THE EMERGING FUTURE OF
SEARCH
Tom Anthony
Head of R&D
@TomAnthonySEO
Will Critchlow
Founder & CEO
@WillCritchlow
I WANT TO TALK ABOUT A SHIFT AS BIG AS MOBILE
(which is pretty big:
Google sees billions of
searches a day)
(and mobile has
probably doubled that)
There's another
multiple-billion
searches per day
opportunity out there
and it's
"searches people
aren't yet doing"
‘DAILY INFORMATION NEEDS’ STUDY
GOOGLE ASKED:
“What did you want to know recently?”
THEY FOUND:
25,000 needs
21 broad categories
39% of needs:
"participants were
unable to address the
need"
Why needs were unaddressed %
Did not know how to find the answer 27
Insufficient time 25
Need not important 15
No internet access 12
Other e.g. driving 21
15% WERE UNIMPORTANT
So 1/3 of all important
information needs are
left unfulfilled
(85% of 39%)
POWER AND CONVENIENCE MEAN INCREMENTAL SEARCHES
X-RAY
FORCE CLICK
LONG CLICK
EVEN BEFORE WE GET TO PASSIVE VOICE
GOOGLE PUBLICATION ON IPAS
Source: http://dl.acm.org/citation.cfm?id=2685309
EXAMPLE #1: CONTEXTUAL ASSISTANCE
“The personal assistant should be able
to detect a change in the user context,
and make useful recommendations for
that context.”
EXAMPLE #2: FULLY PERSONALIZED
“when suggesting restaurants, in addition
to the spatio-temporal context, the
assistant personalizes the suggestions
based on the user’s cuisine preferences and
price sensitivity.”
WE TALK ABOUT CONTEXT A LOT
“A context is a set of user actions
corresponding to a single
information need.”
GOOGLE’S DEFINITION OF CONTEXT
EXAMPLES OF “EXTENDED CONTEXTS”
planning a trip
to Paris
organising a
wedding
buying a
camera
CONTEXTS: NOT ALWAYS ABOUT NOW
Half of search queries belonged to
contexts lasting longer than a month.
“top 10 things to do in Berlin”
“top 10 things to do in Paris”
“top 10 things to do in Zagreb”
“weather in Paris”
“easyjet Paris”
“louvre”
“airports in France”
“top 10 things to do in Berlin”
“top 10 things to do in Paris”
“top 10 things to do in Zagreb”
“weather in Paris”
“easyjet Paris”
“louvre”
“airports in France”
“top 10 things to do in Berlin”
“cheapest vacation deals Paris”
“best coffee near the Champs-Élysées”
“moulin rouge”
“Eiffel Tower entry price”
“top 10 things to do in Paris”
“top 10 things to do in Zagreb”
“weather in Paris”
“easyjet Paris”
“louvre”
“airports in France”
“top 10 things to do in Berlin”
“cheapest vacation deals Paris”
“best coffee near the Champs-Élysées”
“moulin rouge”
“Eiffel Tower entry price”
“top 10 things to do in Paris”
“top 10 things to do in Zagreb”
2nd February
15th February
“weather in Paris”
“easyjet Paris”
“louvre”
“airports in France”
“top 10 things to do in Berlin”
“cheapest vacation deals Paris”
“best coffee near the Champs-Élysées”
“moulin rouge”
“Eiffel Tower entry price”
“top 10 things to do in Paris”
“top 10 things to do in Zagreb”
CONTEXT:
Planning a
trip to
Paris
2nd February
15th February
UPDATING YOUR CONTEXT IN REAL-TIME
“best time to visit France”
“cheap flights to Paris”
“what to see in Paris”
GOOGLE CAN LEARN CONTEXTS
“best time to visit France”
“cheap flights to Paris”
“what to see in Paris”
GOOGLE CAN LEARN CONTEXTS
Google might anticipate your searches
weeks before you do them.
GOOGLE TRENDS: [CHATBOTS]
YOU PROBABLY SAW ALL THE BUZZ AT F8
And you can't help but have
noticed that the 5 biggest tech
companies have all introduced an
Intelligent Personal Assistant.
