Did Google’s Site Diversity Update Live Up to its Promise?


How could average diversity be improved?

With the 10,000 keyword MozCast Set of 10,000 keywords to study the diversity of the page-one SERPs. In simple terms, we determined the number of unique sub-domains present on each result page. Because page one on Google may contain less than ten organic results, the number was measured in a percentage, specifically the proportion of sub-domains with unique domains to the all organic results on the page. Here’s the graph for 30 days (May 19 to June 17):

A diversity of 90 percent in a 10 results SERP means that 9/10 sub-domains are unique, with one repetition. It’s hard to discern however, from June 6th to July 7th the average diversity increased only marginally, going increasing from 90.23 per cent to 90.72 percentage (a 0.49 percent increase). If we look in quite a little (10x) in the Y-axis, it is possible to observe the pattern across the course of time:

Zooming in to a 10 percent interval (85-95 percent of diversity) You can see that the majority of the changes was made in a one day. The improvement remains in place throughout the entire week following the update. When zoomed in however it’s not remarkable.

Is the change more limited?

As honest with Google as we can is a must, so we should take a look at one of their subsequent assertions:

What would happen if Google enhanced the worst-case scenarios, however it wasn’t clear at the time we compared the SERPs in all categories? We can identify situations that contain greater than two pages on the same website by focusing only on SERPs that have a a website diversity score of at least 80 percent or more (eight in ten subdomains are distinct). Here’s the 30-day graph of the specific cases:

On June 6, 84.58 percent of sites in our database were found to have a range of at least 80 percent. The following day, the percentage went up from 84.58 percent to 86.68 percent -that’s a 2.1 percent increase. Let’s dive even deeper to determine what’s happening with individual counts.

What caused the impact to disintegrate?

One data point won’t provide much insight into the events happening in each of the buckets. In this study I’m going to utilize the exact number of duplicates, because percentages can become a bit confusing when we put them in buckets. Another problem is that sometimes, two websites possess more than one organic search resultwhich reduces the overall variety of the SERP. However, it doesn’t mean that only one site is dominant.

So, what happens if we just look at the totals of the top site (the one with the most duplications) in the SERPs with 10,000 results? Let’s compare the June 6th (blue) to the 7th of June (purple):

In the case of just over half of the SERPs in our database there were no duplicates (every site had a listing) This number didn’t significantly change following the update. The number of websites with more than two listing (i.e. two listings) was significantly increased following updating (up to 346 SERPs). This was balanced by a reduction in SERPs, which ranged from the three to five results (down in 345 SERPs in the 3 bins).

The numbers become too small for 5K scale after four-count SERPs, therefore I’ll limit the Y-axis to:

SERPs for dominant websites with up to 10 organic listings, accounted for just 117 of the 10,000 SERPs (just more than 1.1%) in June 6. Following the update it actually increased a little, to 119 SERPs.

We continue to see SERPs where one site is dominant and that narrative did not change much following the update. However, these SERPs with a count of six to ten are not common. When we look at the keywords, we see that a majority of them have a brand or navigational intention. Here are a few of the keywords that we can still find a 10-count:

  • “kohl’s hours”
  • “macy’s hotel collection”
  • “lowes outlet”
  • “dillard’s sales”
  • “edd unemployment”

Site-links are often seen in the top spot (which permit the creation of up to 6 additional hyperlinks to a website). It’s difficult to determine the reason Google does not use links from sites in these cases of extreme. It could be that the purpose isn’t so apparent however we can guess from looking at a few examples. Keep in mind, as well that Google determines intent through algorithms which means that it could change with the course of.

It’s not an easy issue. It’s not a tool you can pull in isolation, particularly when left to the algorithms. The reduction of repetition could cause harm to quality, in some cases (especially SERPs that are branded). In the same way, numerous algorithm updates that are not related to diversity appear to result in unintended effects on the diversity of sites.

What’s the final decision?

When looking at the site’s diversity it is important to be wary of relying too heavily on anecdotes. In my experience, there are certain SERPs in which a single domain is believed to have too much influence. For instance Here’s the primary results page on a Google search of “pure green coffee extract” (I’ve taken out a local pack for this article):

The results for shopping in the top row suggest commercial intent, however these organic outcomes are a mixture of commercial and informational results. Amazon offers a Block of five product results and this isn’t a scenario that suggests the brand or navigational intention (I haven’t stated any specific desire in Amazon).

It’s simple to cherry-pick and it’s true that Google hasn’t resolved the issue But what do the overall findings reveal? There’s no doubt that there was a bit of improvement. This improvements were in line with Google’s public announcements. SERPs with three-five outcomes (two up to 4 duplicates) from the main site were reduced a small amount but in most cases, those SERPs were still showing two results on the dominant website (one two results).

In spite of the fact that the change was isolated however, the change was comparatively small and there was no change in SERPs where 6 to 10 results were from the most popular domain. This could be due to the fact that many of these queries had a significant brand or navigational intention and SERPs that had a six-to-ten count were scarce before and following the change.

Next Post