02 web performance
10 likes1,021 views
The document discusses the importance of website performance and the impact of speed on key metrics like customer satisfaction, conversion rates, and revenue. It outlines challenges to performance from increasing traffic levels, mobile usage, different browsers, and third-party components. The presentation argues for a holistic approach to web performance management to identify and address bottlenecks across infrastructure, applications, and real user experience.
02 web performance
- 2. Performability, Could they do it? Complete Web Monitoring @jeroentjepkema #WAC2010
- 3. The Internet is getting complex…. "
- 4. The End-User: Everywhere! !
- 7. The end-user is in control "
- 9. Back to the fundamentals: Quality of Service"
- 10. Performance matters! "
- 11. Google loses 1% of traffic for every 100 milliseconds of delay "
- 13. Impact of additional delay on business metrics !
- 15. Shopzilla had another angle" Big, High-traffic site" 16 Month re-engineering" 100M impressions a day! Page load from 6 seconds to 1.2! 8000 searches a second! Uptime from 99.65% to 99.97%! 20-29M unique visitors a month! 10% of previous hardware 100M products! needs!
- 16. 12% increase in revenue," 25% increase in page views for" every 5 seconds of gain"
- 18. As of April 9th 2010:! Site speed co-determines " Google page ranking!"
- 19. Google Pagerank"
- 21. Evolving Customer Expectations" 8 seconds 4 seconds 2 or less ? Response Time Source: Forrester Consulting, 2009
- 22. Availability and Stability are! KEY ISSUES!!
- 23. Performance impacts Customer Loyalty!
- 24. Average Impact of 1 second delay in response time for web users" Source: Aberdeen Group, 2009 24
- 25. Impact of Poor Experience" Source: Gartner, 2008!
- 27. Objective: Influence visitor engagement" - Visitor engagement 10 ms +
- 28. Objective: Influence visitor engagement" - Visitor engagement 100 ms 10 ms +
- 29. Objective: Influence visitor engagement" - Visitor engagement 1 sec 100 ms 10 ms +
- 30. Objective: Influence visitor engagement" - 10 sec Visitor engagement 1 sec 100 ms 10 ms +
- 31. A/B Performance Test Setup" Visitor Strangeloop Website Decide whether to optimize Normal Receive Accelerated content page Insert Process segment scripts Optimize? marker Send analytics Unaccelerated Google Analytics
- 32. Technical Problem Outlined - Before" 32
- 33. Technical Problem Outlined - Before" HTML (BackEnd) 63 Round Trips (Front End) 1.975sec 9.537sec 769KB 33
- 34. Strangeloop Solution - After " Before" After" Round trips: 63! Round trips: 8! Start Render: 1.975 Sec! Start Render: 1.198 Sec! Load Time: 9.537 Sec" Load Time: 3.69 Sec" Payload : 769 KB! Payload: 456 KB! 34"
- 35. So hereʼs what weʼve discovered…."
- 36. Traffic Levels" 18! 9000 8.505! 16! 14! 6750 12! Total number of visits" 4.740! 10! 4500 8! 6! 2250 4! 2! 00! Optimized! Unoptimized!
- 37. Bounce rate" 14,60%! 14,40%! 14,35%! Bounce rate" 14,20%! 14,00%! 13,80%! 13,60%! 13,38%! 13,40%! 13,20%! 13,00%! 12,80%! Optimized! Unoptimized! 37
- 38. Average Time on Site" 00:36:00! 00:30:10! Time (seconds)" 00:28:48! 00:23:50! 00:21:36! 00:14:24! 00:07:12! 00:00:00! Optimized! Unoptimized! 38
- 39. Pages per Visit " 18! 15,64! 16! 14! 12! 11,04! Pages" 10! 8! 6! 4! 2! 0! Optimized! Unoptimized! 39
- 40. Conversion rate and order value" 20,00%! 16,07%! Change in shopping cart metric" 15,00%! 10,00%! 5,51%! 5,00%! 0,00%! Conversion rate! Order value! -5,00%! -10,00%! -15,00%! -20,00%! 40
- 41. Customers want more and they want it faster "
- 42. 42 Entire contents © 2009 Forrester Research, Inc. All rights reserved.
- 43. Time to speed up!"
- 44. Compared to the competition!
