1 of 190

what we talk about�when we talk about

WEB PERFORMANCE

May 2017, Dusseldorf

2 of 190

3 of 190

design & front-end code, focus on PE & RWD

WE DO

4 of 190

everyone

MAKE THE WEB WORK FOR

5 of 190

founded 2001

170+ WEB SITE/APP PROJECTS FOR 80 CLIENTS

6 of 190

2010 to today

40+ RESPONSIVE SITES/APPS WITH 30 CLIENT TEAMS

7 of 190

we’ve had lots of conversations about

HOW TO APPROACH WEB DESIGN & DEV

8 of 190

Average web page size more than tripled in the past 5 years

http://httparchive.org/trends.php?s=All&minlabel=Jul+1+2011&maxlabel=Dec+2+2016#bytesTotal&reqTotal

9 of 190

https://aasaindustryanalysis.wordpress.com/2015/08/04/smartphone-market-continues-upswing-in-q2/

10 of 190

http://www.google.com/publicdata/explore?ds=d5bncppjof8f9_#!ctype=l&strail=false&bcs=d&nselm=h&met_y=it_net_user_p2&scale_y=lin&ind_y=false&rdim=region&idim=region:EAS:ECS:LCN:MEA:NAC:SAS:SSF&ifdim=region&hl=en_US&dl=en_US&ind=false

11 of 190

iOS mobile landscape, Sept 2014-present

12 of 190

Samsung touchscreen devices, 2014

13 of 190

Android mobile landscape, 2012

http://opensignal.com/reports/fragmentation/

14 of 190

Android mobile landscape, 2013

http://opensignal.com/reports/fragmentation-2013/

15 of 190

Android mobile landscape, 2014

http://opensignal.com/reports/2014/android-fragmentation/

16 of 190

Android mobile landscape, 2015

https://opensignal.com/reports/2015/08/android-fragmentation/

17 of 190

AT&T 4G ‘plan’ 2014

http://arstechnica.com/business/2014/04/fcc-chairman-regrets-that-att-and-verizon-control-the-best-spectrum/

18 of 190

AT&T coverage map 2016

19 of 190

Verizon 4G coverage map 2013

20 of 190

Verizon coverage map 2016

21 of 190

22 of 190

23 of 190

so much data says

BIGGER (SCREENS), �BETTER (NETWORKS), �MORE (PIXELS)

24 of 190

so, of courseBIGGER-BETTER-MORE WEBSITES

http://httparchive.org/trends.php?s=All&minlabel=Jul+1+2011&maxlabel=Dec+2+2016#bytesTotal&reqTotal

25 of 190

so where’s the problem?

26 of 190

https://twitter.com/jaffathecake/status/709411154150473728

27 of 190

28 of 190

common misperception that slow networks are a

THIRD WORLD PROBLEM

29 of 190

https://twitter.com/BenedictEvans/status/513017790920290304

30 of 190

http://www.businessinsider.com/facebook-2g-tuesdays-to-slow-employee-internet-speeds-down-2015-10

“Facebook … will give employees an opportunity to see what using the app with an incredibly slow connection feels like and help close the "empathy gap" between Silicon Valley and emerging markets.”

31 of 190

“Engineers … testing their apps on a 2G Nigerian network … [or] an overloaded network in Jakarta … at the first Internet.org hackathon.”

32 of 190

https://twitter.com/stephenpeas/status/708759155150966785

33 of 190

too easy for clients �to think

“NOT REALLY OUR PROBLEM”

34 of 190

except...

35 of 190

except...

2G NETWORKS ARE NOT EXCLUSIVE TO THE THIRD WORLD

36 of 190

except...

ALL NETWORK SPEED DATA REFLECT THE BEST CASE SCENARIO

37 of 190

“Map displays approximate outdoor coverage and coverage may vary. Coverage is not guaranteed and is subject to change without notice.”

