Keywords Are Dead – Long Live Concepts, Entities & Audiences! – #SMX London Day 1

Keywords Are Dead – Long Live Concepts, Entities & Audiences! – #SMX London Day 1

21st May 2015

With the introduction of Hummingbird and Google’s extensive ongoing research (Research Areas & Publications) on natural language processing, AI and machine learning we should start looking further than just individual keywords. This session is presented by my SEO friends Dixon Jones, Mike King and Kirsty Hulse!

Kirsty is kicking off the session with an introduction to voice search. It currently a hot topic but not completely new: Google introduced it in 2007. In 2014 voice search in doubled, according to Google but exact numbers are not available unfortunately. 57% of the users use it when with friends, 59% during watching TV and 39% use it during cooking.

The interesting development is the evolution of the queries used during voice search. People are actually having conversations with their devices instead of entering a single, individual keyword. Search is fundamentally changing, talking to a device will be intuitive instead of embarrassing in a few year.

Each mobile search triggers two follow up actions. Are websites still useful? Google is using their Knowledge Graph to directly serve users with answers. But the Knowledge graph and structured data are difficult and sometimes costly to implement. Luckily for us, JSON-LD is making it a bit more easy and in combination with Google Tag Manager, you can add structured data without needing access to the source code. Simo Ahava wrote a clear guide about implementing: Enrich SERP results using GTM.

Make sure you start optimizing your local stores, business entities and websites or apps for the changing search behaviour through voice search because it is growing fast. Make sure Google understands your content, use structured data and stop looking at individual keywords since users aren’t doing that either.

How Search Engines analyze and define concepts, entities and audiences

Informational retrieval by search engines is not only about crawling and indexing content, but grouping that information is key. If you want to find a fact, you first search for a group of information and go deeper down the rabbit hole every step. Grouping and categorizing the web into topics prevent spam: you can categorize for country IP / TLD, images, videos, local listings, maps, travel or content type like a personal blog, news, company website etcetera.

Humans and Websites also do that, think about oldskool directories like DMOZ or directories 2.0 like Yelp, Tripadvisor. Approaching the web by defining topics is a real challenge if not done manually and it needs to be scalable. The way vocabulary and links connect content is a way of categorizing. The original concept of PageRank connected pages via links and determined specific value of specific nodes. The more links a document has, the higher the probability of that document being important or relevant. This system can also be used to map topical relevance.


Young-Ho Eom did research on determining the most influential person on Wikipedia. By using PageRank, he came to an obvious wrong conclusion: PageRank, TrustFlow and the Search Universe. The reason for this is the fact that every small signal is attributed. Instead of using plain relationships, like links are, topical relevance is more relevant and gives better results. On of the practical uses is Topical TrustFlow by Majestic: making it possible to group influential people according to topic.

Persona Driven keyword research

Bad keyword research is just a list of search volumes. Better keyword research accounts for possible outcome like revenue. Good keyword research includes intent. But that is not enough nowadays.

Start with mapping out the complete customer life cycle and develop personas. Identify the needs for every individual stage and persona and build the right content for it.

Some tools Mike mentioned to help you get the right data:
Adwords Keyword Planner gives an idea how Google sees entities
SEMRush for rankings and market data
Bottlenose Sonar – Discover current trends on Twitter
Topsy – Twitter search & monitoring
Keyword Studio
Alchemy API – create smart apps, classify entities
nTopic – Content Relevancy scores
Searchmetrics Content optimization Tool
PostMan – Supercharge your API workflow

After collecting a lot of data you can built a data driven user journey. Mike wrote a complete guide on how to work with this data to develop a strategy: The Persona-Driven Keyword Research Process


Written By
Jan-Willem Bobbink got addicted with online marketing in 2004, since he build his first international webshop when he was 16. He is currently working as Freelance SEO for global clients and is ambassador for Majestic. His blog can be found at and he shares his cycling adventures at CATW!
Recap Friends of Search 2016  #FOS16
Latest Post from Jan-Willem
Business Recap Friends of Search 2016 #FOS16
16th February 2016
  • This field is for validation purposes and should be left unchanged.