Actionable analytics: Why you can't separate customer behaviour and attitudes
What separates a successful product from one that isn’t? The answer lies in truly understanding what customers want. It’s what makes brands like Apple the huge success story they are and why so many others have ended up as nothing more than ‘rotten apples’. Customer insights are fundamental in defining what a product or service should look like, as well ensuring sales and marketing teams can accurately focus on defining what the three other components of the marketing mix should be – selling at the right price, place and using the right promotion.
Traditionally, surveys provided the main mechanism for generating customer insights. But thanks to the advent of digital media, the depth and scope of information that can now be obtained has grown immeasurably. Digital media has opened the door to understanding customer behaviour at a very granular level. This has augmented the traditional survey approach by adding in behavioural data that can now provide a much bigger picture of who your customer is – and crucially what they want.
Marketers, researchers and analysts have started to combine data on both behaviour and attitude, progressing beyond the previous debate on what came first – attitude or behaviour? Behavioural insights now gathered are not only incredibly detailed and accurate, but also come in real-time. Attitudinal insights, on the other hand, can be measured against business KPIs. The result is vastly greater and insights beneficiaries. But for it to truly have a positive influence over decision-making at an individual customer or segment level, it requires an integrated approach.
Integration: why it’s essential
Thanks to recent innovations in technology and the changing tools available to analysts, marketers and the like, achieving the goal of complete customer understanding is now easier, and here’s why:
- Consumers: They leave an extensive digital footprint, and don’t shy away from sharing their behavioural and social activities.
- Analytics: This is increasingly becoming the core to decision-making in business and the need for more data sources is soaring.
- Culture: Organisations are changing; encouraging experimentation and data discovery.
- Technology: The most obvious one here, and with advancements, technology is helping to capture various data signals in a more cost-effective manner.
Below you’ll see that the traditional view of the customer is limited – the ‘who’ and ‘why’ questions reveal only part of their interactive experience with an organisation. To achieve the 360˚ view, it requires the integration of behavioural data (interactions and usage logs) to also provide answers to the ‘how’ and ‘what’ questions.
Why isn’t everyone doing this already?
Sometimes gaining solid insights from big data is as difficult as trying to find a needle in a haystack. The task of combining behavioural and attitudinal data provides some real challenges.
The first obstacle is processing big data. Behavioural data representing a large sample size of the population will usually consist of petabytes of data. Data with such volume requires looking far beyond traditional data management sources to handle, process and analyse. The solution would be to use big data platforms such as Hadoop.
The next complication is business consumption. Despite academic ideas offering good solutions, it’s important to remember that the main reason for such an exercise is to make it easy to implement and business-ready when it’s time to take action and consume insights. The hardest part is convincing businesses of the validity of this approach and actually being able to explain in clear, comprehensible terms the benefits through impact (quantified) and findings. The resolution is to set up a channel for frequent and easily comprehensible insights.
Lastly, you need to ask the right questions. If the problem requires understanding the scope of the issue, the typical hypothesis-driven approach might not be appropriate, as the right framing of questions might not be possible.
How to: the steps to integration
Conventionally, a lack of unstructured data is a deal breaker in the hypothesis-driven approach to customer understanding, while having a ‘wealthy’ amount can be a boost for a discovery-driven approach. Thus, we face a predicament in using a discovery-driven approach against the hypothesis-driven approach. It is crucial, therefore, that when solving these unchartered and innovate analytical business problems organisations will need to adopt the right approach to problem solving.
As you can see from the figure above, having the right skillsets and datasets to process data is quite straightforward, but what some forget to consider, is the importance of having the right mind-set required to solve a problem efficiently.
An illustration of combining data sets
Let’s imagine that a products-based company is interested in understanding their customer’s usage pattern along with sentiments shown towards their products. The integration of behaviour and attitude is what they would need to create such a program. The outcome would better enable the company to design new features and improve targeting strategies.
A closed-loop system (shown below) would be required to make the integration, analysis and incorporation of behavioural and attitudinal components a success (people, process and technology).
Machine log data and survey data, both from the same set of respondents would need to be merged here. Typically the machine log data might make up a large chunk of the analysis (becoming terabytes of data). This could lead to a big data challenge.
Integrating the disparate data sources are at the core of this data exercise. Both the data from survey respondents and machine users need a common identifier as a way to create a bridge between the multiple inputs. The program would struggle to function without a common element, so it’s critical to make sure one exists.
Once you’ve overcome this hurdle, the real logistics challenge starts: integration. Though new data paradigms are unconventional, they work very efficiently, so big data constructs such as Hadoop, Hive and R are needed for good support.
During the program, the following questions might need to be asked:
- What is the relationship between user activity, user profile and attitudes?
- How does use and perception of competitor products impact product usage and attitudes?
These questions would work to establish a link between perception and behaviour between the two different sources. We must also consider the notion that correlation does not mean causation, so when relationships are recognised, they need to be both causal and correlated in nature.
To put this in to practice, organisations need to consider the following:
- Clarity is key – so make sure your business objectives have been defined in advance.
- Estimate infrastructure needs, especially for big data – this includes the capacity to capture, store and process.
- Have you got the right people and partners in place? Remember that the mindset aspect is critical to this initiative.
- Accurate communication means accurate consumption. Consistent and engaging communication with your audience helps to overcome any cognitive biases that might exist.
The beauty of this integration program is that we no longer have to rely on surveys to understand customers’ behaviour and attitude. Ultimately, what has been established is that customer usage actually influences perceptions, which eventually turn into actions through recommendations and repurchases.
What we’ve learned, and now advocate, is that organisations that are rightly turning to data as a means to help drive better business decisions, shouldn’t look at customer behaviours and attitudes from different angles – they need to ensure the two mix. The sooner organisations understand how to integrate these, the closer they’ll come to influencing usage and brand perception of their customers.
Kshira Sagaar is manager and Deepinder Dhingra, head of products and strategy at Mu Sigma.