The state of third-party tag performance

The State of Third-Party Tag Performance Simon Hearne @SimonHearne Web Performance Solutions Engineer @ Akamai

We pitch ad blockers on speed “ “ “ Experience a cleaner, faster web AdBlock Plus Opera will speed you up on the web and make sure ads don’t distract you from what matters Opera ..control over ads and tracking technologies to speed up page loads Ghostery @SimonHearne

Users want speed Top Motivations for Using an Ad Blocker: 1. Too many ads (48%) 2. Irrelevant ads (47%) 3. Ads are intrusive (44%) 4. Viruses / bugs (38%) 5. Ads take up space (38%) 6. Improve page load speed (33%) 7. Avoid video ads (29%) 8. Privacy concerns (26%) 9. Preserve data budget (23%) 10. Preserve device battery (23%) @SimonHearne GlobalWebIndex Q3 2018

Coming Up: 1. The Importance of Speed 2. Getting Speed Data 3. Tags? 4. Tag Performance 5. The Impact of Ad Blockers @SimonHearne

Data Source 1 @SimonHearne Akamai mPulse

Data Source 2 @SimonHearne HTTP Archive

The Importance of Speed @SimonHearne

Speed matters The BBC loses an additional 10% of users for every additional second it takes to load Pinterest improved load time by 40% and saw 15% increase in SEO traf c and 15% increase in conversions The Financial times increased user engagement by 30% when they released the new, performance focused FT.com @SimonHearne wpostats.com/tags/engagement/

Speed Correlates with Engagement 10 Timer TTFB PLT TTI FCP TTVR DOM Ready FID 9 Session Length (pages) 8 7 6 5 4 3 2 1 0 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 First Page Speed (ms) @SimonHearne ~1Bn mPulse pageview beacons - early March 2019

Getting Speed Data @SimonHearne ⚡

Collecting Data Ad Hoc Developer Tools PageSpeed Insights WebPageTest @SimonHearne Synthetics Rigor Catchpoint SpeedCurve RUM Analytics? APM mPulse

Data Points @SimonHearne Resource Timing API Level 2 - W3.org

A Willing Victim @SimonHearne

Gotchas 1/3 “ The Timing-Allow-Origin response header speci es origins that are allowed to see values [..] which would otherwise be reported as zero due to cross-origin restrictions. @SimonHearne Timing-Allow-Origin

Gotchas 2/3 “ @SimonHearne all content loaded from cross-origin IFRAMEs (and below) are completely hidden to ResourceTiming. Cross-origin IFRAME opting-in to sharing ResourceTiming data

Gotchas 3/3 Ad Block detection on one national publisher, with a narrow demographic. @SimonHearne

Tags? @SimonHearne

Tags! @SimonHearne

1/4 of web pages have >90% third-party requests @SimonHearne HTTP Archive

@SimonHearne

Tags serve business goals Measurement & Analytics Personalization & Testing Comments & Live Chat Tag Management Ads & Retargeting @SimonHearne

“Who put Facebook on the site?!” @SimonHearne

Tags can get out of control 352 Requests 147 Domains 2,300kB (1,700kB tags) @SimonHearne RequestMap

Tag managers are enablers @SimonHearne

Ad Networks are greedy @SimonHearne

So what? ¯_(ツ)_/¯ @SimonHearne

Tag Performance @SimonHearne

What’s the Risk? SPOF @SimonHearne

What’s the Risk? OnLoad SPOF @SimonHearne WebPageTest.org

800 600 400 200 0 Host p90 Load Time st0.mngbcn.com lsco.scene7.com cdn.dynamicyield.com res.cloudinary.com cdn.optimizely.com js-agent.newrelic.com apps.bazaarvoice.com lptag.liveperson.net fast.fonts.net analytics.snidigital.com cdn.gubagoo.io s.yimg.com ajax.googleapis.com service.maxymiser.net mc.yandex.ru dmp.theadex.com ing-district.clicktale.net secure-ds.serving-sys.com cdn.taboola.com www.youtube.com static.ads-twitter.com display.ugc.bazaarvoice.com ad.doubleclick.net vars.hotjar.com secure.adnxs.com www.jumpstarttaggingsolutions.com as-sec.casalemedia.com secure.widget.cloud.opta.net script.hotjar.com sslwidget.criteo.com va.v.liveperson.net www.google.co.jp cm.everesttech.net siteintercept.qualtrics.com ssl.google-analytics.com t.co static.hotjar.com ib.adnxs.com px.moatads.com c.evidon.com fundingchoicesmessages.google.com cdnssl.clicktale.net client.schwab.com static.foxnews.com vinlens.com tag.contactatonce.com www.gstatic.com use.typekit.net cdnjs.cloud are.com s.thebrighttag.com fastlane.rubiconproject.com pagead2.googlesyndication.com st.mngbcn.com static.criteo.net adserver-us.adtech.advertising.com trc.taboola.com adservice.google.com logx.optimizely.com cdn.krxd.net s.pinimg.com c.amazon-adsystem.com sb.scorecardresearch.com async-px.dynamicyield.com dpm.demdex.net maps.googleapis.com cdn.ampproject.org image.s5a.com network.bazaarvoice.com fonts.googleapis.com pubads.g.doubleclick.net s7d5.scene7.com nexus.ensighten.com tpc.googlesyndication.com www.googletagservices.com translate.googleapis.com fonts.gstatic.com stats.g.doubleclick.net ct.pinterest.com www.googleadservices.com securepubads.g.doubleclick.net s.btstatic.com bat.bing.com pixall.esm1.net tags.tiqcdn.com www.googletagmanager.com googleads.g.doubleclick.net www.google.com assets.adobedtm.com connect.facebook.net www.facebook.com www.google-analytics.com @SimonHearne 1,000 Volume Lucky they’re always fast, right? 90th Percentile Load Time & frequency for top 100 third-party hosts. n=1,080M 1,800 1,600 1,400 1,200

What’s the Risk? Interactivity @SimonHearne

The Impact of Ad Blockers @SimonHearne

Positive, right? @SimonHearne

A simple test @SimonHearne webtest.app

What about the wild? @SimonHearne

Results 8.98% pageviews with ads blocked (n=50M) @SimonHearne Metric With Ads Without Ads Difference Page Load Time (ms) 6,130 4,600 -25% Time to Interactive (ms) 5,480 4,260 -22% LongTasks Time (ms) 1,270 241 -81% First Input Delay (ms) 86 8 -93% Page Size (MB) 2.69 1.99 -26% Bounce Rate 68.8% 67.0% -2% Time on Site (s) 410 708 +72% Session Length (pages) 2.80 4.41 +58% JavaScript Errors per Page 1.92 0.18 -91%

Conclusions Third-party tags are pernicious Ad Blockers signi cantly improve performance Users with Ad Blocking enabled engage more @SimonHearne

Thank you shearne@akamai.com @SimonHearne simonhearne.com @SimonHearne simonhearne.com/presentations/ad-block-perf/