What is a Data App?

Data apps are applications that rely heavily on data and have an easy to use, accessible UI that enables data-driven decision making. Think of them as a marriage between dashboards and web applications. A few other names that define them quite well: data-intensive apps and data-driven apps.

Yes, every app needs data to function, but this is an emerging new category of apps, and it is one of the next big things in the data world. If you watch the modern data stack and the data space closely, you can see a trend forming around cloud data warehouses as the new “backend” for all of this. 

So what do data apps enable? From a user perspective, they allow you to analyze, slice & dice and take action on top of massive amounts of data through an easy-to-use user interface. They are usually customer-facing (external) and can be full applications or components/modules within existing applications. 

Do you want a one-liner definition of a data app? Here it is:

“Any web or mobile app delivering a packaged, sub-second analytics experience over large datasets to its end users”

Unpacking this a bit, what is a “packaged, sub-second analytics experience”? Users are a discontent bunch, how do you provide them a simple, responsive analytics experience? A packaged experience is: 1. tailored to the interactive needs of the user in terms of datasets and query patterns; 2. Responsive, irrespective of the complexity of the query or the amount of data that needs to be examined to generate insights. What is a large dataset? Any dataset that can stretch the limits of infrastructure in the form of compute, memory, storage, network. This could be in the form of total amount of data, number of records, number of attributes (columns) to examine, or the need to combine several interconnected datasets. Bottom-line, a data app will stretch the limits of your infrastructure, budget and tuning skills when it comes to the only thing that matters - insights that the business or end-user desires.

Examples of Data Apps

Google Analytics (GA) is an app that most marketers are familiar with. It provides insights into website performance and visitors. Marketers using GA can identify sources of user traffic, manage and measure campaigns. Various KPIs such as bounce rate, session duration, pageviews per session can be used to measure user engagement. 

Intuit Mint is a consumer financial app that is available at no cost. How do you get a single pane of glass for all your financial accounts and get insights on your spend patterns ? Intuit Mint provides just that.

Still too exotic and far fetched? Think of your banking app or cell phone bill, there are data components that let you understand spend, projected spend, savings or lack there-of etc. These are all elements of a data app. 

While some of these are consumer centric, data apps are applicable in B2B interactions as well. For example, a FinTech start-up that provides a no-code payments integration platform to thousands of E-Commerce shops delivers payment visibility and fraud detection on a data app.  

Irrespective of whether it is a consumer or B2B app, data apps deliver insights to the user directly. But more importantly, data apps have common attributes that you should be aware of.  A follow-up question that comes up, hey what is not a data app? Isn’t Facebook a data app? According to ChatGPT:’Facebook is a social networking platform that allows users to connect with others and share information, but it is not specifically designed as a data app. While Facebook does collect and store large amounts of data about its users and their activities on the platform, it primarily provides a space for users to connect and communicate with one another, rather than offering tools for analyzing and interpreting data.’ But, any application used to analyze and understand your Facebook insights could be a data app.  Next, let us look at common characteristics of data apps.

Common characteristics of Data Apps

1. Data apps are about the use case and the value you deliver through insights. Data Warehouse workloads have evolved through batch, traditional BI and embedded analytics use cases. This evolution now includes data apps. The concept of a data app is expanding the last mile of analytics to both internal and external customers. As with any internal or external customer facing analytics use case, understanding the customer needs and experience is foundational.

2. Data apps work on large volumes of data, think GB, TB, PB range. One man’s TB is another man’s GB. The point is data volumes vary by industry and use case. But what we find is that data volumes are only growing. If you are in the GB range today, get ready for a TB future. As data volumes grow, the challenges of preserving the user experience will increase as well. When dealing with large amounts of data that stretch the limits of your infrastructure, data apps invariably require efficiency in storing, processing, analyzing and visualizing insights. 

3. Data apps deliver value through fresh, high quality data. Freshness of data is relative. A Fitness app may not have the same freshness requirement that a Fraud detection data app may need. However, freshness of data matters in both cases. With a fitness app, you might never go back to a data app that is 48-hours behind on reflecting your calorie tracking. On the subject of data quality, it is not something that happens overnight and only on production. Every environment, every stage needs to have quality checks built into it.

4. Data apps have an easy-to-use, intuitive and dynamic user interface and can be built as custom apps or on top of frameworks or low-code / no-code tools or in the case of various internal cases through popular business intelligence tools or embedded analytics.  Nothing more to elaborate on this one.

5. Data apps use SQL queries to interact with the underlying data store. However, data sources increasingly deal with semi-structured data. Data stores should provide the flexibility in dealing with semi-structured data in the form of JSON. 

6. Data apps require low latency because an actual user is sitting in front of their smartphone waiting for the result. And it's well known that users lose interest if the response time is not subsecond. We can no longer hide behind the “it's only a nonessential internal use case” argument. The only answer is that data apps need to be fast and need to deliver sub-second analytics. What if your app is incredibly successful? What if your user count grows from a hundred to a thousand? What if the number of concurrent queries jumps from 10 to 800? Can you still continue to deliver the same level of performance? Data apps are redefining performance.

So then, what is a data app? Let me spare you the trouble of scrolling all the way back.

“Any web or mobile app delivering a packaged, sub-second analytics experience over large datasets to its end users”

Question, how do you deliver a data app … we’ll do that in a follow-up blog…

‍