CelerData Blog

What Is User-Facing Analytics?

Written by Sida Shen | Jun 28, 2023 1:41:05 AM

The democratization of data analytics continues to rank as a top priority for enterprises, and this in turn has made user-facing analytics a major initiative for many engineering teams. But what is user-facing analytics and what should you know about it? This article will help you build your foundation in user-facing analytics, its benefits, use cases, and challenges.

 

What is User-Facing Analytics?

User-facing analytics is the process of providing data and insights to end-users of a product or service, rather than to its internal stakeholders or developers. User-facing analytics can help users make data-driven decisions or actions, increase their engagement and retention, and enhance their loyalty and advocacy.

This approach democratizes data analytics, putting the power of data analysis into the hands of those who need it most. It involves the use of interactive dashboards, data visualization tools, and self-service analytics software that users can use to access, analyze, and visualize data.

 

Why Use User-Facing Analytics?

User-facing analytics brings the power of data-driven decision making (DDDM) to the end user and empowers individuals with actionable insights without requiring them to have extensive knowledge of data science or programming. Some typical user-facing analytics scenarios include:

  • Dashboards and reports: Users can access various charts, tables, graphs, and metrics that summarize their data and performance. For example, a SaaS company can provide dashboards and reports to their customers to show them how they are using the product, what features they are engaging with, how they are achieving their goals.

  • Personalization and recommendations: Analytics tailors the customer experience, providing personalized results and recommendations based on their behavior and preferences. For example, an e-commerce website can use user-facing analytics to recommend products, offers, or content that match the user’s interests, preferences, or purchase history.

  • Alerts and notifications: These are use cases where user-facing analytics can help inform and update users by sending them timely and relevant messages based on their data and behavior. For example, a banking app can use user-facing analytics to alert users about their account balance, transactions, spending habits, or fraud detection.

  • Predictive analytics: This involves using historical data to predict future events or outcomes. For example, a fitness app can analyze a user's workout history and health data to predict their future performance or potential health issues. Similarly, a travel app can use predictive analytics to suggest the best time to book a flight or hotel based on past price trends.

 

User-Facing Analytics Challenges

Despite the advantages of user-facing analytics, current approaches still face numerous challenges and difficulties. Let's explore the common pain points and limitations that impact user experience and business outcomes.

 

Low-latency queries, especially with JOINs

Quick response times are essential for end-users when interacting with applications, and waiting several seconds for a query to complete can be frustrating for the following reasons:

  • User experience: Users expect fast and immediate feedback when interacting with an application or dashboard. If queries take too long to execute, users may become frustrated and lose interest, resulting in a poor user experience.

  • Interactivity: User-facing analytics often involves complex queries and filtering options, such as searching and sorting data by multiple dimensions. Low latency queries enable users to interact with the data in real time and explore different scenarios, which can lead to new insights and better decision-making.

  • Timeliness: In some cases, user-facing analytics involves time-sensitive data, such as stock prices or social media mentions. Low latency queries ensure that the data is up-to-date and relevant.

 

Addressing challenges with low-latency queries

While there are existing approaches that can provide low query latency, they are limited to single table queries, which require denormalization in the upstream pipeline. This approach is complex and expensive, and makes the data pipeline inflexible to business changes.

To address this issue, prioritize solutions that can offer excellent query performance not only for flat table schemas but also for star and snowflake schemas, without the need for denormalization.

 

High concurrency

User-facing analytics platforms must support a high number of concurrent users without sacrificing performance. High concurrency means that many users can access and query the same data source at the same time, generating a high volume of queries per second (QPS). High concurrency poses a challenge for user-facing analytics platforms because it can cause resource contention, query queueing, and performance degradation.

 

Addressing challenges with high concurrency

Existing approaches can provide high concurrency after some manual configuration. However, there are still factors to be aware of:

  • Query performance degradation: As the number of concurrent queries increases, especially with complex queries that involve JOINs and aggregations, the overall query performance may degrade.

  • Scalability limitations: Existing solutions may have limitations and may not scale linearly as machines are added.

 

Real-time mutable data

Fresh data is essential, especially in user-facing analytics. Delayed data can result in missed opportunities or incorrect decision-making. User-facing analytics systems need to support data freshness in seconds, which means they need to ingest, process, and visualize data streams that can change rapidly and unpredictably. This poses significant technical challenges for user-facing analytics systems in terms of scalability, reliability, and consistency.

 

Addressing challenges with real-time mutable data

To deal with real-time mutable data, many systems use the merge-on-read (MOR) technique to update data. In this technique, data changes are applied to a separate copy of the data, and the original data is only updated when it is merged with the new copy later. This technique can work well for data freshness in the minutes range, but it fails for anything faster. This is because the merge operations can affect query performance significantly. It can make the query slow, unstable, and unpredictable, which can cause uncertainty and frustration for users.

 

User-Facing Analytics Done Right

While it is relatively straightforward to solve any one of these challenges, finding a solution that can address all three simultaneously has proven to be difficult. Recognizing these limitations in user-facing analytics, the StarRocks project was developed. StarRocks is a high-performance analytical database designed with user-facing analytics in mind. You can learn more about it in this feature article.

If you’re a more hands-on learner, however, then you can get started with StarRocks by downloading it here and joining our Slack community. If you want an even more streamlined experience, try the commercial cloud version of StarRocks for free at Cloud.CelerData.com