Data-driven stable systems v/s Feature-Rich Ecosystems
We today live in a world where we now have an app for everything. Today we have more than 5.5 million apps available across all app stores with over 178.1 billion downloads every year, suggesting that on an average each of the 2.7 billion mobile users try more than 65 apps.
This clearly tells us that we have a technological solution available to solve every type of challenge faced by an organization or by you.
For all of us who are end-users, we witness these applications continuously introducing new features and updates to accommodate our needs as well as to increase the MAU.
However, when applying this specifically to solving business problems, this leaves all of us with a few questions
- Can feature-rich apps fix business problems?
- Are good apps just a sum total of great features?
- What gives a better competitive advantage: better features or robust organizational structure?
The induction of technologies like the Internet of Things and connected devices have led to an increase in the amount of data generated. For instance, the current model of A-350 generates 2.5 terabytes of data per day. As a result, the challenges of getting actual insights from data are enormous. Teams across organizations today are trying to win this never-ending race by building point solutions or features that help them solve a business problem in the short-term.
The industry is currently working with data in a more feature-centric manner. So, the approach works on the feature-to-data methodology. On the contrary, systems need to be data-centric. So, the focus must be on the different components of the data lifecycle. The churned data must then be analyzed to see the features that it can best support. Eventually, this approach is expected to provide insights, which will be more productive and accurate.
The Tradeoff Between Feature-Rich and Data-Centric Ecosystems
With the rising wave to develop technologically driven ecosystems, providers are forced to focus their attention on features instead of strong frameworks. According to a report, the artificial intelligence market for airlines is expected to hit USD 2.2 billion by 2025, with the focus on passenger experience and the Internet of Things, among top trends.
While Digital teams plan for the implementation and use-cases of these technologies, most airlines have not really figured out how they would process data across these systems and derive the right insight.
Most airlines work on the approach that advocates collating data and acting on it with the help of visualizations and prediction agendas, missing out on the acquisition and processing part, to a great degree.
With airlines wanting it all and supporting practically everything from social media analytics, speech analytics, contact centers, and revenue model analytics, the issue is particularly escalated in the scenario where the number of data sources is ever increasing. It is no longer possible to collate all the data and just act. It is more important to process data correctly before actionable results can be made available.
Besides this, the current ecosystem depends on disparate systems for supporting different features, which requires vertical integration. These individual systems are offered by a multitude of technology partners. This undeniably makes the process of integration rather challenging.
TAcquisition and Processing for System Accuracy
The data lifecycle is largely divided into four phases. In the first phase, data is acquired and collated. Storage and processing of data occur in the second and third phases, respectively. Lastly, only after the results are available that are visualized and presented to the user. Typically, in order to evaluate a system, two evaluation metrics are computed, which include accuracy and performance.
Depending on the application concerned, it is determined as to which of these evaluation metrics is more significant. With that said, the performance metric for a system comes into picture only once the desired accuracy has been achieved. For all data analytical applications, accuracy largely depends on two phases namely, acquisition and processing.
Storage, processing, and visualization majorly affect the performance of the system. While storage infrastructure and applied processing techniques have a significant impact on response time and other similar performance metrics, visualizations determine user-friendliness and usability of the system.
If the collected data is irrelevant, insufficient or inaccurate, the results cannot be expected to be any better. According to Gartner’s research, businesses lose an annual average of USD 9.7 million from their revenue because of poor data quality. The same holds true for the applied processing techniques. This makes it unavoidable for airlines and providers to create ecosystems with robust acquisition and processing components instead of focusing on visualizations in view of the fact that quality and accuracy of visualizations depending on the efficacy of the backend system.
Although the market is showing a drift towards feature-rich ecosystems, compromising on the strength of the ecosystem to support them, is not advisable. Data acquisition and processing are the most crucial sub-systems for a robust ecosystem. It is only if the data is collected suitably and processed aptly that it will be used for analytics. Feature-rich ecosystems are mere deceptions as far as a competitive advantage is concerned. Airlines and providers need to focus on the strength of the core system. Data-driven ecosystems can play a significant role in this, to create a lasting impact on the industry.
About the Author
Vice President – Product Management
Leading Product Management Group and driving Engineering/DevOps team with a vision of creating most innovative price intelligence and revenue management products in airline technology.