Web marketers have never faced a more dangerous time to have their visitors hit that "back" button. Facebook, Google, and even Twitter are measuring engagement and punishing the sites and brands that lose too much of their audience to an "I'm outta here" click. We've moved from a world where conversion rate optimization happens only in the purchase funnel to one where converting from a 5-second visitor to a 45-second, more engaged visitor who leaves with an answer is hugely important.
In this presentation, Rand will show how Google's Rankbrain, Facebook's engagement algorithms, and other platforms' metrics are transforming this landscape and what web marketers can do to fight back.
[CXL Live 16] Fight Back Against Back by Rand Fishkin
1. Rand Fishkin, Wizard of Moz | @randfish | rand@moz.com
Fight Back Against Back
Why the back button has become web marketing’s
greatest enemy (and how to defeat it)
15. In 2012, Google Published a Paper About
How they Use ML to Predict Ad CTRs:
Via Google
16. 2012
“Our SmartASS system is a
machine learning system. It
learns whether our users are
interested in that ad, and
whether users are going to
click on them.”
17. By 2013, It Was
Something
Google’s Search
Folks Talked
About Publicly
Via SELand
18. In October of2015,
they finally revealed
RankBrain, an AI-
system input to the
search rankings
Via Bloomberg Business
19. As ML Takes Over More ofGoogle’s Algo, the
Underpinnings of the Rankings Change
Via Colossal
20. Google is Public About How They Use MLin
Image Recognition & Classification
Potential ID
Factors
(e.g. color, shapes, gradients,
perspective, interlacing, alt
tags, surrounding text, etc)
Training Data
(i.e. human-labeled images)
Learning
Process
Best
Match
Algo
21. Google is Public About How They Use MLin
Image Recognition & Classification
ViaJeff Dean’s Slides onDeep Learning; aMustReadforSEOs
22. Machine Learning in Search Could Work Like This:
Potential
Ranking Factors
(e.g. PageRank, TF*IDF,
TopicModeling, QDF,Clicks,
Entity Association, etc.)
Training Data
(i.e. good&badsearch results)
Learning
Process
Best Fit
Algo
23. Training Data
(e.g. goodsearch results)
This is a good SERP – searchers
rarely bounce, rarely short-
click, and rarely need to enter
other queries or go to page 2.
24. Training Data
(e.g. bad search results!)
This is a bad SERP –
searchers bounce often, click
other results, rarely long-
click, and try other queries.
They’re definitely not happy.
25. The Machines Learn to Emulate the Good Results &
Try to Fix or Tweak the Bad Results
Potential
Ranking Factors
(e.g. PageRank, TF*IDF,
TopicModeling, QDF,Clicks,
Entity Association, etc.)
Training Data
(i.e. good&badsearch results)
Learning
Process
Best Fit
Algo
26. Deep Learning is Even More Advanced:
Deansaysbyusingdeep
learning,they don’thave totell
thesystemwhata catis,the
machineslearn, unsupervised,
forthemselves…
28. Googlers Don’t Feed in Ranking Factors… The
Machines Determine Those Themselves.
Potential
Ranking Factors
(e.g. PageRank, TF*IDF,
TopicModeling, QDF,Clicks,
Entity Association, etc.)
Training Data
(i.e. goodsearch results)
Learning
Process
Best Fit
Algo
29. Last October, Google Finally Went Public
with Their Use of ML-Based RankBrain
Via Bloomberg Business
54. Ask “What Are All theNeeds ofThese Searchers?”
Then Serve As Many as Possible
Youmightbetryingtoselldesks,butsearchers
are seekinganswerstoallof theseandmore.
55. If the Competition Delivers Value to Searchers
Who Aren’t Buyers, But You Don’t…
58. Optimize the Title, Meta Description, & URL
a Little for Keywords, but aLot for Clicks
If yourank#3,buthave a higher-
than-averageCTRforthatposition,
youmightgetmoved up.
Via Philip Petrescu on Moz
60. Given Google Often Tests New Results Briefly on Page One…
ItMayBeWorth Repeated Publication onaTopic toEarn that High CTR
Shoot!My postonlymade itto#15…
PerhapsI’lltryagain ina fewmonths.
61. Driving Up CTR Through Branding Or
Branded Searches May Give An Extra Boost
63. With Google
Trends’ new, more
accurate, more
customizable
ranges, youcan
actually watchthe
effects ofevents
and adsonsearch
query volume
Fitbitwasrunningads onSunday
NFLgamesthatclearlyshowinthe
searchtrends data.
65. Better Content > More Content
A lotofSEOusedtobeaboutestablishing
authoritythroughbrutequantity,butPanda,
andnowRankbrain,are changingthat.
66. Better Social Shares > More Social Shares
Via Rand’s Facebook Page
WhenI have a successful
postonFacebook,itboosts
Facebook’slikelihoodto
showmy postsinthe future…
67. Better Social Shares > More Social Shares
High engagementgrows
my reach potential.
Lowengagementshrinks
my reach potential.
72. Speed, speed, and more speed
Delivers an easy, enjoyable experience on every device
Compels visitors to engage, share, &return
Avoids features that dissuade or annoy visitors
Authoritative, comprehensive content that’s uniquely
valuable vs.what anyone else in your space provides
The Marketer’s User Experience Checklist
73. Uniquely Valuable Content
Via R2D3
Lotsofarticlestrytoexplainmachine learning,butthis
oneSHOWShowitworksina wayanyonecangrasp.
82. Top-of-Funnel Content Can’t Be Used Solely
to Filter Out the Non-Customers
Tryingtorank w/content
thatonlyservesoneniche
ofyour searchaudience
may bea recipe for failure
83. Fighting Against Back Means Serving a
Broader Audience
AngelList’stoolmakes
salarycomparison
easy,fast,andservesa
huge rangeofroles,
locations,andmarkets
Via AngelList
84. Or, Getting More Precise with Your Search
Query -> Content Targeting
By targetinga less
competitive,lowervolume
query,Compasscanreach
theaudience they’reseeking
85. Either Way, Engagement Metrics on Content
Must Become KPIs
ImprovingPages/Sessionand
loweringBounceRateshould
probablyplay a “link-building-
like”role inyourSEOarsenal
86. Our Content CTAs Deserve to Be
Customized, Tested, & Refined
(just like conversion-focused landing pages)
e.g. IbetI couldmake a
betterCTA forthe
comparisontoolthan
this(whichlooksfar too
muchlike anad IMO)
Via Talkpay (Comparably’s Blog)