The importance of VoC analytics

This isn’t a theoretical board room exercise: improving revenue means developing better customer experience which itself requires close listening to the customers themselves. In short, listening to the customer pays for itself. It delivers:

• Better customer experience
• Better products and services 
• Improved decision making 
• Improved operational process
• Growth top-line revenue

Before you start, you’ll need to know what precisely you need to answer. This may sound obvious, but the questions you ask now will determine the nature of the data you collect, the type of analysis you use, and the kind of tools you will need. As a tip, these should be open questions, beginning with the words ‘How…’, ‘What…’ or ‘Why…’ rather than the question which can be answered with a ‘Yes’, ‘No’ or a single number. 

So good questions will look like this: Why are our sales down in Germany? What is wrong with our high end product? How do we sell more insurance add-ons in France?

You want questions which will themselves help answer the core question: what do I need to know that will help me improve our products and services?

The next step is to start gathering the information to answer your questions. For this you will need to look at every data source available. Examples might include:

• Online feedback from Google, Amazon or app stores
• Contact centre logs
• Online surveys
• Social media comments and interaction
• Emails
• Chatbot transcripts
• Sales data
• CRM logs

The process of preparing data is also important. As most will have been assembled from different databases there is a considerable task involved in cleaning, deduping and unifying everything the organisation has into a ‘single view’ of the customer. 

It’s likely that if you are an organisation of some size that at this stage you could have a huge volume of both structured and unstructured data. Some will be checkboxes on surveys but some will be text entirely written by the customer with no guidance. Clearly, just using a spreadsheet will not be enough to find trends or anomalies, let alone answer your core questions.

Although it’s possible to run VoC analytics manually or even with standard data warehouse tools, it’s awkward and very time consuming.This is where tools like Wonderflow come in. AI, harnessing the power of Natural Language Processing (NLP) enables analysis of both structured and unstructured data. Wonderflow’s NLP engine translates customer language (sometimes literally, if it’s a multilingual environment) into a business language, relevant to the organisation. Rules are developed to perform a linguistic analysis, in which humans teach the machine to find and categorize mentions into topics.

As an example your question may have been “why don’t customers in Spain like our new car model?”

If you have performed the above steps you will have collected local Spanish data from Facebook and Twitter, perhaps even Spanish review websites or CRM data from dealers. Because you want to see how potential customers feel you’ll be interested in clustering responses around specific themes, topic expressed and sentiments relating to the products.

You may also be interested in running reports based on intention. So for our example you may need to see sentiment scores against the topic “comfort” or “storage space” or even mapped against a competitor. The right tool will be able to cluster effectively and help identify the areas which you need to look at.

Sentiment analysis is an automated process that uses AI to identify positive, negative and neutral opinions from text. Usually, besides identifying the opinion, these systems extract attributes of the expression in order to make the sentiment meaningful and establish context.. Intention analysis looks for action behind the sentiment for example a tweet “I waited 30 minutes for a salesman!” tells you the customer needed something specific to happen which wasn’t delivered.

Sometimes insights will be obvious, based on word frequency or location or channel. Frequently you might have to do finer analysis like aggregating sentiments into clusters and further analysis to see how they interrelate and if the results are actionable. For example in the car example above, frequent clusters based around seats, position, or visibility may indicate a physical product issue. 

All your analyses should be bought together into a visualisation tool which not only integrates into normal business reporting but also allows ad-hoc analysis by being able to zoom in and out of the data or aggregate it in different ways.

Many tools aren’t flexible enough to do this, often requiring complex ticket systems when requests are submitted to a data science team. Wonderflow’s Wonderboard makes a point of being able to be set up quickly, with a setup time up to 80% shorter than the market average.

 There are essentially only three types of result from any kind of analytics program:

Yes, sometimes the data will tell you to do nothing, or at least that there is nothing worth doing. This isn’t a waste of resource as it may have validated some of your initial assumptions in your original list of questions. For example you may have assumed that engine power was an issue for your customers but if the analysis does not mention this, it may not be worth doing anything.

Beware though as “do nothing” is the default setting for many organizations so always seek validation.

This can be the most exciting result. For example you may find that Spanish drivers are used to a particular feature on a car which you haven’t developed yet. So, this means you urgently need to engage design teams. Or, you might find that dealers are not well placed for your target customer group which leads to examining distribution strategy.

This can be the most exciting result. For example you may find that Spanish drivers are used to a particular feature on a car which you haven’t developed yet. So, this means you urgently need to engage design teams. Or, you might find that dealers are not well placed for your target customer group which leads to examining distribution strategy.

VoC analysis can help understand if your roadmaps and strategies are working or not. For example if last year’s action plan required a re-examination of the supply chain to Spain, it can help understand if this is working after taking measures to improve the situation.

The whole ethos behind understanding VoC is to drive change throughout your organisation. There are a number of ‘quick wins’ where the initial insights can make an impact quickly:

This is the clearest opportunity to apply learnings. Applying VoC alaystsics lets you understand how your customers feel at different touch points and helps map the overall customer journey. The aim is to affect not just your NPS score but the customer’s overall loyalty

VoC allows you to delve deeper into the reasons why customers feel the way they do, especially to understand particular product features which may be faulty or missing. It also helps understand price sensitivity behind particular features or even when certain features may be redundant. Hence creating a VoC program plugs directly into your product development roadmap. 

VoC analytics can be highly successful when applied to social media, for example collecting and analysing tweets and social media comments to build a better picture of how your brand stands.

It’s easily possible to use VoC analytics to check for competitor brand comparisons or mentions in your reviews and social media. It’s also just as possible to run the same analysis the other way, checking your competitors for success of their products and mentions of your own brand.

A VoC analysis program is not a theoretical or stand alone effort. It needs its own roadmap and needs to connect to wider company aims. This includes not just overall strategy but also product, marketing and technology roadmaps. Like any other roadmap or plan, VoC needs an owner who is also a point of accountability. It’s this person who needs to take VoC questions and aspirations and boil them down into actual attainable goals and KPIs

The results of your program should not be held on a need-to-know basis or on a password protected file on your desktop. The insights need to be rolled out to the right people. Even if one person owns the goals and KPIs, it will be up to others to make actions meaningful. And this means more than a quarterly presentation. Here at Wonderflow we make a point of ensuring the Wonderboard is as easy to understand as possible and that training is rolled out to as many people as possible. Our innovative pricing model ensures this doesn’t penalize the company!

VoC analytics is about how you act. There is no point in listening unless you do so regularly and then you need to be seen to act upon what you have heard.This means making hard links between your VoC program and your business context. At this stage you will need to integrate your learnings with your company’s project management methodology and reporting systems so you can act on the change. It’s at this stage many find the company isn’t set up to learn from VoC and needs better tools or working practices to be more responsive.

We have left this last as it’s the most difficult to change. Are you flexible enough to meet the needs of your customers over time? Are you truly customer centric? 

Throughout our experience we’ve found that many companies have realized that they need to change not just systems and processes but sometimes the entire company culture from sales teams to board level. Truly listening to the Voice of the Customer often means cultural change in every team and at all levels. It also means being able to map out the influence of everyone’s role on the customer journey. 

It’s for this reason the Wonderboard is designed to be used by anyone in the company, and not just by researchers and scientists. In fact the adoption rate of the Wonderboard is 1000% higher than market average, with users that vary from interns to CEOs of Fortune 500.