{"id":7682,"date":"2018-11-20T16:02:50","date_gmt":"2018-11-20T16:02:50","guid":{"rendered":"https:\/\/www.brainlabsdigital.com\/?p=7682"},"modified":"2025-06-20T19:26:41","modified_gmt":"2025-06-20T19:26:41","slug":"how-to-rank-for-head-terms","status":"publish","type":"post","link":"https:\/\/www.brainlabsdigital.com\/how-to-rank-for-head-terms\/","title":{"rendered":"How to rank for head terms"},"content":{"rendered":"\n

Over the last few years, my mental model for what does and doesn\u2019t rank has changed significantly, and this is especially true for head terms \u2013 competitive, high volume, \u201cbig money\u201d keywords like \u201ccar insurance\u201d, \u201claptops\u201d, \u201cflights\u201d, and so on. This post is based on a bunch of real-world experience that confounded my old mental model, as well as some statistical research that I did for my presentation at SearchLove London in early October. I\u2019ll explain my hypothesis in this post, but I\u2019ll also explain how I think you should react to it as SEOs \u2013 in other words, how to rank for head terms.<\/p>\n\n\n\n

My hypothesis in both cases is that head terms are no longer about ranking factors, and by ranking factors I mean static metrics you can source by crawling the web and weight to decide who ranks. Many before me have made the claim that user signals are increasingly influential for competitive keywords, but this is still an extension of the ranking factors model, whereby data goes in, and rankings come out. My research and experience are leading me increasingly towards a more dynamic and responsive model, in which Google systematically tests, reshuffles and refines rankings over short periods, even when sites themselves do not change.<\/p>\n\n\n\n

Before we go any further, this isn\u2019t an \u201cSEO is dead\u201d, \u201clinks are dead\u201d, or \u201cranking factors are dead\u201d post \u2013 rather, I think those \u201ctraditional\u201d measures are the table stakes that qualify you for a different game.<\/strong><\/p>\n\n\n\n

Evidence 1: Links are less relevant in the top 5 positions<\/h2>\n\n\n\n

Back in early 2017, I was looking into the relationship between links and rankings, and I ran a mini ranking factor study which I published over on Moz<\/a>. It wasn\u2019t the question I was asking at the time, but one of the interesting outcomes of that study was that I found a far weaker correlation between DA and rankings than Moz had done in mid-2015.<\/p>\n\n\n\n

\"Moz<\/figure>\n\n\n\n

The main difference between our studies, besides the time that had elapsed, was that Moz used the top 50 ranking positions to establish correlations, whereas I used the top 10, figuring that I wasn\u2019t too interested in any magical ways of getting a site to jump from position 45 to position 40 \u2013 the click-through rate drop-off is quite steep enough just on the first page.<\/p>\n\n\n\n

Statistically speaking, I\u2019d maybe expect a weaker correlation when using fewer positions, but I wondered if perhaps there was more to it than that \u2013 maybe Moz had found a stronger relationship because ranking factors in general mattered more for lower rankings, where Google has less user data. Obviously, this wasn\u2019t a fair comparison, though, so I decided to re-run my own study and compare correlations in positions 1-5 with correlations in positions 6-10. (You can read more about my methodology in the aforementioned post<\/a> documenting that previous study.) I found even stronger versions of my results from 2 years ago, but this time I was looking for something else:<\/p>\n\n\n\n

Domain Authority vs Rankings mean Spearman correlation by ranking position<\/h3>\n\n\n\n
\"domain<\/figure>\n\n\n\n

The first thing to note here is that these are some extremely low correlation numbers \u2013 that\u2019s to be expected when we\u2019re dealing with only 5 points of data per keyword, and a system with so many other variables. In a regression analysis, the relationship between DA and rankings in positions 6-10 is still 98.5%\u00a0statistically significant<\/a>. However, for positions 1-5, it\u2019s only around 41% statistically significant. In other words, links are fairly irrelevant for positions 1-5 in my data.<\/p>\n\n\n\n

Now, this is only one ranking factor, and ~5,000 keywords, and ranking factor studies have their limitations.<\/p>\n\n\n\n