38 of 190

“These maps are not a guarantee of coverage and contain areas of no service… Wireless service is subject to network and transmission limitations…”

39 of 190

“In mature markets (like the US), LTE’s median speeds have become as much as 50% slower within the last year.”�

“fast” networks don’t always deliver

http://twinprime.com/lte-has-slowed-by-50-in-the-us/

40 of 190

https://www.ericsson.com/assets/local/mobility-report/documents/2016/ericsson-mobility-report-november-2016.pdf

41 of 190

https://www.ericsson.com/assets/local/mobility-report/documents/2016/ericsson-mobility-report-november-2016.pdf

42 of 190

https://www.ericsson.com/assets/local/mobility-report/documents/2016/ericsson-mobility-report-november-2016.pdf

43 of 190

https://www.ericsson.com/assets/local/mobility-report/documents/2016/ericsson-mobility-report-november-2016.pdf

44 of 190

and we know

SPEED ISSUES ARE �NOT EXCLUSIVE TO LOWER END NETWORKS

45 of 190

“Website visitors tend to care more about speed than all the bells and whistles we want to add to our websites”

https://blog.kissmetrics.com/loading-time/?wide=1

46 of 190

https://webmasters.googleblog.com/2015/04/mobilemadness-campaign-to-help-you-go.html

47 of 190

In early 2015,The National Journal tested download speeds on the Acela AmtrakConnect WiFi service

Green = broadband

Red = 0 - .9 Mbps.

Amtrak WiFi: So Slow It Might As Well Not Exist

https://consumerist.com/2015/04/23/amtrak-wifi-so-slow-it-might-as-well-not-exist/

48 of 190

49 of 190

The web’s true potential for democracy, economic growth and human creativity is only just beginning to be glimpsed...

�- Sir Tim Berners-Lee

http://www.theworldin.com/article/10648/wider-world-web

50 of 190

In 2016 all of us must protect and enhance this public space for the benefit of all humankind.

�- Sir Tim Berners-Lee

http://www.theworldin.com/article/10648/wider-world-web

51 of 190

we need data and tools to convince

‘NOT REALLY OUR PROBLEM’ SKEPTICS

52 of 190

https://www.soasta.com/blog/author/teverts/

53 of 190

54 of 190

http://www.pewinternet.org/

55 of 190

56 of 190

http://www.webperformancetoday.com/2012/03/21/neuroscience-page-speed-web-performance/

57 of 190

Sensory memory...works in 100ms bursts. Every time you see something, this visual information is taken in....

Your short-term memory can store information for 10-15 seconds, at most, enough time for your working memory to process, manipulate, and control it.

http://www.webperformancetoday.com/2012/03/21/neuroscience-page-speed-web-performance/

58 of 190

We’ve evolved to perform actions in beautiful sequential series of minute actions that flow more or less seamlessly. We aren’t wired to deal with the fits and starts of human-computer interaction.”

http://www.webperformancetoday.com/2012/03/21/neuroscience-page-speed-web-performance/

59 of 190

“Delays in web page load times and video streaming can be as stressful as watching a horror movie.

Poor network performance increases smartphone users’ heart rate and stress levels

https://www.ericsson.com/res/docs/2016/mobility-report/ericsson-mobility-report-feb-2016-interim.pdf

60 of 190

“Back as far as Robert B. Miller’s classic 1968 paper “Response Time in Man-Computer Conversational Transactions“, studies have found that people are—and always have been—most comfortable, most efficient, and most productive with response times of less than 2 seconds.

Tammy Everts, “Time Is Money: The Business Value of Web Performance” (O’Reilly, 2015)

61 of 190

Analysis of 12 top retailers that cater to the back-to-school shopping crowd, 500 million user sessions

http://www.soasta.com/blog/back-to-school-website-performance-monitoring/

SEPTEMBER 30, 2015

62 of 190

Smartphones dominate retail traffic, doubling in one year

25% DESKTOP

75% PHONE/TABLET

60% DESKTOP

40% PHONE/TABLET

2015

2014

http://www.soasta.com/blog/back-to-school-website-performance-monitoring/

63 of 190

Between 2014 and 2015, the optimal load time for peak conversion shifted from 3.8 seconds to 2.4 seconds