FACEBOOK ‘M’
AMAZON ECHO
HOW DO YOU FIND OUT WHAT YOU CAN ASK?
http://gizmodo.com/facebook-messenger-chatbots-are-more-frustrating-than-h-1770732045
HOW DO YOU EVEN FIND THE BOTS?
We're back to Yahoo! Directory style:
chatbots >> news >> regional >> US >> MA >>
In other words, we still
need search and we still
need discoverability
But there is the promise
of data-driven search
powered by intent
Search that considers:
what I mean,
not what I say
and finds results based on:
what I can do,
not what text is on the page
OBLIGATORY RANKBRAIN REFERENCE
HUMMINGBIRD:RANKBRAIN AS CAFFEINE:PANDA
CREDIT: MIKEBAIRD ON FLICKR
We also see it showing
up in the human quality
rater guidelines
Supplementary Content (SC) is:
Features designed to help shoppers
find other products they might also like
a feature to multiply or divide the
recipe to make the right quantity of
food for a given number of people
[content] which contributes to a
satisfying user experience on the page
and website
Google removed the
section referring to
"helpful SC"
http://www.thesempost.com/updated-google-quality-rater-guidelines-eat/
what I can do,
not what text is on the page
IT ALSO PROBABLY HAS QUITE A LOT TO DO WITH MOBILE
This isn't the algorithm.
It's just the quality rater guidelines.
www.distilled.net/odn/
ONE OF THE REASONS WE BELIEVE IN SPLIT-TESTING FOR SEO
Context
Current
Situation
Time
Personalisation
Search History
Current
Location
Mode of Travel
Likes/Dislikes
Schedule
Companions
Query
Ancestor
Queries
Implicit query
aspect
Explicit query
aspect
Keyword
Spoken phrase Image
Search ‘mode’
(regular / maps / images)
Search
settings
Device
Anticipated
Situation
Past activities
Estimated
Intent
Context
Current
Situation
Time
Personalisation
Search History
Current
Location
Mode of Travel
Likes/Dislikes
Schedule
Companions
Query
Ancestor
Queries
Implicit query
aspect
Explicit query
aspect
Keyword
Spoken phrase Image
Search ‘mode’
(regular / maps / images)
Search
settings
Device
Anticipated
Situation
Past activities
Estimated
Intent
1
IMPLICIT QUERY SIGNALS
1
search query
“london tube stations”
explicit aspect of query
implicit aspect of query
iPhone user, on street in London
“london tube stations”
IMPLICIT SIGNALS: MODE OF TRANSPORT
TIME
TOTALSIGNALINFORMATION
EXPLICIT SIGNAL
IMPLICIT SIGNAL
My vision when we started Google 15 years ago was
that eventually you wouldn't have to have
a search query at all.
SERGEY BRIN
GOOGLE
IT'S BY NO MEANS EASY
On the same day in April, Kobe played his
last game
IT'S BY NO MEANS EASY
...and the Warriors won their 73rd game of
the season
THE TWITTER MOMENTS WERE GREAT
...and Twitter knows so much about me and
my interests
IMPLICIT SPORTS SEARCH SHOULDN'T BE HARD
IMPLICIT SPORTS SEARCH SHOULDN'T BE HARD
SOCCER
IMPLICIT SPORTS SEARCH SHOULDN'T BE HARD
SOCCER
SOCCER
IMPLICIT SPORTS SEARCH SHOULDN'T BE HARD
SOCCER
SOCCER
SOCCER
IMPLICIT SPORTS SEARCH SHOULDN'T BE HARD
SOCCER
SOCCER
SOCCER
SOCCER
COMPOUND QUERIES
2
Query
Algorithm
Response
Revised
Query Algorithm
Response
OLD MODEL: INDEPENDENT QUERIES
HUMMINGBIRD
CREDIT: MIKEBAIR
brought to you by…
https://youtu.be/JfJOFvKDNu4
NEW MODEL: DEPENDENT QUERIES
Query
Algorithm
Response
Additional
Input
Algorithm
Revised
Response
https://youtu.be/M1ONXea0mXg
KEYWORDS
TO INTENTS
3
SOURCE: /U/UPBOATOARKNOTUP
SEARCH HISTORY OF A 5 YEAR OLD
KEYWORDS
INTENT
IMPLICIT
SIGNALS
COMPOUND
QUERIES
CONTEXTUAL SEARCHES
http://selnd.com/1MnLlF8
A NOTE OF CAUTION:
You can't yet assume
Google is ranking on
intent alone
SEE THIS POST BY MY COLLEAGUE @SAMNEMZER
http://dis.tl/keyword-data
METHODOLOGY
DIFFERENT KEYWORDS STILL TREATED DIFFERENTLY
IT'S NOT JUST REORDERING
1
PERSONAL DEVICE
4
Screen size and keyboard type
are not important.
