Delta Airlines

Case study - Flight Result Load Time Experience h4> Date: Nov. 2023 - Current

Overview

Normally, when a user opens a webpage, the entire page’s contents are downloaded and rendered in a single go. While this allows the browser to cache the web page, there’s no guarantee that a user will consume all of the data that is downloaded onto a page. Thus slowing the download speed of the page down. Instead of dumping all the content on one single load, the content will load when the user accesses part of the page when called. With lazy loading, pages are created with placeholders and displayed when inside the user's viewpoint.

The Goal

My goal was to improve the response time on the Delta Flight Results page and decrease the load time under 6 secs.

The Process

First, I had to understand exactly what data was rendering on the search results page. By conducting usability tests and working cross-functional with the Shop and BF team, I can record the exact data being called to the page.

Flight Results Load Time 9.17 seconds

I tested a series of load scenarios conducted between Nov. 2023 through Dec. 2023, parallel with existing load time results.
Observations:

The average response time of a one way flight was 9.17 seconds
23% or 2.149 seconds was taken before the overall search API was called.
BFF took an additional 6% or 0.83 seconds to render results

The Solution

Non-lazy load vs lazy load

Positive Findings

1 | The average response time of 5 search result per page (initial) was slightly better than the 15 search result per page

2 | The average response time with 5 search results is improved by 800ms compared to the existing search.

3 | he subsequent pages (load more page) average response times are similar in improvement.