How Google Decides Which Brands It Likes: 50 Site Case Study

In the world of SEO, there are always two sides to the story. The first is what Google says is happening. The second is what the SEO community sees happening in practice.

If you asked Google what’s happened over the last couple of years, they will tell you that they have been focusing on rewarding “helpful, reliable, people-first content” that demonstrates E-E-A-T.

It’s clear that Google have been making significant adjustments to the search algorithm. But what that looks like in practice is somewhat different to how they’ve described it.

In practice, what the SEO community has witnessed includes: the introduction of AI overviews on a huge percentage of informational queries; dramatic increases in visibility for large forum sites and select legacy media sites; and drastic decreases in visibility for independent media sites and many small businesses.

At the same time, few in the SEO community would report any improvement in the “helpfulness” of the content that Google is serving in the search results. On the contrary, there have been thousands of cases of Google punishing top-quality expert-written content, and replacing it with content written by anonymous accounts on forums (like Reddit), or with generic low-value content on non-expert sites (like Forbes).

So what’s really going on with the algorithm? I went in search of an explanation.

Like many in the SEO community, I had the hunch that the strength of a brand had a lot to do with whether Google likes a website or not.

But how does an algorithm evaluate the strength of brand? This study provides a partial answer to that question.

I sifted though a tonne of data on 50 brands, and ended up focusing on these key metrics:

• Year-on-year organic user growth. This is the metric by which I sort the “winners” and “losers”
• Organic traffic as a % of total of all site traffic
• Non-branded organic traffic as a % of total of all site traffic
• % of branded vs non-branded organic traffic
• % of paid Google traffic as a % of total of all site traffic (including search, cross-network, display, shopping)
• Domain Rating (Ahrefs)
• Domain age

Key Findings

As I mentioned, my hunch was that the algorithm has a way of understanding which brands it likes, and which it doesn’t like. But what data does the algorithm use to decide which brands to reward?

My key three findings were:

• The algorithm likes sites that have an overall low % of non-branded organic traffic
• The algorithm dislikes sites that have a higher % of non-branded organic traffic
• The algorithm likes sites that use Google ads for a large % of their overall traffic

I will break down the data on each of these below, and then expand on the less conclusive data.

Naturally, I’ll save the big takeaway for last.. The final section highlights what I think is the most important finding–what I’ll call the “golden ratio” of overall traffic make-up.

About the sample

The study is based on 50 brands, so any conclusions are, of course, tentative. A larger sample could potentially confirm the findings. On the other hand, it’s possible that a larger sample could reveal that the findings were inaccurate.

The brands in the sample cover a very wide range of sectors. However, it is perhaps significant that the sample is slightly skewed towards small and medium businesses (fewer than 250 employees), although it does include ten large businesses (greater than 250 employees).

The majority of the businesses in the study (70%) have annual revenues below £10m, with the other 30% of the businesses exceeding £10m in revenue.

This table demonstrates how I have categorised the different brands in the sample. The findings focus chiefly on the “very bad performers” and the “excellent performers,” which is where trends were most noticeable.

Performance is determined by percentage change in organic users year-on-year. At the bad end of the scale, brands have experienced a dip in organic traffic whereas at the other end of the scale, brands have experienced significant organic growth year-on-year.

Trend 1: the algorithm likes sites with overall low % of non-branded organic traffic

A strong trend amongst the excellent performers was a low percentage of non-branded organic traffic in relation to the site’s total traffic. To calculate this figure, I looked at the site’s total organic traffic, and used Semrush to get an estimate of how much of that traffic was non-branded (i.e. generic).

8 out of 10 of the excellent performers received under 20% of their traffic from the non-branded organic channel. In reality, I think that 9 out of 10 would actually be the more accurate figure, as there are some inaccuracies in the data. Either way, 8 of 10 is enough to demonstrate the trend.

Trend 2: the algorithm likes sites that pay Google for a large % of overall traffic

The second trend is that the excellent performers tend to pay Google (through Google Ads) for a large percentage of their overall traffic. The data showed that 7 out of 10 excellent performers were using Google ads for at least a third of their traffic. In fact, 3 out of these 10 were paying Google for over two-thirds of their total traffic. On the other hand, the worst performers were generally only paying Google for a very small amount of traffic, or none at all.

The algorithm dislikes sites that have a higher overall % of non-branded organic traffic

Based on the above assertion that Google likes sites with a low percentage of non-branded organic traffic, you might suspect that the inverse is true, and, indeed, it is. Aside from two outliers amongst the “good performers,”² high non-branded organic traffic correlated closely with very poor performance.

Inconclusive findings

Naturally, I came upon many dead-ends in this study before patterns started to emerge. But a couple of these dead-ends are worth sharing, so here they are.

