HeadSpin Proposes Speed Optimizations as Perfect Item for Flipkart
Overview and Methodology For Our Analysis
Moto G3 / Airtel / Bengaluru
HeadSpin Connected Intelligence Platform
- Review multiple 3rd party SDKs that cause slowdown in launch time
- Ensure that immediate content is loaded first before future content.
- Optimize connection reuse
- HeadSpin analyzed the latest version of the Flipkart Android app and mobile site on a real Moto G3 in Bengaluru, on Airtel Network.
- NO SDK, code change or instrumentation was needed to run the remote performance sessions.
- HeadSpin’s AI engine sifted through network traffic, client-side metrics, and videos of the test execution to find areas of poor user experience and performance bottlenecks.
- Recommendations were provided for every issue that was surfaced.
HeadSpin looked at Flipkart reviews in the Play Store and found some users struggling with functional issues and technical errors.
Flipkart ratings is currently 4.5/5 on the App Store.
Insights and Recommendations
HeadSpin’s proprietary AI technology analyzes and prioritizes the core issues impacting critical user journeys. Using computer vision techniques to quantify blank screens, time to interact, loading time, loading or buffering animations, and content quality, our AI models can automatically diagnose server-side issues whether they arise from infrastructure deployment issues, poor performance, or API errors. For a custom analysis of your app, request a Performance Insight report from HeadSpin.
- Launch time to interact was 3.76 seconds on Android.
- Flipkart client app waited 508 ms before receiving a 404 response from the API server, on the main Shop view. GET request failed when attempting to receive a JSON payload of product data.
- There were 3 instances of slow server responses, cumulative impact being 1.56 seconds
- 18 instances where the Flipkart app created new TCP connections instead of reusing existing ones
- Old OkHttp library detected
- Launch time was 4.8 seconds
- Duplicate fetches of the same store content detected
- 5 instances where the Flipkart mobile website created new TCP connections instead of reusing existing ones.
FLIPKART KPIs VS PEERS
All KPIs are from sessions ran on a OnePlus 2 located in New Delhi, on Vodafone, end of Feb 2020.
|App Name||Launch Time||Search||Add to Cart|
|Tata CLiQ||2.63 sec||4 sec||1.87 sec|
|Amazon||6.5 sec||3.1 sec||0.67 sec|
|Flipkart||6.6 sec||1.6 sec||0.77 sec|
In this session, the load time of the Flipkart Android app was 3.76 seconds. Our Waterfall UI allows you to see all the API calls made during this time. The shaded orange area corresponds to the time in the session where the app was launching.
Failed API Calls
On the main Shop view there was a GET request that failed when attempting to receive a JSON payload of product data. The Flipkart client app waited 508 ms before receiving a 404 response from the API server. The API server was an nginx web server and the request had no-cache in the HTTP response content headers, thus the resource would have to be revalidated if requested again.
There were 3 instances detected in this session where Flipkart servers were slow to return a response to the client. The cumulative impact to the client was 1.56 seconds.
Optimizing TCP Connections Created
The Flipkart Android build analyzed was using OkHttp Version 3.12.1, which is 1 year and 10 days old compared to the recent release of the library. Significant improvements have been made in the library, such as fixing connection degradation issues after a request timeout, preventing dropped callbacks on unexpected connections, improving connection pooling, and much more.
In the conducted session, after the Flipkart mobile website fetched the JSON payload to the main Store page index, 46 seconds later it made the same request for the same JSON payload. From taking a look at the HTTP Content Headers, we can see this behavior occured because of no-cache and a max-age of 0, causing a revalidation of the request and no reuse.
|Cache-Control||private, no-cache, no-store, must-revalidate, max-age=0|
Optimizing TCP Connections Created
How to Work With HeadSpin
HeadSpin Powers Flawless Mobile App Performance
HeadSpin’s state-of-the-art Digital Experience Intelligence Platform™ will empower your engineering teams to:
- Execute automation tests on real carrier-enabled devices worldwide
- Discover performance bottlenecks
- Collect and visualize your performance data
- Seamlessly integrate with your CI/CD workflow
- Monitor every critical user flow of your mobile experience
- Pinpoint issues that cause performance regressions and impact user experience earlier in your development cycle
- Launch new features confidently by monitoring and analyzing root causes of performance issues throughout the development life cycle
- Drive new user installs and user engagement with optimized performance
- Forge user trust and safety, accelerating consumer services for critical user paths across your apps and website
- Benchmark KPIs for mobile performance to understand how your user experience for your mobile app and website compares to your peers in the same industry
- Support localized testing across thousands of devices and 70+ countries around the world, on real devices under real-world conditions
HeadSpin has the product and solutions expertise to help you perfect the digital experience for your customers! Our team hails from digital pioneers, such as Google and Facebook, and has already delivered enhanced mobile-centric experiences for a diverse set of Fortune 1000 clients. HeadSpin’s strategic insights can help you:
- Drive new user acquisition, user engagement, and feature adoption
- Deploy app security measures on iOS and Android
- Build consistent performance standards and experiences across all your mobile apps and websites