SOASTA 2016 Holiday Retail Insights Report, September 2016�https://www.soasta.com/wp-content/uploads/2016/10/19036-Holiday-Retail-Insights-Report-PQ.pdf

64 of 190

In 2015, most users experienced page loads between 2.2-4.2 sec, compared to 2014 when the bulk experienced load times between 3-5.8 sec

SOASTA 2016 Holiday Retail Insights Report, September 2016�https://www.soasta.com/wp-content/uploads/2016/10/19036-Holiday-Retail-Insights-Report-PQ.pdf

65 of 190

In 2014, converted shoppers fell neatly within the sweet spot

SOASTA 2016 Holiday Retail Insights Report, September 2016�https://www.soasta.com/wp-content/uploads/2016/10/19036-Holiday-Retail-Insights-Report-PQ.pdf

66 of 190

In 2015, there is a pronounced gap between the bulk of converted traffic and the 2.4-second peak conversion rate

SOASTA 2016 Holiday Retail Insights Report, September 2016�https://www.soasta.com/wp-content/uploads/2016/10/19036-Holiday-Retail-Insights-Report-PQ.pdf

67 of 190

TOP SITES KNOW: Performance is critical

page load +500ms = 3% drop in trafficpage load +1000ms = 6% drop

page load +100ms = 1% loss in sales

page load +500ms = 25% fewer searches

page load +400ms = 5–9% drop in traffic

68 of 190

Pew Research Center study of American device ownership trends

Multi-wave, demographically diverse

2014-15

http://www.pewinternet.org/files/2015/10/PI_2015-10-29_device-ownership_FINAL.pdf

69 of 190

Pew Research Center detailed research on �U.S. smartphone ownership and usage

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

70 of 190

Cell phones, and especially smartphones & tablets are growing dramatically.

Other devices (laptops, e-readers) flat or declining

http://www.pewinternet.org/files/2015/10/PI_2015-10-29_device-ownership_FINAL.pdf

71 of 190

77% of cell internet users say they experience slow download speeds that prevent things from loading as quickly as they would like.

72 of 190

Among our 5,400 panelists, approximately �15% do not have Javascript-enabled devices.

Building Pew Research Center’s American Trends Panel,” April 8, 2015

http://www.pewresearch.org/2015/04/08/results-2/#javascript

73 of 190

Fully 68% of adults now have a smartphone, nearly double the share that Pew Research Center measured in its first survey on smartphone ownership in mid-2011.

74 of 190

19% of American adults are “smartphone dependant” - i.e., they have a smartphone, but limited or no other internet access

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

75 of 190

The trend toward mobile-only access is growing, especially for younger, minority, and lower income populations

http://www.pewinternet.org/2015/12/21/home-broadband-2015/

76 of 190

37% of smartphone owners reach the maximum amount of data on their plan at least on occasion.

Fully 15% say that this happens “frequently.”

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

77 of 190

48% of smartphone-�dependent individuals have had to cancel or suspend service due to financial constraints.

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

78 of 190

79 of 190

http://www.cbsnews.com/news/apple-iphone-wi-fi-assist-blamed-for-teens-2000-phone-bill/

“It turns out Ashton Finegold's bedroom is one such place with a weak signal.

So while he thought he was still connected to his home Wi-Fi while surfing the web in his room, his iPhone was eating up more than 144,000 MB of data..”

80 of 190

terrific, if you

HAVE ABUNDANT / UNLIMITED DATA

WANT THIS DECISION MADE FOR YOU

81 of 190

those are

BIG ASSUMPTIONS

82 of 190

https://twitter.com/roy/status/861563195424681985

83 of 190

For context, in 2014, 51.4% of American wage earners received compensation of �less than $30,000

84 of 190

62% of smartphone owners have used their phone in the past year to look up information about a health condition

85 of 190

  • 2.0mb page, 103 requests
  • 19 second full page load on DSL
  • 7 seconds of blank screen
  • JS-only navigation

86 of 190

  • 1.6mb page, 94 requests
  • 21.5 second page load on 3G
  • 11.8 seconds of blank screen
  • UI not properly optimized for mobile (overlapping buttons, illegible text)

http://www.webpagetest.org/result/160316_C2_1B8A/

87 of 190

88 of 190

68% of smartphone owners use their phone at least occasionally to follow along with breaking news events, with 33% saying that they do this “frequently.”