Old domains

Having an old domain didn’t seem to have a big impact on performance. Although there was a slight tendency for good performers and excellent performers to have an old domain, quite a few bad performers also had old domains. I think this is because the bad performers group contained a lot of big, old companies in competitive niches where change happens very slowly.

High DR

High Domain Rating (according to Ahrefs) did not seem to make much difference to overall performance. I think this is mostly likely because the Domain Rating required to compete in different niches is very different. It is notable that only one of the excellent performers had a Domain Rating of over 40.

Low DR

Remarkably, I found that the high performers tended to have a fairly low Domain Rating. This was quite unexpected, but potentially significant. My explanation would be that many of the top performers are companies who had historically not done much SEO work, such as link-building. Rather than investing in SEO and link-building, they have focused on other channels, such as paid Google traffic, which Google prefers, and tends to reward.

The Golden Ratio

Now here’s the interesting bit. Considering my hunch that “brand” is important, I was expecting to find that branded vs non-branded (generic) traffic data would reveal a trend. 

Namely, I was expecting that brands that received a high percentage of branded searches relative to non-branded searches would perform better, and that brands with a low percentage of branded searches would perform worse. It turns out it’s a little more complex, and a lot more interesting than that.

The table below looks at three of the worst performing brands, who all had a very low percentage of branded traffic. The table also includes three of the top performing brands who happened to also have a very low percentage of branded searches.

What this comparison reveals is that it doesn’t matter if not many people are searching for a brand. As long as organic traffic makes up an overall low proportion of total traffic, and as long as a brand is paying Google for a significant proportion of their traffic, they can outperform in organic search.

Google’s Reasons For This Approach

Don’t get me wrong, I’m not in a hurry to defend Google here, but I do think that all of this makes some sense.

You probably will have noticed the implication that paid traffic is a ranking factor. On the one hand, of course, there is the cynical reason to think that Google likes sites that pay them for ads.

But on the other hand, being able to pay Google tells Google that that particular brand has resources. If they have resources, it’s an indication of sorts that we are dealing with a legitimate brand, rather than just a well-SEOed shell that basically doesn’t exist outside of Google Search.

However, Google liking sites that pay them is only half of the picture. The other essential part of the picture is that Google doesn’t like sites that have high non-branded (generic) organic traffic. Google likes sites with low generic organic traffic.

Why might this be? A great study from Cyrus Shepard suggests that Google seems to be punishing “good SEO.” And I think this makes a lot of sense. Google no longer wants to rank the brands that have the best SEO. It wants to find more reliable, less “gameable” ways of deciding who should rank.

That’s why the non-branded organic traffic metric is so important. This metric needs to be in the correct proportion to other traffic sources, as this demonstrates a balanced and natural-looking make-up of traffic. A healthily low proportion of non-branded organic traffic seems to suggest to Google that it is dealing with a “wholistically” successful brand, as opposed to a brand that just happens to have good SEO.

Because, in fairness, users aren’t looking for “good SEO.” And Google is, arguably, trying to serve its users. At least, it was until recently.

Conclusion

This study was based on the hypothesis that Google has a way of identifying which brands it likes, and rewarding them with greater visibility in organic search.

The hypothesis was based on my own hunches, and the hunches of many others, from what we have seen, anecdotally, on the ground in the world of SEO. I set out to see what the data might reveal–unsure what I would find–and, personally, I think it has revealed a couple of real insights.

Like many in the SEO industry, I don’t think that the Helpful Content Update made Google Search better. And like many, I am doubtful about how serious Google is about serving the best content to its users.

But the point of this study is not to ask whether Google Search is getting better or worse. The point is to understand the mechanisms by which the algorithm decides to reward one site and punish another.

I think this study does point to a certain logic, whether you agree with it or not, by which the algorithm attempts to identify and reward certain brands in organic search.

The findings point to a future of SEO in which SEO professionals must be increasingly mindful of the part that SEO plays within the wider marketing mix. It highlights how increasingly challenging it is for brands to beat their competition at SEO alone. In fact, the findings suggest that the algorithm could be preventing brands from winning with organic search if they are not investing sufficiently in other channels. And whilst that will pose a challenge to laggards in the field of the SEO, in the end, that may not be such a bad thing.

Footnotes

¹ I attribute the relatively large numbers of bad performers with low generic traffic to the fact that many of these are quite big, incumbent businesses in competitive niches. As we will see later, many of these sites have old domains. They also tend not to pay much for Google ads. Many of these sites could be described as entrenched and slow moving, and overall not very proactive with SEO or search marketing generally.

² I looked into these two outliers. Notably, they were both in niches where there was fierce SEO competition, and where no one was paying Google for traffic.