July 22, 2017

Uncover true value in your hidden data

Summer cleaning in the corporate basement

It’s that time of the year again: most people are off for the Summer holidays, and teams are running on low capacity since customers have become surprisingly silent… Blue sky, waiting for the September thunderstorm to arrive, when many marketing teams will be preparing their 2018 segment plans.
Key question to answer with budgets under pressure: how to create the biggest bang for the buck? What are the leverage points to be successful in an increasingly interconnected market environment?

Instead of buying an expensive sectorial market research to obtain new insights that guide your strategic direction for 2018, have you ever considered taking another perspective? Think about having a look at freely available social media data or even your own operational data store. Challenge yourself to descend into the basement of your corporate information environment. While some pre-processing will be needed, the richness of insights to gather is often enormous. Since it entails the very specific characteristics of your loyal and less loyal customers and how they deal with your product & services portfolio (often a mixture of descriptive and behavioral data), it might lead to more actionable insights than a generic, high-level market research based on declarative statements of (possible) customers. When you’ve pinpointed your best customers, try to classify them into meaningful segments: are they truly ideal or are there opportunities ahead for up- and cross-sell, or are there even options for a whole new business model? If you’re kind of happy about their interactions with your brand, and the business you earn on them, then how and where can you find lookalikes? Algorithms are taking over the world, so grab the opportunity and claim your spot!

In this blog we will describe a couple of use cases about how KWARTS® has improved the marketing & sales effectiveness of its customers, by applying data science techniques. We do hope it will inspire you for a paradigm shift towards more innovative practices in your strategy setting exercise.

Case 1: Pinpoint your most valuable distributors to grow a niche target audience

One of our B2B clients is active in retail via a network of distributors. These distributors manage the relationship with its niche end customers since local presence & network are the main drivers in the company’s business model. Aside from the centrally managed branding & communication materials, they are self-sufficient and manage their own marketing budgets, strategies & tactics.

Forced by a shareholder agreement to grow its end customer base exponentially on a very short term, our client has had a hard time in determining the best go-to-market strategy, given the power relations in its distributors network. Which distributors are doing better, which are doing worse? What is better, what is worse? How to convince independent resellers to adhere to HQ growth ambitions? You can’t command a tree to grow faster.

Historical efforts have led to the conclusion that attracting new, nation-wide distributors is no guarantee for exponential growth in end customers, since the local touch & credibility is lost by increasing the scale of the sales partner, plus the distributor power quickly gets too big for a scale-up to handle. Regional attacks in dense geographical areas didn’t return the expected results neither, since it is very hard to target specifically on the most valuable customers when building up a critical mass. Population density does not equal target group density, so more specific commercial engineering would be needed to address the right customers. And even if we could find the perfect way around, bypassing the distributors via other (online) channels would disregard the aspect of local integration, and potentially cannibalize the distributors’ business model – still the most important channel in the mix that you don’t want to offend too much.

KWARTS® decided to dive into the client’s operational data store and analyzed the transactions history between distributors and end customers, literally gigabytes of data. We conducted a quantitative network analysis, in order to uncover patterns in transactional behavior and interrelations between different actors in the network.

We identified both characteristics inherent to the distributor (let’s call them “properties”), as environmental ones (let’s call them “circumstances”) by explorative network analysis. For example, some distributors appeared to be far more popular than others, since they were visited by many more customers. On the other hand, some customers showed social influencer potential, since they were the unique link to more isolated customers at less popular distributors. They could carry the positive word-of-mouth from more dense target group communities to the more scattered communities. This could benefit the brand reputation and thus lead to extra sales. Via predictive data mining, we identified the discriminative properties of distributors with the most optimal characteristics in terms of popularity and social influencer customer capital. We labeled them as the key aspects of distributors with a high “customer virality”. These insights were turned into tactical guidelines for the client’s key account managers. By pampering & observing these distributors on their marketing strategy & tactics, and by sharing their best practices with sub-performing distributors, the key account managers were able to improve the latter’s performance, and meanwhile grow their end customer base. On the other hand, by defining the right targeting criteria for new distributors, the sales effectiveness has improved dramatically, leading to lower CPA and higher return on marketing investment. To put a long story short: while you can’t command a tree to grow faster, you can create the right environment and cultivate those types of trees that grow the sweetest fruits…

Case 2: Social listening on digital channels to reveal market trends

Many companies are active on social media, and so are their customers. Social listening can therefore be an interesting brand barometer, aside from focus groups and other more traditional market research techniques.

One way to look at it is to detect what people are talking about in the context of your brand and whether your brand is put in a positive or negative perspective, the so-called sentiment analysis. This is done by translating the social media posts containing your brand name, into a “bag of words”, where stopping words have been deleted from and where frequent discriminative terms are being matched with a list of positive and negative sentiments. This leads to a sentiment scoring that enables brands to rank themselves in comparison to competing brands (by applying exactly the same technique for competing brands). Since social media channels are so reactive to the latest word-of-month, brands can be far more proactive in customer care than with traditional research techniques that tend to bring insights many months after the facts.

Another application of social listening we have conducted for a customer at KWARTS® is revealing market trends & events. As an illustration of the value that social media mining can bring for these purposes, we have scraped the 200 most recent Twitter posts containing McDonald’s or Burger King. We applied the “bag of words”-technique described above, and created a word cloud with the 25 most frequent words associated with both brands.

