuriated with a product that’s not up to spec.
To do social media listening properly, you must examine a lot of different types of input – text, audio, hashtags, text-on-screen, and even facial expressions. Fortunately, companies such as Repustate offer AI-driven sentiment analysis APIs to perform these queries for you. These tools can even search inside video content to extract vital VoC data.
- User Review Platforms
One source of VoC data that’s quite often simple to render into metrics is the user reviews that appear on Amazon, or Trustpilot, together with optional product reviews. Star ratings are easy to interpret, of course, since they already represent numerical values.
Our review sentiment analysis can help sift through the high volume of reviews on these sites and deliver a convincing summary of what’s being said. These sites can be invaluable in understanding how purchasers of your product react. The only downside is that, once again, a certain type of personality tends to offer online reviews. Casual customers may be harder to track this way.
- News Monitoring
News columns and PR mentions in consumer sites can be tracked the same way as user reviews, although they may not come at scale, or with handy star rating appended. They will, however, tend to be written in a more grammatical, typo-free manner that’s easier to run a sentiment analysis API over. Make sure you survey sufficient sites because some smaller platforms use only one or two reviewers and you wouldn’t want to be reporting the opinions of the same two journalists, after all.
We’ve written a whole guide to news sentiment analysis you can check out.
- Website Use
You’ll be able to gather a lot more information than you might imagine, even from customers who don’t make a purchase or interact verbally. Time on site is a good indicator of interest, and returning customers are evidently intrigued by what they find. It’s possible to interrogate your CRM to observe consumer behavior as leads move from casual visitors to consideration and commitment. Talk to your sales and marketing staff about the strategies they have adopted to retain or convert customers.
Product pages whose visitors far outstrip their sales, compared to other pages, may not be offering what customers want. Perhaps the price point is wrong or you’re not offering the right options. You can also glean a lot from which sites visitors left to locate yours. Were they on a competitor site or did some piece of marketing drive them to your home page?
- Purchase History
Lastly, as Amazon knows all too well, your purchase history provides clear predictions of the kinds of affiliated products you may be looking for. Purchase a camping stove and hiking boots and you may well find yourself presented with hillwalking guides and insect repellant!
It’s also about what price you paid, as well as when you bought the item. VoC data collected from aggregated sales can help target marketing campaigns as well as suggesting adjacent product lines you might develop.
AI-Driven Customer Analytics with Repustate
Above, we’ve provided a rundown of the top sources of data and information that will inform your Voice of the Customer data collection strategy - from sentiment identification to video content analysis. With an AI-powered text analytics and sentiment analysis platform like Repustate’s, you can capture the Voice of the Customer and create a composite customer portrait while identifying sticking points, trends, and opportunities. No longer do you need to be wracked with high costs and issues that arise from the manual processing of customer data or from inaccurate sentiment analysis ools.
The Repustate engine is a scalable sentiment analysis solution that is powered by native language NLP. This means you can harness customer sentiment regardless of the language your customers write reviews in or post user-generated videos in the form of vlogs or YouTube reviews.