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

89 of 190

  • 3.8mb page, 604 requests
  • 34.5 second full page load on DSL
  • 2.5 seconds for text, 4.5 for images

http://www.webpagetest.org/result/160318_W2_153W/

90 of 190

  • 2.3mb page, 421 requests
  • 43 second full page load on 3G
  • 7.4 seconds of blank screen before initial view
  • 10.5 seconds to initial text content

http://www.webpagetest.org/result/160318_PQ_15FR/

91 of 190

92 of 190

93 of 190

“Visiting the home page of Boston.com every day for a month would cost the equivalent of about $9.50 in data usage just for the ads.”

http://www.nytimes.com/interactive/2015/10/01/business/cost-of-mobile-ads.html?_r=1

94 of 190

95 of 190

Some 28% of Americans—including 53% of 18- to 29-year-olds—have used a smartphone in one way or another as part of a job search

http://www.pewinternet.org/2015/04/01/us-smartphone-use-in-2015/

96 of 190

  • 2.1mb page, 131 requests
  • 16 second full page load on DSL
  • 5.5 seconds of blank(-ish) screen
  • JS-only navigation, content

http://www.webpagetest.org/result/160322_Y4_12G1/#run4

97 of 190

  • 2.6mb page, 219 requests
  • 36.1 second page load on 3G
  • 7.7 seconds of blank screen before initial view

http://www.webpagetest.org/video/compare.php?tests=160317_E7_10KB-r:1-c:0

98 of 190

Fully 53% of smartphone owners indicate that they have been in an emergency situation where having their phone available helped resolve the situation

http://www.pewinternet.org/2015/12/21/home-broadband-2015/

99 of 190

  • 2.1mb page, 72 requests
  • 14.6 second full page load on DSL
  • 4 seconds of blank screen

100 of 190

  • 1.2mb page, 94 requests
  • 24.1 second page load on 3G
  • 7.1 seconds of blank screen before initial view

http://www.webpagetest.org/video/compare.php?tests=160317_E7_10KB-r:1-c:0

101 of 190

  • 1.9mb page, 111 requests
  • 14.2 second full page load on DSL
  • 3.5 seconds of blank screen

http://www.webpagetest.org/result/160318_7P_1CSD/

102 of 190

  • 1.24mb page, 69 requests
  • 16.5 second page load on 3G
  • 8 seconds of blank screen before initial view

http://www.webpagetest.org/result/160322_70_11P8/

103 of 190

without JavaScript

with JavaScript

Completely non-functioning navigation without Javascript enabled

104 of 190

sobering data emerging about

ENVIRONMENTAL IMPACTS

105 of 190

http://www.theatlantic.com/technology/archive/2013/04/cell-networks-are-energy-hogs/274961/

Cell Networks Are Energy Hogs

“Wireless �infrastructure, �not data centers, �may prove to be �the larger energy �consumer.”

106 of 190

“...energy consumption associated with 'wireless cloud' will ... increase in carbon footprint from 6 megatonnes of CO2 in 2012 up to 30 megatonnes in 2015, which is the equivalent of an additional 4.9 million cars on the road.”

107 of 190

IEA: $80 Billion in Power Wasted by Connected ‘Things’

http://www.forbes.com/sites/heatherclancy/2014/07/24/iea-80-billion-in-power-wasted-by-connected-things/#2c81a20374f7

108 of 190

these are big audiences

AND SERIOUS CONSEQUENCES

109 of 190

large web pages impose an

INVISIBLE BUT �VERY REAL COST

110 of 190

our design and �code decisions

NEED TO BE MINDFUL

111 of 190

the future is internet

THAT DELIVERS ON THE PROMISE IT STARTED WITH

112 of 190

113 of 190

(Special thanks to Eric and Sara for that reminder, and for so many other very good ideas)

114 of 190

how do we make our designs & code

CONSERVATIVE AND ACCEPTING?

115 of 190

filesize and �page speed

ARE DESIGN CHOICES

116 of 190

take a critical eye �to everything

PARE DOWN TO MOST ESSENTIAL, & OPTIMIZE

117 of 190

make speed part of the design process:

SET PERFORMANCE GOALS IN THE DESIGN SPEC

118 of 190

consider perceived performance:

RENDER SOMETHING USEFUL AS QUICKLY AS POSSIBLE

119 of 190

performance opportunities:

  • IMAGES, VIDEO, AUDIO
  • CUSTOM FONTS
  • THIRD-PARTY TOOLS (ADVERTISING, SOCIAL MEDIA, TRACKING)
  • CODE FRAMEWORKS
  • DATA NETWORKS

120 of 190

we can make performance a

COMPETITIVE ADVANTAGE

121 of 190

we can make performance a

PATH TO DEMOCRACY, GROWTH & CREATIVITY

122 of 190

#1:

OPTIMIZE THOSE IMAGES

123 of 190

FORMATS AND METHODS

VECTOR GRAPHICS (SVG)�Infinitely scalable, bitmap fallbacks

COMPRESSIVE IMAGES�2x bitmaps at high compression

RESPONSIVE IMAGES�Different images per breakpoint

124 of 190

svg

THE JOY OF

125 of 190

Scalable Vector Graphic (SVG) is an XML-based vector image format for two-dimensional graphics with support for interactivity and animation. The SVG specification is an open standard developed by the World Wide Web Consortium (W3C) ...since 1999.

126 of 190

FULL COLOR VECTORS

127 of 190

CSS STYLING

128 of 190

RESPONSIVE

OPTICAL CORRECTION

129 of 190

INTERACTIVE

130 of 190

Automate animated SVGs using Adobe AI layers

131 of 190

LEARN HERE

http://filamentgroup.github.io/layersnap/demo/

132 of 190

LEARN HERE

https://sarasoueidan.com/articles/

133 of 190

LEARN HERE

https://docs.google.com/presentation/d/1CNQLbqC0krocy_fZrM5fZ-YmQ2JgEADRh3qR6RbOOGk/edit#slide=id.p

134 of 190

bitmaps

BETTER

135 of 190

1x Display

2x Retina

Balancing Bitmap Quality

136 of 190

137 of 190

2x images are very large

~300% larger file at “standard” compression

55% Quality

55% Quality

Name gate_2x_55.jpg

Kind JPEG image

Size 43k

Name gate_1x_55.jpg

Kind JPEG image

Size 14k

138 of 190

Compressive images are much smaller

120-150% larger file than 1x at high compression

55% Quality

20% Quality

55% Quality

20% Quality

Name gate_2x_20.jpg

Kind JPEG image

Size 19k

Name gate_1x_55.jpg

Kind JPEG image

Size 14k

139 of 190

Compressive images

Over-sized and over-compressed JPEG images. This single image then works for both SD and HD screens with a simple img tag.

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

140 of 190

Typical settings

  • Size images 1.4-2x of expected size in context
  • Progressive checked
  • Adjust quality - usually 25-45% works
  • Touch of blur to help compression algo (but not enough to be visible)

141 of 190

For images, especially with Responsive, one size does not fit all

For larger image, consider multiples

640px in 768px viewport

0.8x

640px in 568px viewport

1.1x

142 of 190

600px

1200px

1800px

143 of 190

Responsive images

Dynamically load the appropriate image source based on image size in layout and/or screen density (DPI)

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

Typical export settings: 2x, Progressive JPEG, 20% quality, 0.1% blur

144 of 190

There are great resources, and a huge community, around image optimization and delivery:

https://responsiveimages.org/

145 of 190

#2:

OPTIMIZE ASSET LOADING

146 of 190

In addition to download time, JavaScript can be very slow to parse and execute on mobile devices.

