The Knowledge Panel Course: Managing People Also Search For and Related Searches
Script from the lesson The Knowledge Panel Course
Managing People Also Search For and Related Searches
Jason Barnard speaking: Hi and welcome! People Also Search For and Related Searches. Unfortunately, these are almost impossible to control or even significantly influence. This lesson contains mostly an explanation of how they work and the tips and techniques we currently know. We are working hard analysing the data in Kalicube Pro and doing experiments to see how we can better exert influence over these two Knowledge Elements on the SERP.
Jason Barnard speaking: The results for People Also Search For and Related Searches are often different. Google recently changed the dropdown result from the search box from Related Searches to People Also Search For. But for the moment, in order to distinguish them, I will use Related Searches.
Jason Barnard speaking: Now, that change in name may be a signal that they are going to merge the results, or perhaps use the same algorithm, or perhaps even change the layout of the SERP. I’ll need to re-record this lesson in a year or so when we see what Google does. And at Kalicube, we have figured out exactly how this works. That’s pretty exciting. I’ll also talk about what we call Entity Boxes at the bottom of some Brand SERPs. Google includes these entity based Related Searches, like these three for the Beatles.
Jason Barnard speaking: Right now, they are obviously driven by different algorithms. Related Searches is more Web Index, cross-search volume, and name similarity based, and the People Also Search For and Entity Boxes are more Knowledge Graph and category ontology based. But it appears all those factors and more are used in the different algorithms. They are just weighted differently.
Jason Barnard speaking: Firstly, Related Searches in the dropdown menu. When looking at the results for myself and multiple experiments we have done, not a large data set, but enough to make a good educated guess. In Related Searches, Google seems to be looking more at the closeness, strength, and duration of the relationship. In this case, my sister and my mother are close, strong, and long.
Jason Barnard speaking: Another thing it looks at is the possibility the user got it wrong. Jason Bernard is included because of the similarity in the names, since there is no other meaningful relationship between us. Kalicube is included principally due to the cross-search volume. A large number of people who search Jason Barnard will also search Kalicube at some point.
Jason Barnard speaking: Samantha Weinberg is there because she has a parallel career as a fiction writer using the pseudonym Kate Westbrook. Rand Fishkin is included due to cross-searches, but also because I focused on communicating my admittedly distant relationship with Rand when I started rebuilding my Knowledge Panel from scratch after my Wikipedia page was deleted.
Jason Barnard speaking: That’s a long story and was a huge learning experience, but that story is for another day.
Jason Barnard speaking: The point here really is that Related Searches are all about offering both disambiguation options and research opportunities. You cannot influence the disambiguation entities, in this case Jason Bernard, except by potentially convincing Google that there is no ambiguity. Increasing the search volume on the correct name here would be a solution. The influence you have here is always indirect, and you can’t aim for anything very specific. Educating Google about strong, close, and long lasting relationships with known entities and demonstrating their value to your audience is the technique.
Jason Barnard speaking: For example, I could work to improve Google’s understanding of my relationship to my other sister, who is a writer. She could potentially be a candidate to replace Samantha Weinberg. From a professional perspective, I could work on educating more about a long term business partner or companies I have founded.
Jason Barnard speaking: With Rand Fishkin, I made more online noise about the Kalicube Tuesdays with him than the other guests, mentioning him in my entity description, and generally ensuring Google clearly sees the relationship repeatedly. Whether there is truly cause and effect here is debatable. We are still figuring this out at Kalicube, but this seems like a good start.
Jason Barnard speaking: Ironically, People Also Search For is more heavily weighted towards similar entities using ontologies. So in fact, it seems to me that Google is using a misnomer here. These are less People Also Search For and more Related Searches.
Jason Barnard speaking: In my case, as of late 2022, the People Also Search For are other digital marketers, which makes sense from an Entity Equivalent standpoint. In this specific case, geo doesn’t apply since digital marketing is an international occupation by definition. However, for most companies and people, these will vary from country to country, and you’ll see the Entity Equivalent aspect more clearly, same entity type, same geo region, and same industry.
