Sunday, April 05, 2020

Kmart Queuing for Website: A bad Idea Implemented Badly

News reports indicated Kmart Australia had implemented queuing for use of its website, to handle the volume of traffic due to the   At first I assumed I had misread this and it was online queuing for physical store entry, which would not be a bad idea. But this is online queuing for entry to the Kmart website. The customer is kept waiting at a screen which counts down where they are in the queue. The process worked as promised: after a two minute wait I was rededicated to the Kmat website. However, anyone with any training in human factors should know that you need to give people something to do in a queue. Kmart's designers seem to have gone out of their way to make the wait seem longer, and annoy customers. Also the Kmart website is inefficient in implementation, slowing the customer down further.

I suggest Kmart offer customers a catalog of specials to browse while waiting. Customers could be encouraged to make a note of what they would like to buy while flipping through the catalog. There could also be information about Kmart's efforts to deal with COVID-19, procedures for shoppers in store, and special coupons . This could be provided with efficiently encoded web pages, which place limited load on system, and keep customers engaged while they wait.

Also once I got to the Kmart site, after the predicted two minutes, I found I sub-optimal website implementation. The Google Page Speed Insights test rated this at 30 out of 100, for the mobile version. A few simple changes suggested by the test would improve the experience for customers:
Opportunity and Estimated Savings

 Eliminate render-blocking resources 1.06 s
Serve images in next-gen formats 0.15 s
Enable text compression 0.15 s


Ensure text remains visible during webfont load
Serve static assets with an efficient cache policy 18 resources found
Minimize main-thread work 3.3 s
Reduce JavaScript execution time 2.1 s
Avoid chaining critical requests 20 chains found
Keep request counts low and transfer sizes small 49 requests • 805 KB

No comments: