NEW Game developers: check out PacketZoom Mobile Connect
Important announcement: PacketZoom is now part of Roblox   

Mobile Travel Performance Index

Is Your Mobile App Fast Enough for Travelers?

PacketZoom's Travel App Performance Index benchmarks the performance of selected popular travel and hospitality mobile applications. The Index focuses on the home (landing) screen of each app and ranks them by average time-to-search (amount of time it takes to allow the user to start their travel search). The variance in time-to-search is a result of both: (1) design methodology and (2) sequencing of screen elements. Brands that are ranked high are likely to make the right user experience decisions.

PacketZoom Mobile Expresslane was able to reduce time-to-search, with speedup factors of up to 2.84x

Network Analytics and Monitoring

Travel App Retail index for iOS devices over WiFi

Test Methodology:

  1. Downloaded each app from Apple's AppStore on iOS device.
  2. Launched the app and interacted with it while examining its network traffic.
  3. Captured url flow and network request pattern of all the static assets that were downloaded before the app showed a search box. Saved this session flow as an "executable script". (Note: These scripts are for use on our proprietary First Look platform, The only one of it's kind in the world. This platform allows us to gather RUM data on any url of our choice from thousands of actual mobile devices operating on real-world networks).
  4. Using our First Look Test Platform, executed the script on hundreds of mobile devices globally on real mobile user sessions. Only sessions connected to WiFi were used for this index.
  5. Executed this script via PZP (PacketZoom Protocol), measuring total time taken for all downloads in the script.
  6. Executed the exact same script from the exact same mobile device during the exact same user sessions via HTTP(S) protocol. Same time measurements were done for this case.
  7. Averaged the results from step 5 to find out the existing "Time to Search with PacketZoom" from all the test devices.
  8. Averaged the results from step 6 to find out the existing "Time to Search" from all the test devices.

Notes:

  • Steps 5 and 6 were executed sequentially on each device back-to-back on the same user session. Guaranteeing apples to apples comparison. Note that PZP script was run first to absorb any radio warmup penalty in the PZ column.
  • Testing was performed over WiFi networks on the second week of April during daytime (PST).
  • The "total page size" and "number of images" are just counting assets downloaded by the app *before* user conducts any searches using precious mobile network bandwidth.
  • The test script recorded only the images that were requested *before* the search box was rendered. This required careful manual effort for each and every app in the index.