The Knowledge Panel Course: Identifying the Relevant Corroborative Sources
Script from the lesson The Knowledge Panel Course
Jason Barnard speaking: Hi and welcome. You need corroboration from relevant first, second, and third party authoritative sites. There is a simple but important distinction between first, second, and third party corroboration.
Jason Barnard speaking: First party is any web page you control. Corroboration on these has a very fast rate of diminishing returns. By that, I mean each additional piece of corroboration on a first party site will bring significantly less value than the previous one.
Jason Barnard speaking: Second party is web pages that you partially control. Social profiles, review platforms, for example, where you control parts of the content such as the title and the description, but the content can be added by others and the actual page is owned and controlled by another party. Corroboration on these has a slower rate of diminishing returns than the first party, with each additional corroboration having slightly less value than the last.
Jason Barnard speaking: Third party is web pages where you have no control. Articles about the entity on independent media sites, interviews, appearances on webinars, and any other content that is controlled by an entity over which you have no control. Corroboration on these has no rate of diminishing returns. Each additional corroboration will bring the amount of additional confidence the third party inspires in Google’s mind.
Jason Barnard speaking: Human curated collaborative sites such as Wikipedia, Wikidata, Crunchbase, IMDb are technically second party, but can be considered to be third party since they are collaborative and ultimate control is with the human editors and admins.
Jason Barnard speaking: What affects the additional confidence in the understanding for Google? The main factors here are the consistency of the information, the relevancy, and the authoritativeness of the source. The number of corroborations, the frequency, sustained corroborative activity around the entity being incredibly helpful.
Jason Barnard speaking: There is absolutely not a one-rule-fits-all here. Each entity will need different doses of each factor, and each factor will play off each other, as well as Google’s initial perception of the Entity Identity and the related named entities. Generally speaking, 30 is a reasonable number to aim for. In terms of proportions, the second party corroboration will generally dominate quite significantly.
Jason Barnard speaking: There are simple and general rules that you can follow to ensure that you are choosing the right corroboration. To make it easier, here is a table that you can use as a guide when deciding what you should focus on and what should be added and what should not be added to your Schema Markup.
Jason Barnard speaking: What you shouldn’t add is outdated profiles that the entity has no control of, articles written by the entity that don’t contain relevant information about the entity, content from dubious and spam sites, profiles of other entities, and social media posts and replies. What you should add is up-to-date profiles controlled by the entity, interviews about the entity, articles about the history or life of the entity, and videos exclusively about the entity.
Jason Barnard speaking: Bear in mind that Google is really comfortable with pages that talk 80% or more about that specific entity, so focus on those only. Anything where the entity shares with other entities isn’t going to be incredibly powerful for Google’s understanding at this stage. And also bear in mind that the most important thing is to always remember that you are educating the child that is Google, and any corroboration must contain truthful facts about the entity. An article written by the entity can be interesting and impress your audience with a great sales pitch. But if it does not contain relevant factual information about the entity, it isn’t very helpful to Google the child as it tries to understand the facts about you.
Jason Barnard speaking: That said, we made a series of videos about Kalicube’s offers. And although first party, they have proven to be a big help. We provide a factual presentation of who we are, what we offer, and who our audience is, and presented it in a manner that is very attractive to our audience and also helpful and factual for Google. You can watch them on the homepage of kalicube.com to get an idea of what we did.
Jason Barnard speaking: The next thing you should do is make sure that the different sources of corroboration that you have don’t contradict each other. Contradictions lead to confusion, and confusion is very troublesome for Google.
Jason Barnard speaking: So, now that we have established the initial guidelines, you need to start working. The initial work in making the easiest plate puzzle for Google is to do a spring clean. Correct all of the information on all the first and second party sites. This is usually a massive job that will take weeks or perhaps months. If you are using Kalicube Pro, it will be easier and quicker since the platform automatically finds all the corroborative sources and lists them by order of importance to Google and therefore, priority for you.
Jason Barnard speaking: As for how you will correct the information, many SEOs will worry about duplicate content. They might talk about Google’s Panda updates, whereby Google penalised duplicate content when it found multiple versions of the same text around the web.
Jason Barnard speaking: However, in this specific case, duplicate content is actually an advantage. That repetition is what’s going to build the child’s confidence that the contents of the description on your Entity Home are in fact true. There is no need to worry that there will be repetition on your Brand SERP. Google automatically takes care of that. It will pick and choose and mix and match the parts that it wants to show, so rarely do you get a case where a Brand SERP is overly repetitive.
Jason Barnard speaking: And this is also the advantage of the modular system we use at Kalicube. On different platforms around the web, you can use different parts of that modular entity description according to the audience that will be reading it. So on Crunchbase, you could potentially use the module about the history of the company and the module about the key people, but not the one about the company’s podcast. On your Facebook profile, you might want to use the module about the mission statement and the module about the customer service.