Jason Barnard speaking: There probably is some crossover of searches. People who search my name probably also search Rand Fiskin, Joost de Valk, Cindy Krum, or Lily Ray, but it is Entity Equivalents that is the main driving factor here.
Jason Barnard speaking: A good illustration of this is that when my subtitle in the Knowledge Panel was musician, the People Also Search For were Boowa and Kwala as a music group and three other musical artists. I didn’t keep a screenshot of the Google SERP from that time. Luckily, Kalicube Pro tracks my Brand SERP and Knowledge Panel, so I have the data going back several years. The other musical artists have no connection to me or even my previous musical career, which was music for children and French alternative punk folk.
Jason Barnard speaking: So, People Also Search For in the Knowledge Panel is very much ontological and would be more appropriately referred to as ontologically Related Searches or in Kalicube speak, close Entity Equivalents.
Jason Barnard speaking: Obviously, this is just one example, but it is very illustrative and helpful. We have experiments that show the same general pattern. And I look at Brand SERPs all day, every day, and I see these same distinctions between Related Searches and People Also Search For all the time.
Jason Barnard speaking: In this example with Google’s own Brand SERP, we can see the geo differences in both Related Searches and in PASF. The PASF shows the geo aspect of Entity Equivalents. Google’s perception of the equivalent for a searcher in the US is different to that for a searcher in Australia.
Jason Barnard speaking: One interesting quirk in People Also Search For is trending. Google will sometimes temporarily promote an equivalent entity when search volume for that equivalent entity spikes. Here is an example. There is nothing you can do to influence trending entities appearing in your PASF. However, to appear as a trending in other people’s PASF, you simply need to create a spike in searches on the name.
Jason Barnard speaking: To influence what appears in People Also Search For in a Knowledge Panel is, once again, using your Entity Home, entity description, and corroborative sources to insist on relationships with specific Entity Equivalents. Getting multiple relevant, authoritative third parties to mention your entity in a semantic triple with a specific Entity Equivalent is certainly going to help.
Jason Barnard speaking: Multiple repetitions of the semantic triple “Jason Barnard works with Andrea Volpini” on first, second, and third party sites would increase the probability that Andrea would appear in my People Also Search For, since he is also in the digital marketing industry. In fact, two years ago, he was in my PASF. Once again, cause and effect is difficult to establish, but that was during a period when his company, WordLift, started sponsoring my podcast, and we both appeared on several popular webinars and online conferences together.
Jason Barnard speaking: The last item here is Entity Boxes, which are part of the Related Searches at the bottom of the Brand SERP. This example for the Beatles shows an obvious ontological bias. It is aimed at taking the searcher on a research journey, a little like Knowledge Panel hopping but with Entity Boxes.
Jason Barnard speaking: But games aside, Google adds these Entity Boxes whenever it has a sufficiently good understanding of the entity and its context. These Entity Boxes are built in a similar manner to People Also Search For. The algorithm is very similar, perhaps even the same, except that it finds a related topic and shows Entity Equivalents from that topic.
Jason Barnard speaking: To trigger and manage these, you’ll need to take educating Google to the next level beyond Knowledge Panels. In fact, for all three of these SERP Elements, for any level of control, you’ll need to establish a very solid understanding of your entity and its relationships to other entities, including topic entities.
Jason Barnard speaking: Having said I’d have to re-record this lesson in the future, I now see I will probably have to research fully and create a whole course about building your mini-Knowledge Graph as a sub-section that is included in Google’s own Knowledge Vault. At Kalicube, we will soon build a system to analyse these patterns and figure out how to do exactly that and gain more control over related entities in Brand SERPs.
Jason Barnard speaking: Thank you so much, and I’ll see you in the next lesson.
