Google's Hummingbird update was a major rewrite of its search algorithm focused on understanding search queries through entities and relationships rather than just keywords. This has implications for e-commerce sites, which should optimize to be recognized as entities, build relationships between products and categories, and focus on answering queries through in-depth content rather than just keywords. They also need to optimize for voice and mobile search through measures like prioritizing local search results and ensuring their content is accessible across devices.
4. @KunleTCampbell
“…a new engine built on
both existing and new
parts, organized in a way
to especially serve the
search demands of today
(from mobiles), rather than
ten years ago”
What was
Hummingbird all
about?
7. @KunleTCampbell
What was Hummingbird all about?
Precision
& Speed
“answer your questions about the world”
- Tamar Yehoshua, VP, Search
Answers
Queries
Knowledge
Graph
Conversational
Queries
Voice
Search
Anticipate
Queries
Google Now
follow up
context queries
12. @KunleTCampbell
Context of a Query rather than String Match
Information Card
A major change in
the way Google
Interprets the way
we Search
Knowledge
Graph
19. @KunleTCampbell
Is Google not a
scraper site?
No…a search engine like
Google is ‘an amazing Swiss
Army Knife’ ;)
h#ps://www.youtube.com/watch?v=HViSQjZxhnY
23. @KunleTCampbell
Search used to be about
using queries that hopefully
matched content that was
out there…
Conversational Queries…
Search today is also
about asking complex
questions in a conversational
format with the hope of
getting a direct answer ?
Conversational
Queries
Voice
Search
26. @KunleTCampbell
Google is also teaching us a new set of commands
Read
more:
h#p://bit.ly/PQg3zq
Conversational
Queries
Voice
Search
27. @KunleTCampbell
As we learn these commands, Google might
better Anticipate our follow up Queries...
Anticipate
Queries
Google Now
follow up
context queries
28. @KunleTCampbell
Query Reviser Re-Writing Engine Based on
Identifying ENTITIES and the SYNONYM ENGINE...
Anticipate
Queries
Google Now
follow up
context queries
via:
h#p://www.seobythesea.com/2013/09/google-‐hummingbird-‐patent/
SYNONYM IDENTIFICATION
BASED ON CO-OCCURRING
TERMS
United States Patent: 8,538,984
Filled on: September 17, 2013
Assignee: Google Inc.
(Mountain View, CA)
http://1.usa.gov/1i900HL
29. @KunleTCampbell
The Vast and Ever Expanding Size of Knowledge Graph and
the Semantic Web is Constantly Improving Query Re-Writing
Anticipate
Queries
Google Now
follow up
context queries
via:
h#p://www.seobythesea.com/2013/09/google-‐hummingbird-‐patent/
SEARCH QUERIES IMPROVED
BASED ON QUERY SEMANTIC
INFORMATION
United States Patent: 8,577,907
Filled on: November 5, 2013
Assignee: Google Inc.
(Mountain View, CA)
http://1.usa.gov/1nosfFP
31. @KunleTCampbell
Advances with Google Now, shows
Google’s Ambitious long-term goal
of progressing from a search engine
to an ubiquitous artificial-
intelligence answer machine
Anticipate
Queries
Google Now
follow up
context queries
34. @KunleTCampbell
User Data from Query logs
Here’s How Google Attempts to Understand the‘Layers
of Context’in a Query
Search Entity information – Knowledge Graph
has 570 million objects with data on 18 billion+
relationships
Clicks and CTR history on SERPs
Co-occurrences of words within
queries and query sessions
Queries and query refinements with a
query session
Location and device cues
Was the search via Voice or typed in?
35. @KunleTCampbell
Here’s How Google Attempts to Understand the‘Layers of
Context’in a Query
“A search query for a search engine may be
improved by incorporating alternate terms into the
search query that are semantically similar to terms
of the search query, taking into account information
derived from the search query.”
- U.S. Patent 8,577,907 Abstract
Search queries improved based on query semantic information
http://1.usa.gov/1nosfFP
36. @KunleTCampbell
“the context for a particular query term included at the
beginning of the search query may be defined by a
query term located at the end of the search query”
Co-occurrences of words within queries and
query sessions
SYNONYM IDENTIFICATION BASED ON CO-OCCURRING TERMS
United States Patent: 8,538,984
September 17, 2013
Assignee: Google Inc. (Mountain View, CA)
http://1.usa.gov/1i900HL
where can I buy a playstation 4
37. @KunleTCampbell
Mobile OR Desktop?
Local business OR on an e-Tailer?
Context might be different…
Co-occurrences of words within queries and
query sessions
The defining query
where can I buy a playstation 4
49. @KunleTCampbell
Don’t Just List your Retail Business on
Wikipedia; ensure that it is in the right
category and that it has as many
schema attributes are completed
57. @KunleTCampbell
Avoid Keyword Cannibalisation
URL Singularity Is Key
Especially on Category and Product Pages
Rethink the excessive use of tag pages
What THING does your Category Page
Represent?
Mark-up Product & Category pages
with Schema.org, Microformats, Open
Graph
58. @KunleTCampbell
With Schema.org – Go Over and Beyond mark-up
Required by Google
http://schema.org/Product
Also consider the
data highlighter
tool to help
establish entities
66. @KunleTCampbell
If Google Changed its Engine in
preparation for voice and mobile
search, prepare for the storm
ahead by going
mobile
67. @KunleTCampbell
Check the growth and share of
mobile traffic and study your
Multi-Device Attribution
i.e. with Universal Analytics
+
+
68. @KunleTCampbell
Google is striving to become an
Answers’Engine – rather than a
Search Engine
with is gear to cover any and every
computing device
+
+
servers, mobile phones, tablet computers, notebook computers, music players, e-book readers, laptop or
desktop computers, PDAs, smart phones, or other stationary or portable devices
75. @KunleTCampbell
Align On-site Content Marketing with Content that Addresses
Pain Points at Each Stage of the Purchase Funnel
A
I
D
A
AWARENESS
INTEREST
DESIRE
ACTION
Brand Awareness Efforts: Viral
Video, Image, Advertising,
Sponsorship, Social
Create Interest: PR, Events,
Guides, Blog, YouTube Video
Series, eNewsletter, Q&As
Desire for Your Products: Brand
Name Search, Product Search,
Direct Traffic
Action: Buy Product, Voucher
Codes
THE AIDA MODEL