Measuring Image Performance
Steve Souders
Chief SpeedCurver, SpeedCurve
Designers and developers instinctively know that faster websites make happier users. Thankfully, there's a wealth of case studies that prove our hunch. Thus, tracking website speed is necessary for knowing where our users are on the happiness meter. Join Steve as he takes us on an exploration of the best ways to measure performance of images.
USERS DON'T CARE ABOUT THOSE GRANULAR METRICS
THEY CARE ABOUT SEEING THE IMAGES
Image size doesn't even matter for mobile data plans – images are nothing compared to video.
https://www.flickr.com/photos/kk/6863172432/in/photolist-bsty4b-3HDYJE-wiSVMX-eVWzMJ-9ftJBe-kYWQ8F-aucUmh-diNSS9-qZg4Rp-kjFung-dmuJKm-pH4iVp-bnTdAq-9LFXXN-ba67YV-dhth5h-oaaZKo-nd9Fs1-29RVSQ-pJGAjW-i29xbH-ajNXL9-o8WqV6-uqMyAg-fdPuAo-srZ5Yr-nwMJDC-jibyGv-hBcuis-vKiMxj-pu9xhj-rwJSDL-dnDq8k-swkmVC-o151uM-hBbGwi-gfwits-cHtV9-61nXN5-bAZ5za-pU4HhE-kBkvPH-pn47sx-hTFhB9-mnGW1V-o8wD5y-rRvX4y-9B1Qwy-dQuEdE-rDsEGb
THIS IS A BAD UX
SO THE IMPORTANT METRIC IS WHEN IMAGES RENDER
https://www.pexels.com/photo/brushes-oil-paint-table-94736/
no attribution needed
WHY DO WE SEE THAT BLANK WHITE SCREEN HAPPEN?!
https://www.flickr.com/photos/ksayer/5614813544/sizes/l
SCRIPTS AND STYLESHEETS
https://www.flickr.com/photos/tipsterhog/1066224195/sizes/l
METRICS LIKE IMAGE SIZE AND RESOURCE TIMING DON'T EXPOSE THIS TERRIBLE USER EXPERIENCE!
THIS IS 20 SECONDS BECAUSE IT'S A CATASTROPHIC OUTAGE. BUT A 2-3 SECOND DELAY, WHICH IS MUCH MORE LIKELY, WOULD PRODUCE A 2-3 SECOND DELAY IN IMAGE RENDERING AS WELL.
AVG WEBSITE HAS 14 CRITICAL BLOCKING RESOURCES (HA).
IF PROB OF ONE TAKING 3+ SECONDS IS ONLY 10%, THERE'S A 77% PROB THAT AT LEAST ONE OF 14 WILL TAKE 3+ SECONDS.
https://www.webpagetest.org/video/compare.php?tests=170228_4Y_4dcfe7bca1ffe45ea896c8fa7f4bf2fc%2C170228_TG_ebdbb5adab5a95aeef0035873dd43ab8&thumbSize=200&ival=5000&end=visual
https://www.webpagetest.org/video/compare.php?tests=170228_FM_5a6994df455a12de0608a022845b7a79%2C170228_BS_3bc72be813135f3c9f9e7350fc9329c5&thumbSize=200&ival=5000&end=visual
https://www.webpagetest.org/video/compare.php?tests=170228_5F_83d78664c918eac2aac5a123ea385ecf%2C170228_PC_7ccfc6842e73923c6d279824af26eca0&thumbSize=200&ival=5000&end=visual
PROB OF AT LEAST 1 OF 8 TAKING 3+ SECONDS IS 57%.
1 OF 10 IS 65%
(IF PROB OF 3+ SECONDS IS 10%)
https://speedcurve.com/speedcurve-enterprise/top-sites/test/170223_2N_98309f17a27ee546c8c5709fcbb48ee7/