PERFORMANCE ISSUE:

147 of 190

jQuery 2.1.1 Parse + Execution Times*

Macbook Air (2014), Chrome

Lumia 520 (WP8)

5ms

29ms

34ms

97ms

56ms

153ms

Android 2.3.6 Phone, Browser

168ms

484ms

652ms

JS Parse and Execution Time” Tim Kadlec, September 24, 2014�http://timkadlec.com/2014/09/js-parse-and-execution-time/

* jQuery is 82K (29K gzipped)

PARSE

EXECUTE

TOTAL

148 of 190

https://twitter.com/addyosmani/status/861166051517620225

149 of 190

Load times for 1MB of uncompressed JS over WebPageTest + real devices.

150 of 190

When anything over 100ms stops feeling instantaneous and anything over 1000ms breaks the users flow, taking 700ms to parse your JavaScript cripples the user experience before it really has a chance to get started.

JS Parse and Execution Time” Tim Kadlec, September 24, 2014�http://timkadlec.com/2014/09/js-parse-and-execution-time/

151 of 190

Many techniques for�SPEED OPTIMIZATION

  • Extract and Inline “Critical” CSS
  • Load full StyleSheets asynchronously (cached for return visits)
  • Load all JavaScript asynchronously (yes, ads and tracking too!)
  • Load fonts asynchronously, and apply them in a progressive manner

https://www.filamentgroup.com/lab/weight-wait.html

152 of 190

More weight doesn’t mean more wait�SAME ASSETS, SMARTER LOADING

https://www.filamentgroup.com/lab/weight-wait.html

153 of 190

  • Audit your JS: Is all this needed? For all pages?
  • Add a Cut the Mustard (CTM) test so scripts are only loaded in capable browsers
  • Ensure site works without JS. Render first. JS second.
  • Ensure 3rd party scripts (social, tracking) are non-blocking, and perhaps delayed to prioritize rendering

Optimize for speed & caching

JS RECOMMENDATIONS

154 of 190

155 of 190

Our “Cut the Mustard”

We use Enhance.js to provide a cleaner way to determine what devices should get the enhanced experience.

156 of 190

Making the cut

If a browser passes the test, it requests the enhanced experience JavaScript (including shoestring, our DOM Library and the component JS) and web fonts.

If it fails to cut the mustard, it isn’t encumbered with any extra JS or web fonts.

157 of 190

Custom fonts: how long will they wait?

When @font-face loading hangs, some browsers have long timeouts while they wait for the font. Your content will be invisible until...

158 of 190

Current: Flash of Invisible Text (FOIT)

The site now is unusable for longer than needed as we wait for custom fonts to download. A blank UI with invisible text feels slow to a visitor.

INVISIBLE TEXT WHILE LOADING

CUSTOM FONTS LOADED

CHALLENGE

159 of 190

Flash of Unstyled Text (FOUT)

For the first visit only: show a local fallback font so text is immediately visible and site is usable much faster on a 3G connection. Use cached fonts on subsequent views.

FALLBACK NATIVE FONT (GEORGIA + ARIAL) WHILE LOADING

CUSTOM FONTS LOADED

FONT RECOMMENDATION

160 of 190

Load fonts asynchronously

RECOMMENDATION

161 of 190

  • Minimize the number of fonts & weights (bold, italic, semibold) used
  • Subset fonts to only characters you need w/ FontSquirrel

Tuning Up Fonts

RECOMMENDATION

162 of 190

SHAMELESS PLUG:

Some great technique recommendations here:

http://filamentgroup.com/lab/performance-rwd.html

163 of 190

SHAMELESS PLUG:

This man has forgotten more about optimization that I will ever know.

