The battleground of entities & reviews

For anyone familiar with my articles, you’ll know I like to write a lot on a couple of specific topics:

1. Entities
2. The future of search

Today, we’re going to look at an area where both apply: reviews.

In this article, we’re not going to dive into specific strategies for acquiring reviews, as those change over time (though I will be linking below to a couple of fantastic pieces that cover well some current approaches). Instead, we’re going to look at why reviews are important and how Google looks at them — and likely will be looking at them in the months and years to come. We’re going to be looking at business reviews, obviously, but we’re further going to consider reviews of specific products and similar areas.

What are ‘entities?’

Before we get to any of the above, we need to cover what an entity is to really start to wrap our heads around how they play their role. If you’ve not yet heard of entities as they relate to search algorithms, they are defined by Google as follows:

[A]n entity is a thing or concept that is singular, unique, well-defined and distinguishable. For example, an entity may be a person, place, item, idea, abstract concept, concrete element, other suitable thing, or any combination thereof.

This seems like a fairly straightforward concept, and it is. Essentially, an entity is a thing. It may be a specific person, like “Danny Sullivan,” or it may be a singular and defined idea, like “evolution.”

While simple, the impact of entities on search is massive — and it’s sadly one of the most overlooked areas of discussion in SEO. So today, we’ll take steps to remedy that in at least one area.

Let’s talk about reviews

We’re going to begin our discussion in an area we all tend to think of when we think of reviews…

Business entity reviews

From a search standpoint, it can be useful to think of your business the way the law does (if you’re incorporated, at least): it is a thing that is unique and autonomous. It may be connected with other entities, but it is not the same as them, nor does an adjustment of those connections necessarily impact the business entity itself (a business may change its CEO while changing very little, for example).

Let’s get a feel for how this all works — and since an image is worth 1,000 words, let’s look at a graphical representation of our business in Google’s eyes:

OK, perhaps this picture isn’t worth 1,000 words, but let’s assume this is your business. Now let’s add in some connections that are natural. Entities connected with your business will appear in dashed red circles, and blue arrows will establish the relationships between these entities.

Now we’re getting started in illustrating how entities work. Your business entity is connected to other entities in ways that define many of its characteristics. If you want to simplify it, you can think of them like links to and from that entity. We’ll get a into that further below; for now, it’s enough to understand that a business entity is connected to other entities that define what that business is, where it’s located, who and what it’s connected to and so on.

Now, let’s add in some reviews in green dotted circles…

Now we can start to see how reviews fit into the picture. They’re not simply an unknowable ranking factor that’s “good because it’s good,” but rather a simple-to-understand addition to a business entity calculation. The more reviews you have, the more trusted the global review average will be — but further, the reviewers themselves are entities that factor in. In this area, we’re just starting to witness the first implementations of the entity status of the reviewer factoring in, but this will push forward dramatically in the coming months and years.

At this point, you may be asking what I’m referring to regarding the reviewer entity status. Great questions, hypothetical you! As was reported last week, Google has changed the way they display reviews for hotels on mobile to look like:

The key part here is the information related to the type of visitor (e.g., Families, Couples).  This requires taking in entity information related to the reviewer and adjusting specific review scores based on it. So let’s look at how that fits into our graph:

This is extremely limited in its scope to include only the number of reviews someone has done and their marital status — in reality, there would be dozens or hundreds of different connections.

With just this limited example, however, we can see that if the searcher is married, they are highly likely to enjoy their experience with Acme Business Entity, whereas a single person may not like it. These are the types of expressions of entity metrics we’re seeing presently in hotel reviews, but let’s flash forward a bit.

Dave and Bill have also done a lot of reviews compared with Jane’s 2, indicating they are less likely to be spammers and they understand how the review system functions. Inevitably, other areas of their own entity metrics will factor in, such as their other reviews and ratings, age, location and so on, and many of these will invisibly influence the rating system.

The idea that the algorithm will be adjusted to weight reviews from people with similar demographic or interest-based characteristics higher is not a big reach. In the example above, does it make more sense for me as a married guy reading reviews to see the total average of 3.6/5 or the adjusted average only considering people with characteristics similar to my own, which would yield a 4.5/5?

What we’re seeing with hotels is fine, but it isn’t broad enough in scope to hit the nail on the head across all sectors. It’s a proof of concept, and it’s interesting. But there is more to me than whether I’m solo or married, traveling for business or with my family — and to believe Google will not be taking this into account is short-sighted. And here’s why that’s great…

The vast majority of businesses could not (and should not) attain a 5/5 rating from every demographic. They cater to their audience, and that’s what they should do. A hipster restaurant with craft beer would suit me well now, but back when I was a starving student… not so much. Understanding who’s writing a review and what they expect and enjoy needs to factor in strongly.

This recent step with hotels makes sense, but it cannot possibly cover all the variables that would go into a review being fully applicable to me. Rather, Google can weight all the various entity information they have and come up with what they determine to be the most applicable reviews for me.

For example, let’s take a review for a Mexican restaurant and look at just a few characteristics Google might consider if I were personally searching. Some of my core characteristics include:

• Male
• 40s
• Has favorably reviewed Mexican restaurants
• Has written and rated many locations
• Lives in Victoria, Canada
• Has reviewed and rated various restaurants with mid-to-higher price points

Armed with this data, Google is going to know that when I’m looking up a Mexican restaurant in a new city, the rating given by a middle-aged person who tends to like good food and is willing to pay for it is going to be a lot more relevant than a review from a student who tends to hit up cheaper places to save money. Both may give a five-star review to different locations, but what they recommend is not equally applicable to me — and thus, their impact on reviews and the weight they pass to an entity needs to be adjusted.