- 45. ABN Amro 1.81 sec Bijenkorf 0.76 sec D-reizen 1.79 sec Alex 3.43 sec Bol.com 1.20 sec Neckermann 0.61 sec ING 0.86 sec Hema 1.19 sec NS 0.86 sec Rabobank 1.42 sec Mediamarkt 2.39 sec Sunweb 0.39 sec SNS Reaal 1.64 sec TomTom 1.46 sec Vliegwinkel 5.04 sec
- 46. ABN Amro 1.81 sec Bijenkorf 0.76 sec D-reizen 1.79 sec Alex 3.43 sec Bol.com 1.20 sec Neckermann 0.61 sec ING 0.86 sec Hema 1.19 sec NS 0.86 sec Rabobank 1.42 sec Mediamarkt 2.39 sec Sunweb 0.39 sec SNS Reaal 1.64 sec TomTom 1.46 sec Vliegwinkel 5.04 sec
- 47. Retail 2.31 sec Average speed per sector, Finance 3.30 sec Gomez Benchmark 2009 Travel 3.55 sec ABN Amro 1.81 sec Bijenkorf 0.76 sec D-reizen 1.79 sec Alex 3.43 sec Bol.com 1.20 sec Neckermann 0.61 sec ING 0.86 sec Hema 1.19 sec NS 0.86 sec Rabobank 1.42 sec Mediamarkt 2.39 sec Sunweb 0.39 sec SNS Reaal 1.64 sec TomTom 1.46 sec Vliegwinkel 5.04 sec
- 48. Retail 2.31 sec Average speed per sector, Finance 3.30 sec Gomez Benchmark 2009 Travel 3.55 sec ABN Amro 1.81 sec Bijenkorf 0.76 sec D-reizen 1.79 sec Alex 3.43 sec Bol.com 1.20 sec Neckermann 0.61 sec ING 0.86 sec Hema 1.19 sec NS 0.86 sec Rabobank 1.42 sec Mediamarkt 2.39 sec Sunweb 0.39 sec SNS Reaal 1.64 sec TomTom 1.46 sec Vliegwinkel 5.04 sec
- 51. The userʼs distance from the content matters"
- 52. 6.54 sec 4.89 sec 4.07 sec 5.49 sec 6.09 sec 7.19 sec
- 53. Forrester research shows that over 40% of consumers will wait no more than 3 seconds for a Web page to load before clicking away to a compeHtor 6.54 sec 4.89 sec 4.07 sec 5.49 sec 6.09 sec 7.19 sec
- 55. 58% of mobile phone users expect websites to load as quickly, almost as quickly or faster on their mobile phone, compared to the computer they use at home Among adults who accessed the internet with a mobile phone in the past 12 months, Gomez Mobile Web Experience Survey
- 56. desktop 1.17 sec Iphone
- 57. desktop 1.17 sec 20.19 sec Iphone
- 58. desktop 1.17 sec 20.19 sec Iphone http://m.bol.com 5.02 sec
- 59. desktop 1.94 sec Iphone
- 60. desktop 1.94 sec 15.60 sec Iphone
- 61. desktop 1.17 sec 20.19 sec http://m.nu.nl 4.64 sec
- 63. Performance differs per browser "
- 64. Each browser acts different" Firefox Internet Explorer
- 65. Raw page load time"
- 66. Raw page load time vs. Perceived Performance"
- 67. 0.95 sec 1.82 sec 1.00 sec 1.19 sec 2.637 sec 2.59 sec
- 68. Availability & performance issues can also be browser specific
- 69. Assembly within the browser: it all has to work !
- 71. ?
- 72. ?
- 73. Real User Experience !
- 80. Congestion!
- 81. Watch out for Peak Traffic!"
- 82. Watch out for Peak Traffic!"
- 83. Conclusion: Bottlenecks are everywhere!
- 84. Bottlenecks are everywhere" How fast is fast enough? Performance by region Mobile Performance Browser Experience 3d Party Performance Real User Experience CongesHon
- 85. How to make your website faster?"
- 87. How does the performance of Infrastructure and Application…. Management Tool! Load Web App DB! balancer! server! server! Internet! Client!
- 88. …relate to the customer experience Management Tool! Load Web App DB! balancer! server! server! Internet! Client!
- 89. Introducing the synthetic user Management Tool! Active Monitoring! Load Web App DB! balancer! server! server! DB! Internet! Client!