Jason Barnard speaking: The aim isn’t to get exactly the same description on every single platform. You will naturally end up with parts of that description on multiple platforms. Once again, the fragmented plate analogy, but here all the pieces fit neatly together. If the descriptions and facts differ from platform to platform, then Google not only has a broken plate problem, it has a problem with a broken plate where the pieces don’t fit together. Ensuring that the corroboration around the web is consistent ensures the pieces all fit together correctly, and Google can easily and reliably build the complete puzzle that will look exactly like the one you present to it on your Entity Home.
Jason Barnard speaking: Once you have completed the spring clean, you can expand, obviously prioritising third party content about the entity since corroboration from a relevant, authoritative third party site is by far the most powerful. When working with third party sites, it is harder to get them to include exactly the description that you propose. If you can, it is great since it is a confirmation that an independent source agrees that what you say is true. The important thing is to get at least some of the facts mentioned, so be flexible.
Jason Barnard speaking: But whatever the type of site, first, second, or third party, you want to aim to keep the texts clear, use semantic triples, and mention closely related entities where reasonable and possible. As a quick reminder about writing, a semantic triple is simply subject-verb-object. And a sentence will be clearer for a machine like Google when the sentence is relatively short, well focused, and the subject-verb-object are close together. Kalicube provides digital marketing services. Jason Barnard works at Kalicube. Those are both semantic triples. The sentences are very short, too short, of course.
Jason Barnard speaking: So if a third party such as entrepreneur.com were to write this: “Based in France, Kalicube provides digital marketing services to businesses who want to control their brand message on Google,” I would be very happy. We have the semantic triple that describes what we do, an explanation of who we serve and what we offer to them, and the location, which helps with disambiguation, and also, rather cheekily, a mention of a related entity, Google itself.
Jason Barnard speaking: Repetition of what you are saying on the Entity Home on first, second, and third party sources will build Google’s confidence in its understanding. And as always, remember first party sites will have a very fast rate of diminishing returns in terms of building confidence, second party websites will have a slower rate of diminishing returns, and third parties will tend not to have a rate of diminishing returns at all.
Jason Barnard speaking: And that means, after you’ve done the spring cleaning and ensured a consistency on first and second party sites, building Google’s confidence will be part of an ongoing digital PR strategy aimed at positive, accurate corroboration on relevant third party sites.
Jason Barnard speaking: On Kalicube Pro, we offer a list of third party sites that Google trusts. You can filter by country and by industry. And it is absolutely free, but it is not a definitive list and it is not complete. It will give you some idea of some initial targets and a good idea
of the type of site Google is looking at in your industry, though. You’ll notice that Kalicube’s sites and my own personal site appears in this list regularly. This is because over the years, we have built Google’s trust in us for our industry and therefore, Google sees us as an authority and trusts what we say about the entities in our industry.
Jason Barnard speaking: And finally, a word of warning about spam. There are services that will create literally hundreds of second party profiles for you. That is not a good idea. Because after the first half a dozen, a fast rate of diminishing returns will kick in and those profiles are going to be very weak. They’re going to bring very, very little additional value, probably no value at all. Even if this strategy might sometimes work short term, they will be very difficult for you to maintain over time. And as your entity evolves, the facts and therefore your description changes, and you will forget to update many of these profiles. That will mean inconsistency and contradiction for Google, and Google the child hates that.
Jason Barnard speaking: Plus, since over time, there will be very little or no activity and no audience engagement, Google will see this, identify them as spam, and that will damage your credibility. And Google hates spam.
Jason Barnard speaking: Another warning is to be aware of websites that try to trick the machine. Some of them do it. Wikitia has done a good job of tricking the machine, but Google has a massive team dedicated to cleaning up Knowledge Panel spam, through manual intervention on a case by case basis, plus a huge investment in machine learning for the Knowledge Graph and for the SERPs.
Jason Barnard speaking: You will be coming at this from the perspective of the Entity Home and your initial description about who you are, what you do, and which audience you serve. Google the child now has that information from the parent and is trying to figure out which other sources it can trust. Brothers, sisters, grandmother would represent second party sites in this analogy. Teachers, community leaders, and so on would represent Wikipedia, IMDb, and other sources the child has been told it can trust.
Jason Barnard speaking: For the rest of the world wide web, the scary wider world for our child analogy, Google is training their algorithm to judge for itself. We are in the early days, and sometimes the child can be tricked. Google cannot allow this child to be tricked too often. If it gets information wrong now or it trusts the wrong sources, then it will get progressively worse as time goes by.
Jason Barnard speaking: Google’s entire business model depends on building a Knowledge Graph that it can trust, the machine-readable encyclopedia I mentioned in the first lesson. It needs to understand the world in order to be able to match their users’ needs to the most effective solutions provided by the different entities, your company, or yourself.
Jason Barnard speaking: Knowledge spam management is already a huge priority for Google. And it is definitely going to become an even bigger focus for them over the coming years. Knowledge spam reminds me of link building, creating massive numbers of phony, spammy links on second rate websites used to trick Google. The engineers then implemented the Penguin updates and now have this under control in the core algorithms. Something very similar is going to happen with Knowledge Panels.
Jason Barnard speaking: There will always be short term cheats and tricks you can play. But long term, I don’t see why you wouldn’t do this properly.
Jason Barnard speaking: Thank you, and I’ll see you soon.