It can take
photos
It can take
photos
It knows
where it is
It can take
photos
It knows
where it is
It knows who
your friends are
It can take
photos
It knows
where it is
It knows who
your friends areIt knows
if you are
walking,
running,
etc.
It’s personal,
signed-in,
and taken
everywhere
It’s personal,
signed-in,
and taken
everywhere
It
interacts
with the
world
(beacons,
pay, NFC)
It’s personal,
signed-in,
and taken
everywhere
It has
natural
m
ulti-
touch
It
interacts
with the
world
(beacons,
pay, NFC)
It’s personal,
signed-in,
and taken
everywhere
It has
natural
m
ulti-
touch
It has
notificationsIt
interacts
with the
world
(beacons,
pay, NFC)
It's actually the PC that has the limited, basic, cut-down
version of the internet… it only has the web.
‘mobile first’
via Ben Evans
DATA DRIVEN SEARCH
5
SHIFT AWAY FROM ‘WEB SEARCH’: DIRECT ANSWERS
SHIFT AWAY FROM ‘WEB SEARCH’: GOOGLE NOW
OTHER IPAS ALSO GIVE DIRECT ANSWERS
REFERENCE: 1, 2, 3
INCREASINGLY ANSWERS NEED TO BE DYNAMIC
Time Based Location Based
GOOGLE NEED DATA FOR THESE QUERIES
Knowledge
Graph
Data
Partnerships
Everything Else
?
3RD PARTY INTEGRATION WITH IPAS
(PROBABLY)
FACETED SEARCH: CROSS DEVICE CARDS?
CHALLENGES
& TIPS
How do we check Rankings?
How does Attribution work?
Will there be Analytics?
RANK #0
http://www.slideshare.net/crumplezone/seo-for-answers-ranking-0
STAT RESEARCH: http://bit.ly/STAT-featured-snippets
OPERATE EARLIER IN THE FUNNEL
http://www.slideshare.net/jonoalderson/searching-higher-up-the-funnel-54123155
POP QUIZ
How many web searches have I shown
today?
POP QUIZ
How many web sites have I shown today?
CLUE:
By my count, I've shown you 25 search
screens, apps, bots or devices
ANSWER:
I showed 2 regular web searches
ANSWER:
(when I was showing that conversational
search doesn't work there)
ANSWER:
...and ZERO websites
POWER AND CONVENIENCE MEAN INCREMENTAL SEARCHES
This is where the next trillion searches / year
will come from
COMPOUND
QUERIES
IMPLICIT
SIGNALS
PERSONAL
ASSISTANTS
DATA
DRIVEN
SEARCH
INTENTSKEYWORDS
PERSONAL
INDEX &
DEVICE
AND THIS IS HOW THEY'LL BE POWERED
THANKS
@TOMANTHONYSEO
@WILLCRITCHLOW
LATEST DISTILLED NEWS:
TALK TO US ABOUT OUR ‘OPTIMIZATION DELIVERY NETWORK’:
https://www.distilled.net/odn/
THE ODN ALLOWS SEARCH-FOCUSED SPLIT TESTING ON YOUR
WEBSITE
REGARDLESS OF YOUR TECHNOLOGY STACK.
This is @TomAnthonySEO
We collaborated on a lot of this thinking
Image credits:https://www.flickr.com/photos/daviderickson/4114851120/
https://www.flickr.com/photos/accidentdesigns/14799315568/

SearchLove Boston 2016 | Will Critchlow | The Emerging Future of Search