Every software product has a specific value proposition. For instance, a mobile app or a product creates value for users on the go by providing them with specific content that can be displayed on a small screen despite intermittent Internet connectivity. Similarly, an analytics-based product creates value by solving a business problem using meaningful insights from big data. However, there are risks involved in the development process, which prevent the product from delivering to its potential. An analytics product could overcomplicate a business problem in delivering a solution. In some cases, it may not successfully address the needs of the customer because of the lack of industry-preparedness to use it.
In this article, let’s delve into five tenets for developing a successful analytics product with minimum risks.
Understanding the market
It is necessary to identify the primary users of the product and evaluate their skillsets to assess market readiness vis-a-vis people, processes and tools. Let’s take the example of the emerging multiplex industry in India. To build an analytics product for performing business tasks, such as calculating the average occupancy or revenue forecast of a cinema hall for the next three months, we must review the current business systems in use and the skillsets of those using it. It will help in understanding the maturity of the market.
Now let’s assume that the users are developing reports after the descriptive analytics or diagnostic analytics and seeking answers as to why the occupancy and revenue in the previous month spiralled downward compared to the previous year. The next logical progression would be to use predictive analytics to calculate the occupancy or revenue forecast for the next three months. It would be overambitious to directly transition to prescriptive analytics and develop an analytical solution to forecast occupancy and revenue for the next three months bypassing the earlier stages. The users and the industry, at large, need to move through each stage and develop people, processes and systems to get ready for the next step.
Acquiring the right data
Data relevance is of utmost importance. In a business process, it’s difficult to generate clean data from the manual entries in a ledger. Therefore, it is critical to set up business systems that can yield clean data from the user. A multiplex, for instance, should be able to understand the source of the bookings, whether online or offline, the number of hours prior to which an advanced ticket is booked, the amount of discount offered at the time of booking, etc. It’s only possible to collect such data elements if the industry uses automated ticketing systems with the provision to export the data. Another important aspect is the availability of meaningful patterns in the data. The data must be captured at the right level to extract meaningful insights through analytical processes.
Avoiding complexity
The real world is full of complexities such as the number of variables impacting a business outcome. For a movie to succeed, you can think of several variables that could determine its success in the box office. It is not possible to collect data related to each variable. As analytical experts, we all like to model every corner condition using every possible data element in our algorithms. We must fight the battle between possibilities and the probabilities. If we account for each of the possibilities using potential variables, we will not only consume an enormous amount of data but will also make our models too complex for the users to understand. Strike the right balance and account for achieving the most probable scenarios in the product. For example, it is possible that a regional film release might have an impact on the success of a national movie in the first week. However, including potential releases of all the regional movies into the models and their probability of success will make your product about occupancy predictions extremely complex. It is important to educate the users about the corner case possibilities not being handled by algorithms and how to deal with them within the product itself.
Adapting to change
With the shift from offline to online and the emergence of social media, real-world business outcomes are changing rapidly. Imagine a top-rated restaurant getting below average ratings repeatedly for weeks. The restaurant will see a decrease in its footfall due to the changes in its reputation on social media. Any analytics product related to the restaurant business must adapt to the changing world and incorporate the additional data feed, such as reputation data, into their business model.
Measuring the right value
Several software products are sold in the Software-as-a-Service (SAAS) model. It means that customers could walk out if they found no measurable value from the product. At the development stage itself, it is important for product managers to think about ways to empower the customer to measure value from the product on an ongoing basis. If your product enhances the potential occupancy of a multiplex by five per cent, it is important to educate users about the potential revenue and profit impact of the increased occupancies. These measurable value calculations should be simple for the client to comprehend because it becomes a key differentiator in the buying process.
The five tenets help in analytics product development by addressing important marketing questions such as “Why should I build this analytics product?”, “Why will my client buy my product?” and “Is the market ready for this product?” It also helps avoid complexities by asking “Why do I need this additional data feed?” resulting in continuous product development.