164 of 190

https://www.filamentgroup.com/lab/mv-initial-load-times.html

MVC/JS framework first-load impacts

We know many sites want use frameworks to capitalize on their speed capabilites, but the first page load is a critical gating factor for many visitors.

Data about baseline load speeds is difficult to come by.

CHALLENGE

165 of 190

on-demand

THINK MORE ABOUT

166 of 190

Curbing wasted bandwidth

Many sites will load images, audio, video, JavaScript, and other resources that are:

  • never accessed by the visitor because they don’t scroll the page to see them
  • autoplayed when the visitor isn’t interested
  • autoplayed while out of view in the window, or in another tab

CHALLENGE

167 of 190

  • Load only the lightest initial assets for essential content and functionality
  • Lazy load “hidden” resources —carousels, long product grids—as the visitor explicitly asks/acts
  • Make all multimedia opt-in
  • For features inaccesible for some (images accessed on hover, menus) load based on capabilities test, or at time of request

Put the visitor in control

RECOMMENDATION

168 of 190

  • Super low-res images (~2-3k each) on initial page load
  • If JS is available, lazy-load in higher quality as page scrolls
  • For 48-product grid, changed initial page size from ~2mb to ~450k.
  • If visitor scrolls full page, sees the full 2mb, but only with that explicit action

DESKTOP WITH THROTTLED 2G CONNECTION

AN OPTIMIZATION EXPERIMENT

169 of 190

#3:

DON’T COMPROMISE ACCESSIBILITY

170 of 190

there’s no good reason

TO LIMIT ACCESS TO YOUR BASIC CONTENT

171 of 190

it ensures reach �and larger markets

AND COMMON COURTESY

172 of 190

would you refuse a customer because

THEY DROVE THE WRONG CAR? OR A BIKE? �OR TOOK THE SUBWAY?

173 of 190

$6,000,000�SETTLEMENT

$795,000�SETTLEMENT

http://www.wsj.com/articles/SB10001424127887324373204578374483679498140

174 of 190

Last year alone in �the US, there were >240 law suits against sites for accessibility problems, and thousands of notice letters.

http://www.karlgroves.com/2011/11/15/list-of-web-accessibility-related-litigation-and-settlements/

http://www.adatitleiii.com/2016/10/federal-website-lawsuits-spike-community-banks-get-demand-letters/

175 of 190

Progressive Enhancement

IS YOUR FRIEND

176 of 190

Progressive Enhancement

ENSURES BROAD ACCESS �AND FAULT TOLERANCE

177 of 190

Progressive Enhancement =

SEMANTIC MARKUP, �ADD FEATURES BASED ON CAPABILITIES

178 of 190

not all or nothing

THINKING DIFFERENTLY ABOUT CONTENT, SOURCE ORDER, STRUCTURE, INTERACTION

179 of 190

with JavaScript

without JavaScript

180 of 190

with JavaScript

without JavaScript

181 of 190

with JavaScript

without JavaScript

182 of 190

with JavaScript

without JavaScript

183 of 190

remember: �HTML is inherently accessible

JS & CSS CAN BREAK IT

184 of 190

HTML element semantics often fall short for describing complex UIs.

WAI ARIA is a W3C spec that gives us the semantic tools to describe these.

185 of 190

WAI-ARIA

  • Rolesheader, menu, tree, tab, grid
  • Statesexpanded, hidden, selected
  • Focus & keyboard management�for screen readers (and power users)

186 of 190

SHAMELESS PLUG:

lots of open-source PE-based components here:

http://filamentgroup.com/code/

187 of 190

SHAMELESS PLUG:

400+ pages of �Progressive Enhancement nerdery for the really curious

188 of 190

Progressive Enhancement

LIBERATES US TO FOCUS ON GREAT DESIGN OPPORTUNITIES

189 of 190

In 2016 all of us must protect and enhance this public space for the benefit of all humankind.

�- Sir Tim Berners-Lee

http://www.theworldin.com/article/10648/wider-world-web

190 of 190

@pattytoland

@filamentgroup

filamentgroup.com

THANKS.