- 91. Active Monitoring Continous Check Performance Measurements & Availability
- 92. Active Monitoring Continous Check Performance In case of failure: Measurements & Availability Alerting
- 93. Active Monitoring Continous Check Performance In case of failure: Root Cause Analysis / Measurements & Availability Alerting SLA reporting
- 95. Response time and Availability"
- 96. Page breakdown"
- 97. From where your users are "
- 98. But synthetic isnʼt enough"
- 99. But synthetic isnʼt enough" Gomez Asia Gomez US Gomez UK
- 100. Measure all your users Client! Client! Management Tool! Active Monitoring! Load Web App DB! balancer! server! server! DB! Internet! Real User! Monitoring! Client! Client! Client!
- 101. Collected data: Active Client" Monitoring! Client-‐ip, Server-‐ip, Gateway-‐ip Page Requested URL (Page,Object) request Cookie/Session-‐id Network Response Time Referrer Networkresponse time GET & POST requests User agents (phone type) Real User! Load Monitoring! balancer" Server Response Time Web server" Response? Server error, web site error, Hit OK Content of page (i.e order details, App server" funcHonal errors) DB" Delivered? Aborted? Network Hmeout?
- 102. Business Impact per user Satisfaction Reports Business Transaction Funnel Usage Reports: • Data Browser • Sales by Country/State
- 103. Session Replay User SessionTracking" Session Replay" Business Service Monitoring (e.g. Order Entry, Buy Stock)!
- 104. Root Cause Analysis per user " KPI configuration, monitoring, & alerting! Server, network, & object performance diagnostics! Error reports & analysis!
- 105. Active Monitoring:! Active Monitoring" Simulate users by scripting pages and/or business transactions! Continuous measurements from outside the firewall! Best used for incident, problem and SLA management.! Real User Monitoring! Real User" Monitoring" Measures traffic of ALL users, transactions and pages! Passive solution which gives information about the user experience! Best used for performance and capacity management!
- 106. But, how to become the performance leader?!
- 107. 5 Web Performance Best Practices "
- 108. 1: Create Shared Performance Objectives"
- 109. Business vs. IT
- 110. Business Invest to aeract more and new customers Quick Time to market Conversion rates Realize Sales targets IT Quality of Service Be Cost efficient Follow procedures Standardize
- 111. …..How to work together?"
- 112. Key Business Metrics" AJenKon Engagement Conversion NEW VISITORS GROWTH CONVERSION SEARCHES RATE PAGES TIME TWEETS MENTIONS NUMBER OF VISITS PER VISIT ON SITE x ADS SEEN ORDER LOSS VALUE BOUNCE RATE 112
- 113. Key Business Metrics" AJenKon Engagement Conversion NEW VISITORS GROWTH CONVERSION SEARCHES RATE PAGES TIME NUMBER TWEETS MENTIONS OF VISITS PER VISIT ON SITE x ADS SEEN ORDER LOSS VALUE BOUNCE RATE Performance maeers! Web Quality directly impact business metrics 113
- 115. Business oriented Key Performance objectives !
- 116. Business oriented Key Performance objectives ! System Management Business TransacHons
- 117. Business oriented Key Performance objectives ! System Management Business TransacHons System Availability End User Availability
- 118. Business oriented Key Performance objectives ! System Management Business TransacHons System Availability End User Availability Build Quality Web Stability
- 119. Business oriented Key Performance objectives ! System Management Business TransacHons System Availability End User Availability Build Quality Web Stability Capacity (Peak) Traffic
- 120. Business oriented Key Performance objectives ! System Management Business TransacHons System Availability End User Availability Build Quality Web Stability Capacity (Peak) Traffic Network Page Speed
- 121. 2: Measure from End User Perspective !
- 122. Monitor Key Transactions from where your End-Users are!
- 123. Benchmark against the competition!
- 124. Measure every detail of your traffic" Customers: Global! Access: Getting Mobile! Technologies: Enterprise/Virtualization/Cloud! 12 4!