Similarly, if both reviewed the same restaurant, and if that restaurant is known to have a higher price range, the review of the one known to visit and rate pricier locations should be weighted higher than the review of someone who may have their opinion skewed by feeling the pricing is too high (or they weight it more highly because they paid more for it, not because it’s actually good).

Flash forward in review evolution a bit, and these variables would appear in an equation that would look something like:

Rating Weight Adjustment = Gender * V + Age * W + Rated Mexican * X + Number Of Reviews * Y + Location * Z

In such a scenario, each factor is given a relevancy score (how relevant is gender to the enjoyment of Mexican food?) and then adjusted by machine learning over time to account for personal considerations and the wide array of other factors that would be taken into account on top of this very short list.

Let’s look at the following illustration (these weight numbers are examples and not indicative of what actually is in the algorithm):

We can get a feel for how much weight each of the factors has, with gender hardly impacting them at all and past ratings of Mexican restaurants factoring in heavily. Remember, we’re looking at a person here and the value of their reviews on my results. Rightfully, whether the reviewer is male or female would have very little impact on the weight of their review; however, their writing of past reviews of other Mexican restaurants, their age being close to mine and having written a large number of reviews would cause more emphasis to be placed on their review.

If I’m right, then in the near future we’ll see the review system change to place more weight on reviews where the reviewer is similar to the searcher, and where generic influencer scores will be placed on individuals (human entities). Furthermore, I would suggest it’s highly likely that not only will review weighting be adjusted as a result of personalization, but the actual search results themselves will be more personalized than they are today.

Thinking about products

I’m about to go out on a limb to discuss an area that I feel makes sense, but for which I’m just spit-balling. We’ve been talking a lot about the impact of reviewers on review weighting and relevancy of a site to a specific demographic. But I would suggest that the products a business carries — and how those products are reviewed — may well impact an entity’s overall prominence, too.

Let’s look at a simple example based on our second entity illustration above.

What I would predict we will see in the near future is that the reviews of a specific product, or “product entity,” will impact a business entity’s status if they sell that product (even if the review is from a different site). If a company were selling only products with low reviews across different sites, I would put forth that that business entity’s overall score would be diminished (certainly for queries related to those products or that product category).

One can think of this as tall-tale breadcrumbs. All of these products are understood to be under a specific hierarchy/category, and that category is understood to contain low-quality items (though, again, this could be adjusted based on reviewer demographics). And thus, the Acme Business Entity would be reduced in the value assigned to it for that category of products.

I need to stress, again, that at this time I have not seen any evidence of this. As I said above, I’m just spit-balling here. But if one simply thinks about an environment where Google wants to provide its searchers with results that will meet their needs — and assuming they have the information to connect the reviews of one product with another on a different site — it is a logical and beneficial angle to pursue.

So, what do you do?

We’ve covered a lot here about how entities and reviews can and likely will impact rankings and how review scores will likely be augmented further in the very near future to place more weight on those reviews that more closely match the searcher’s intent and interests. So let’s review what you need to pay attention to…

• Who is reviewing you, and what their reviews are. You can’t please all of the people all of the time, but are you pleasing your target demographic? Be clear on your site or in your business who you are catering to and what they can expect.
• Which of the products and services you offer are being reviewed favorably and poorly across the web. This is simply a good business move (clearing out bad products and focusing on the good); however, if I’m right, and this will start to impact your own rankings and review scores, it will be more important than ever.
• How your site connects with other entities (e.g., authors in your blog, companies you’re affiliated with) and how they are rated. If you’re associated with poorly reviewed and rated entities, this flow of influence (or rather, lack thereof) will impact you.

In the end, the point is that we can no longer focus on simply how our business entity is reviewed but must look at how the entities it’s connected to are reviewed and who is doing that reviewing. We’re being forced into an environment where we need to look at our business as a whole, what we offer, who we partner with and who we cater to. While we need to respond to negative reviews as always, we need to be more conscious of who is doing the reviewing and whether they are part of our target demographic.

Resources

I promised above to link to some resources on how to get reviews and the risks involved, since we didn’t talk much about those specific strategies here.  Here are some of my favorite pieces on the subject:

• Thomas Ballantyne Speaking On Reviews At SMX West — Search Engine Land
• Info On The Consumer Review — Search Engine Land
• How To Get Local Reviews & Rating — Yoast
• Create A Link For Customers To Write Reviews — Google

Conclusion

I hope that if nothing else, this article has given you food for thought. While a lot of this article is based on ideas not yet implemented, most are logical, and we’re starting to see some of the early signs that this is the direction things are about to take. Our job (yours and mine) is to be ready for these things when they come, and being ahead of the curve in understanding what’s happening will help us make business decisions that lead naturally to a better entity status for our companies. Fortunately, there is no downside to following the ideas listed above; it’s simply forcing us to understand the complexity (and simplicity) of the way Google approaches entities as outlined in their many patents on the subject and changes we’re seeing them make every day.

The post The battleground of entities & reviews appeared first on Search Engine Land.