Selling Performance - Bristol WebPerf Meetup 2017-07-20
2 likes1,556 views
The document discusses the critical role of web performance in influencing user behavior and conversions, highlighting that even slight improvements in load times can lead to significant increases in conversion rates. It draws on various case studies, including Walmart and FT, to illustrate how speed impacts user engagement and conversion metrics. Additionally, it emphasizes the need for real user monitoring tools to effectively measure performance and its effects on visitor behavior.
Selling Performance - Bristol WebPerf Meetup 2017-07-20
- 1. Selling Performance @AndyDavies · NCC Group · May 2017 https://www.flickr.com/photos/chrisjohnbeckett/10446312055
- 2. Performance often isn’t at the top of people’s concerns • ~~~~~~~~~~~~~~~~~~~~~ • ~~~~~~~~~~~~~~~~~~~~~ • ~~~~~~~~~~~~~~~~~~~~~ • ~~~~~~~~~~~~~~~~~~~~~ • ~~~~~~~~~~~~~~~~~~~~~ • Security • Performance • Accessibility ☹
- 3. So what do they care about?
- 5. But it’s not just about getting visitors to buy things! - sign ups / subscriptions - completing tasks - interaction with adverts - sharing things - finding a store / restaurant - reducing costs - hosting / acquisition
- 6. Science explains visitor behaviour
- 7. Response Time in Man-computer Conversational Transactions Robert B. Miller, 1968 Instant Seamless Yawn! 100ms 1s 10s How we perceive response times
- 8. Response Time in Man-computer Conversational Transactions Robert B. Miller, 1968 Instant Seamless Yawn! 100ms 1s 10s Median start render time for 600+ UK retailers = 1.68s How we perceive response times
- 9. So yes, our attention span is shorter than a goldfish https://www.flickr.com/photos/watts_photos/23124797946
- 10. https://www.flickr.com/photos/ericparker/5416356195 “Badly performing web sites make us concentrate harder” Foviance
- 11. https://www.flickr.com/photos/dhollister/2596483147 “Stress levels comparable to watching a horror movie” Ericsson Mobility Report 2016
- 13. Most people don’t have an iPhone % sales by OS, three months to April ’17 - http://uk.kantar.com/tech/mobile/ 0 10 20 30 40 50 60 70 Android iOS Windows Other
- 14. Everyone has fast broadband… right?
- 15. Average UK connection speed is 16.9Mbps! Akamai State of the Internet Q1 2017 92% above 4Mbps 60% above 10Mbps 42% above 15Mbps
- 16. Which will load faster? www.bbc.co.uk/news 10Mbps1Mbps
- 17. Which will load faster? www.bbc.co.uk/news 10Mbps1Mbps
- 18. Which will load faster? www.bbc.co.uk/news 10Mbps1Mbps
- 19. Which will load faster? www.bbc.co.uk/news 10Mbps1Mbps
- 21. Which will load faster? www.bbc.co.uk/news 10Mbps1Mbps / 28ms / 280ms
- 22. Case studies demonstrate what’s possible
- 23. Wallmart demonstrated speed affects conversion in 2012 Walmart 2012
- 24. Cook reduced load times by 1 sec saw 7% increase in conversions https://www.nccgroup.trust/uk/about-us/newsroom-and-events/blogs/2017/march/how-we-predicted-an-increase-in-conversions-from-a-decrease-in-load-times
- 25. http://engineroom.ft.com/2016/04/04/a-faster-ft-com/ FT saw lower engagement when they made the site slower
- 26. Case studies are a great place to start https://wpostats.com
- 27. “Case studies don’t represent our demographic, our audience, our products, our proposition…” Has some truth
- 28. BounceRate 0% 25% 50% 75% 100% Page Load Time (s) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Site 1 Site 2 Site 3 Bounce rates across three retailers
- 29. ConversionRate 0% 2.5% 5% 7.5% 10% Page Load Time (s) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Site 1 Site 3 Site 4 Conversion rates vary between sites too Noisy as few data points at end of range
- 30. So what about real visitor experiences?
- 31. We’re used to the idea that conversion rate varies by device iPhone 6/7 iPhone 5/5s iPad iPhone 6/7 Plus Galaxy S6 Galaxy S7 Edge Galaxy S7 Galaxy S6 Edge Galaxy S5 Galaxy A3 Conversion / Bounce Rate 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% Conversion Rate Bounce Rate
- 32. But existing analytics tools are weak when it comes to speed data
- 33. But existing analytics tools are weak when it comes to speed data Sample contains < 4% of impressions
- 34. They’re no substitute for measuring your own visitor’s experience Performance Analytics aka Real User Monitoring
- 35. To understand how performance affects visitor behaviour
- 36. ConversionRate 0% 1% 2% 3% 4% 5% 6% BounceRate 0% 20% 40% 60% 80% 100% Page Load Time (s) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Bounce Rate Conversion Rate The impact on key business metrics
- 37. Conversion Rate drops 1% over 3 seconds ConversionRate 0% 1% 2% 3% 4% 5% 6% BounceRate 0% 20% 40% 60% 80% 100% Page Load Time (s) 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Bounce Rate Conversion Rate The impact on key business metrics
- 38. And modelling the benefits improved performance delivers Only NCC Group, Soasta and Blue Triangle’s Real User Monitoring products do this
- 39. Real User Monitoring is a strong driver of performance improvement
- 40. Other complementary approaches too
- 41. Filmstrips (and videos) are hugely powerful https://webpagetest.org
- 42. http://heatmap.webperf.tools Can be sumarised into heat maps
- 43. Waterfalls require explanation (I tend to avoid them with non-technical audiences)
- 44. http://requestmap.webperf.tools But enable fascinating insights
- 45. Block third-parties understand how they affect performance Produce comparison videos / filmstrips to show others
- 46. Can also compare against competitors
- 47. Can also compare against competitors
- 48. Videos prolong the painful moment
- 49. Or for fun… …compare against an F1 pit crew
- 54. Wait… they changed all four wheels? https://www.flickr.com/photos/cellardoorfilms/7620375702
- 55. So where next?
- 56. Establish which metrics people care about
- 57. Experiment - Change the performance of a page - Measure how metrics change (Same principles as AB / MV testing) Can build negative experiments too e.g make pages bigger
- 58. Measure Visitors Experience - Use the data to predict how improvements will change the metrics - Improve performance - Verify the result
- 59. Even ‘small’ performance improvements can have a significant impact
- 60. How much was 150ms improvement worth? (was 300ms on Android)