- 125. 3. Plan for Peak Crowds"
- 126. 1. Compare Actual Results
- 127. 1. Compare Actual Results 2. With Actual Performance
- 128. 1. Compare Actual Results 2. With Actual Performance 3. To forecast Capacity & Performance
- 129. Peak Traffic: Itʼs a Business KPI! " Capacity forecast, actuals and maximum Max. number of pageviews per hour 1.200.000 1.000.000 800.000 Christmas/EOY Pageviews 600.000 400.000 Traffic Realized Traffic Forecast 200.000 MAX capacity Safety capacity 0 nov-11 Jan-11 apr110 sep-11 jan-10 jul-10 sep-10 dec-10 mrt-10 jun-10 okt-10 aug-10 jul-09 sep-09 dec-09 nov-10 jun-09 okt-09 aug-09 jul-11 Mei-09 mei-11 okt-11 nov-09 feb-10 jun-11 aug-11 apr-11 dec-11 mrt-11 feb-11 mei10
- 130. 4. Test and Simulate performance"
- 132. Make sure the view your customers are getting is the view you want them to see!
- 133. Make sure the view your customers are getting is the view you want them to see!
- 134. Measure performance for normal traffic…. "
- 135. …. AND Peak Traffic"
- 136. Find your ceilings "
- 137. Source: John Allspaw, Flickr.com what you have left The End Know what your limits are…"
- 138. Plan for the unexpected Measure performance under normal and peak conditions to find all the problems and optimization opportunities
- 139. 5. Tying performance to conversion:" Web Performance Analytics!
- 140. Is it marketing?"
- 141. Or the WEB Experience?" Why do customers exit? Offer? Performance? FuncHonal errors? Technical errors? Design errors? Business impact? In affected users? In missed transacKons? In €?
- 142. Online Success requires a complete picture" IT OperaKons: ? Web AnalyKcs: System Management Conversion CPU, Memory Pageviews Quality of Service Visitors
- 143. Online Success requires a complete picture" IT OperaKons: Web AnalyKcs: System Management Conversion CPU, Memory Pageviews Quality of Service Visitors Web Performance AnalyKcs: Assess impact of performance on user behaviour Tie customer saHsfacHon to quality of delivery QuanHfy revenue risked
- 144. The Web Performance Challenge: Balancing Cost vs. Performance Revenue" Conversion Optimization" Operations" Development"
- 145. The Web Performance Challenge: Balancing Cost vs. Performance Revenue" Conversion Optimization" Operations" Development" Am I opHmizing for performance AND cost?
- 146. The Web Performance Challenge: Balancing Cost vs. Performance Revenue" Are these the most cost effecHve improvements? Conversion Optimization" Operations" Development" Am I opHmizing for performance AND cost?
- 147. The Web Performance Challenge: Balancing Cost vs. Performance Is my performance improvement generaHng conversion? Revenue" Are these the most cost effecHve improvements? Conversion Optimization" Operations" Development" Am I opHmizing for performance AND cost?
- 148. Creating the Big Picture: Break down the Web Application Delivery Chain…! Management Tool! Internet! Load Web App DB! Client! balancer! server! server!
- 149. Creating the Big Picture: ….and relate Customer Experience….. ! Type of visitors Browser type Application Infrastructure Performance Performance Management Tool! Internet! Load Web App DB! Client! balancer! server! server!
- 150. Creating the Big Picture: …to revenue risked! Type of visitors Browser type Application Infrastructure Performance Performance Management Tool! Internet! Load Web App DB! Client! balancer! server! server! Conversion Speed/Availability Functional Technical Quality Quality & Capacity
- 151. Make your own business case" Current daily orders" (! *! Increased" conversions" +! Increased order value" )! ROI =! (days)" Cost of performance enhancement" 151!
- 152. Make your own business case" (! )! $10,000! *! 0.1607! +! 0.0551! $2,158! 23.17 =! =! days" $50,000! $50,000! 152!
- 153. Actionable reporting:" Monetize your performance " Availability Speed Capacity User Experience SLA IT suppliers Shared KPI’s +/- Conversion
- 154. Summary"
- 155. 1. Create Shared Performance Objectives 2. Measure from End User perspective 3. Plan for Peak Crowds 4. Test and simulate performance 5. Create the BIG picture: Web Performance Analytics
- 156. Questions?!
- 157. Analytics Usability Performability (What did they do (How did they (Could they do on the site?)! interact with it?)! what they wanted to do?)! VOC Communilytics Competition (What were their (What were they (What are they motivations?)! saying?)! up to?)!
- 158. Thanks! More questions? M: jtjepkema@measureworks.nl T: @jeroentjepkema W: www.measureworks.nl