  • An immediate eye catcher is “MTV” right in the center of the word cloud for Burger King (with a few variations such as MTV Bananas, Challenge MTV, Bananas Doing Things, Fandom Awards). On the contrary, for McDonald’s the words are far more conventional and tied to the food business, e.g. “nuggets” (from its “Chicken Nuggets”) is in the center of the cloud, surrounded by “fries”, “chickfila”, “favorite”, “can” and “toy”; all what you would expect. For Burger King we only see “chicken”, “burger” and “dinner” appearing in this context.
  • Three strange words appear in the McDonald’s word cloud as well: “police”, “judgment” and “fired”. This is an interesting finding that might need some further investigation.
  • In the Burger King word cloud, its competitors McDonald’s and Wendy’s are mentioned, which is not the case for the McDonald’s word cloud, so there might be a market dynamic that leads to Burger King being often related to the two other incumbents.

After an hour of online desk research, we summarize a few findings that explain some of the specificities in the above word clouds.

  • Johnny Bananas, an opinion leader working for MTV (and followed among others by content marketing guru Gary Vaynerchuck), has recently put Burger King in the picture; this was liked and shared by many people, creating some social buzz.
  • The MTV Fandom Awards in San Diego (an annual pop culture, television and film awards show) were sponsored by Burger King in July 2016, and since they’re currently in the press for the next edition of July 2017, last year’s action appears to still create some buzz for Burger King today.
  • There has been an action recently on the YouTube channel of Challenge MTV, called “The Burger King of Kings”, where a guy ate as many Burger King dishes as he could, which was followed heavily on social channels.
  • When taking a look at the Burger King Twitter page, we also notice why the word “chicken” is so important in recent tweets: they recently launched a new Chicken Parmesan Sandwich. They got the old crew together and shared some images and videos on social media, which appears to be a huge success in terms of likes and retweets.
  • The McDonald’s word cloud contains the words “police”, “judgement” and “fired”. A quick Google search revealed what happened: a co-worker refused to serve a police man a few days ago, creating quite some negative buzz which could be leveraged by competitors. Indeed, the respective employee was “fired”.

Combining the derived insights from our social media mining, we could conclude that Burger King has had slightly more positive buzz on social media than McDonald’s for this specific timeframe (200 most recent tweets).

More importantly our findings also give indications for different targeting strategies applied by both brands: where McDonald’s has more “mainstream” social media presence, Burger King is clearly more focused on the male youngsters segment, through its affiliation with MTV, YouTube all-you-can-eat experiments, etc. Other sources confirm that Burger King has made a clear strategic choice to put less focus on the female and family customers. McDonald’s on its part is focusing on nutritional or food source information in their content strategy.

However, there is no good or bad strategy. The brand intentions might be different. While market trends evolve towards healthy food & living, Burger King’s strength in social lies in inviting user content, welcoming users to create content as well. In an age when office chatter has moved from last night’s TV episode to the latest viral video, part of Burger King’s marketing advantage has been its willingness to move quickly to exploit a constantly churning Internet news cycle. Burger King likes being edgy and it has proved that it doesn’t mind doing things that might make other brands blush.  

McDonald’s remains the world’s largest restaurant company with roughly five times Burger King’s ad budget, and its sheer size has pushed Burger King to take marketing risks to stay competitive. The budget size advantage doesn’t go as far as it used to because of social media. It is clear that if marketers can utilize social media as part of their marketing plan, it can generate more buzz than paid media.

This perspective fits the vision of Burger King’s new CEO Daniel Schwartz, a 32-years old young potential. He has no restaurant experience but does have the strength of doing things differently, esp. in terms of increasing the share of franchising restaurants compared to company-owned restaurants; bringing huge cost savings. Business literature, such as Bloomberg, state the doubters are in the minority now, and many in the investment community would like McDonald's and Wendy's to mimic the kids at Burger King (hence, explaining why we find these 2 brand names in our word cloud as well).

To put a long story short: as does Burger King, a good marketing practice for your company could be to closely monitor social media via text mining techniques, and use the derived insights as a reading key to identify major trends or events impacting your own or competing brands. While you won’t track all events via social media, you’ll certainly track the ones that resonate most. Consequently, you could use that information to your advantage and adapt your marketing tactics proactively. This is what KWARTS® calls an essential element in any growth hacking strategy. Social moves super fast so it’s all about constantly testing and learning from your own successes or mistakes, or the ones of your competitors. At the end of the day you need to make sure what you’re doing is working and pushing your brand and business forward. It’s a small effort that is certainly worth the sanity check!


The world is undoubtedly connected. You’re part of this world. Your customers are too. It makes sense to dive into the granular elements that constitute it, since on the one hand your business is dependent on it, and on the other hand, these elements are directly or indirectly fed by the operational processes of your business.

Data-driven marketing techniques such as network analysis, or social media mining, are useful to gather insights in the product-market fit of your offering, in your brand reputation, and in general market trends affecting all players, or only some of them due to specific events that are happening. Some of these insights might drive the way you define your marketing strategy. Others might have a direct tactical impact.

There are plenty of other data mining techniques that you could apply right away on your operational datastore or online channels, without the need to purchase expensive external data. Think about process mining on your customer journey based on ticketing & email information right in your operational systems: how many steps does it take before a customer’s issue is resolved? What are the bottlenecks in the process flow? When we  visualize these flows, our clients are often surprised about how much reality deviates from the slide decks in the customer experience plan.