Stop using Keyword volumes to inform your content planning!
This presentation looks at a new way to understand User Intent and a new way to plan your content for SEO
2. Purpose
Understand exactly how Google
see your content as a set of
related Entities
Understand how Entity SEO
trumps Keyword SEO
Learn How to optimize your
content using Knowledge
Graphs instead of Keywords.
8. How Search Works in 2022
• tour Eiffel
• Эйфелийн цамхаг
• Tùr Eiffel
• twr Eiffel
• 埃菲爾鐵塔
• අයිෆල් කුළුණ
• Eiffel טורעם
• A Big Metal Tower in Paris
25. Three steps to a
Knowledge Graph
Find the most authoritative content
around a subject (10 pages will do)
Break the content into entities
Tabulate the results & make them
Presentable and usable
42. Wrapping it all
up
• SEO is about matching topics and entities to real
user Intent
• You can create an ad hoc knowledge graph without
tools
• Creating at Scale needs (at least) an NLP algorithm
With a Knowledge Graphs, NLP and Google suggest
you can:
1: Create better content around topic clusters
2: Create great Schema fast
3: Internally Link entities
4: Plan Content using A Gap Analyses of Entities
recommended by Google Suggest, based on your
EXISTING content
45. Does it matter that
My Knowledge
Graph <> Googles?
Yes and no. What matters is that you stay true to your
own map.
46. You Need a Topic
Map
• You do not need the same map /
Knowledge Graph as Google
• But it does have to represent
reality.
• Otherwise, your content can
walk of a cliff
Editor's Notes
Hello Dallas! Thank you for letting me escape the chaos that is British politics!
The benefit of acting on this talk. Here are 3 separate case studies, not all using our tools, but all using this approach for SEO.
Everyone OK with this as a talk?
I am very proud of my new car. Has anyone gone all electric yet? This is the new Mustang and I really love it.
But here’s the issue with the word “Mustang”… This is also a beautiful beast. But when it comes to the keyword “Mustang” it’s an entirely different animal.
I am also the proud owner of this car. Anyone want to guess what it is called?
Bulldog… Pilgrim Bulldog.
The point is that words are just that… words. They LABEL ideas and concepts. As SEOs we call these concepts TOPICS and as scientists, The word is “Entities”.
But sometimes, disambiguation is not always clear!
What are these?
Which ones are right, and which ones are wrong?
They are all right. All wrong.
Do they show relationships between things in Manhatten? Yep.
Do the show the distances between place in Manhatten? Sure.
Are they useful for travel? Sure.
Are they Manhatten? No. in the same way, you do not need Google’s knowledge graph or processing power to optimise content using knowledge graphs. You just need a reasonably accurate view of the world which does not overly contradict Google’s understanding of the world.
So what does a knowledge graph look like?
Well every website’s knowledge graph tends to look very different. It is really a database of the frequency of ideas expressed on a website.
We’ve been talking about Horseshoes in relation to Pubs,
So I do love this poster that I saw in a local pub, which is pretty much a knowledge graph of Beer.
I like to bring complex concept back to real world examples that we can see and digest as humans.
The Entity “German Lager” is part of the lager family.
- It is connected to Munich Lager and Vienna Lager
- Of which Samuel Asams Boston Lager is a type
Google grabs ideas and augments its records by reading YOUR content, whether it is in pictures, text, increasingly spoken word and video, News or countless other formats as Google collates the information.
Google reads content, page by page, noticing entity by entity, then uses that to augment its entity within it Knowledge Graph.
Google Call this their “Natural Language AI” (previously NLP API) but it compresses content into Machine readable nuggets and the also have another one for Pictures and another for spoken word.
Schema helps Google to associate the correct concept to the correct record in ots knowledge graph.
But you HAVE to reflect reality in your schema…
So now it becomes obvious what we mean when SEOs say “context”.
I imagine a web page or web site now as a list of entity ID numbers!But be VERY careful describing your new MachE as being “wild”!Or about talking about its “untamed” setting…
*Talking about keywords JUST BECAUSE they are in a list, is a HUGE mistake.
Keyword tools have done the grouping.
Here is how Ahrefs groups keywords
SemRush groups ideas into separate worksheets.
But where they fall down is understanding context and intent
I’m going to throw in some controversy.
SEOs describe Intent as “Informational Investigational Transactional or Navigational”. I checked the top 10 results for Search Intent and these were mentioned heavily by Moz, Yoast, SEMRush and Ahrefs.
Even Wikipedia highlights these four classification of Intent… EVEN THOUGH they start with an entirely different definition.
They call it the “identification and categorization of what a user inline intended to find”
If you start generic, you get search suggestions… look here – one entity (the picture), then lots of search terms…BUT NO intent…
I think there is an awful lot more to understanding intent than the SEO mantra of these four or five classifications.
Look at what is going on here… Google is
1: Blending entities with search terms.
2: Nothing here suggests Google is yet ready to consider Intent based on purchase vs navigational. It is considering context!
3: Imagine a hotel near Horseshoe falls talking about The Horseshoe in Warlingham! – There is no meaningful overlap between these ideas.
So I put it to you, that when it comes to your website and your keyword research, search Intent has nothing to do with what SEOs have all been saying.
(That includes our own tools, by the way. We have been led on this as well… but it’s time to look at the serps and forge our own view here!)
Without context, Google’s core index on the right goes straight to the heart of the matter. Not to Horseshoe pass or Horseshoe falls.
With Context – namely my location, Google can find local pubs and places, relative to me.
If I am a farrier, I do not want to look like a pub and if I am a pub, I do not want to look like a Farrier.
THAT is how I look at Intent. Did the searcher want food or a farrier? Google doesn’t know, right now.
So doing keyword research by words makes no sense. You need to plan your content by Entity, and you need to connect your entities in a well organized knowledge graph.
So how do you go about building a knowledge Graph?
Building a Knowledge Graph in THEORY is not difficult.
Building it quickly and reliably is difficult.
Choose a keyword.
Take the top X results.
Build a database of all the topics and frequency
Use this to inform your page content…
Then you need to tabulate the results. You could do this in Excel… but essentially your are recording the relative importance of each topic by counting its frequency of use.
If you are feeling adventurous, you could weight use in Page Titles or H tags, but once you start, stick to a framework. Don’t let your human self deviate based on what you are reading, because you will be improperly blending human understanding with machine understanding.
Prior to entities, I used to use Wordclouds to make this easier, I could just cut and paste pages of text into a single Word Cloud generating tool, but the issue with a word cloud is that it loses context very quickly.
You have to stitch back the relationships between words manually.
You are unlikely to see the “alcohol content” as a concept in a WordCloud.
The result should be something like this.
It also helps to ctaegorize the topics… because…
But THEN you can throw this research back into Google Suggest and expand it.PLUS, by seeing the relevance as WELL as KW Vol and CPC, we can assign a priority, making the information much actionable.
ALSO Google suggest throws out some interesting questions.
What is really exciting is that User Intent is now defined by verbs, not by arbitrary concepts.
This means that we can see genuine questions that people might be asking, by topic cluster.
Now THAT is useful keyword research
Having an Knowledge Graph of topics ob the site and page lets you easily write “about” and “mentions” schema like this.
Turning an implicit understanding of the main content into and explicit epression of the main content.
We can then present the data in a way we can all understand.
This turns my Knowledge Graph or Topic Map into a Topic Wheel.
Breaking the topics into categories is also an extra layer of icing on the cake
So now we know how to write a page of content around any key word.
But as soon as we got this live, people asked about when they should create new pages and when they should add content on an existing page.
How can Content Planning be done at the SITE level.
We cracked this in August.
The truth is, as SEOs, it is very rare that we start with a clean slate.
This is where Keyword Research tools completely miss the boat.
Let’s take an example. We have already done a keyword research around the phrase “The Azores”
But how applicable would it be to a site like this?
Keyword Research at the site level sucks, because:
The research is done out of context!
You need to analyze the WEBSITE first!
You need to build a Knowledge graph of EXISTING Content First!
When Fred explained this to me I just didn’t get it, until he built it.
Now I do – and I can show you in my first language, not a French translation.
It’s BRILLIANT… so how does it work?
Instead of analyzing the competitor pages for a given keyword,
You analyze all the important content on the website itself.
-And build a knowledge graph out of THAT.
Then you start putting THESE topics into Google Suggest and start creating a Knowledge Graph of Google Suggestions.
Then you break the SUGGESTION into entities and build a knowledge graph of the suggestions
Then you compare the perfect knowledge Graph to the site’s Knowledge Graph.
Clustered concepts
Last year I was climbing this mountain in Wales. Ahead of me was an ex-marine. Behind me was an experienced navigator with satellite driven GPS mappig.
The blue lines intersect at a summit and the red line is only 100 yards away, leading to the next summit. Because we could SEE the summit, we walked to it assuming we were still on the path. We then just carried on and ended up climbing down almost TOTALLY the wrong way. We didn’t check the GPS OR the compass.
- It did not matter WHICH map we had. What mattered was the we ignored it.