\"moz
Image: https:\/\/moz.com\/blog\/seo-ranking-factors-and-correlation<\/a><\/sub><\/em><\/figcaption><\/figure>\n\n\n\n

<\/p>\n\n\n\n

However, it\u2019s still a compelling bit of evidence for my hypothesis. Links are the archetypal ranking factor, and Moz\u2019s Domain Authority* is explicitly designed and optimised to use link-based data to predict rankings. This drop off in the top 5 fits with a mental model of Google continuously iterating and shuffling these results based on implied user feedback.<\/p>\n\n\n\n

*I could have used Page Authority for this study, but didn\u2019t, partly because I was concerned about URLs that Moz might not have discovered, and partly because I originally needed something that was a fair comparison with branded search volume, which is a site-level metric.<\/p>\n\n\n\n

Evidence 2: SERPs change when they become high volume<\/h2>\n\n\n\n

This is actually the example that first got me thinking about this issue \u2013 seasonal keywords. Seasonal keywords provide, in some ways, the control that we lack in typical ranking factor studies, because they\u2019re keywords that become head terms for certain times of the year, while little else changes. Take this example:<\/p>\n\n\n\n

\"seasonality<\/figure>\n\n\n\n

This keyword gets the overwhelming majority of its volume in a single week every year. It goes from being a backwater search term where Google has little to go on besides \u201cranking factors\u201d to a hotly contested and highly trafficked head term. So it\u2019d be pretty interesting if the rankings changed in the same period, right? Here\u2019s the picture 2 weeks before Mother\u2019s Day this year:<\/p>\n\n\n\n

\"rankings<\/figure>\n\n\n\n

I\u2019ve included a bunch of factors we might consider when assessing these rankings \u2013 I\u2019ve chosen Domain Authority as it\u2019s the site-level link-based metric that best correlates with rankings, and branded search volume (\u201cBSV\u201d) as it\u2019s a metric I\u2019ve found to be a strong predictor of SEO \u201cranking power\u201d, both in the study I mentioned previously and in my experience working with client sites. The \u201cspecialist\u201d column is particularly interesting, as the specialised sites are obviously more focused, but typically also better optimised. M&S (marksandspencer.com, a big high-street department store in the UK) was very late to the HTTPS bandwagon, for example. However, it\u2019s not my aim here to persuade you that these are good or correct rankings, but for what it\u2019s worth, the landing pages are fairly similar (with some exceptions I\u2019ll get to), and I think these are the kinds of question I\u2019d be asking, as a search engine, if I lacked any user-signal-based data.<\/p>\n\n\n\n

Here\u2019s the picture that then unfolds:<\/p>\n\n\n\n

\"2<\/figure>\n\n\n\n

Notice how everything goes to shit about seven days out? I don\u2019t think it is at all a coincidence that that\u2019s when the volume arrives. There are some pretty interesting stories if we dig into this, though. Check out the high-street brands:<\/p>\n\n\n\n

\"major<\/figure>\n\n\n\n

Not bad eh? M&S, in particular, manages to get in above those two specialists that were jostling for 1st and 2nd previously.<\/p>\n\n\n\n

These two specialist sites have a similarly interesting story:<\/p>\n\n\n\n

\"florist<\/figure>\n\n\n\n

These are probably two of the most \u201cSEO’d\u201d sites in this space. They might well have won a \u201cranking factors\u201d competition. They have all the targeting sorted, decent technical and site speed, they use structured data for rich snippets, and so on. But, you\u2019ve never heard of them, right?<\/p>\n\n\n\n

But there are also two sites you\u2019ve probably never heard of that did quite well:<\/p>\n\n\n\n

\"small<\/figure>\n\n\n\n

Obviously, this is a complex picture, but I think it\u2019s interesting that (at the time) the latter two sites had a far cleaner design than the former two. Check out Appleyard vs Serenata:<\/p>\n\n\n\n

\"flower<\/figure>\n\n\n\n

Just look at everything pulling your attention on Serenata, on the right.<\/p>\n\n\n\n

Flying Flowers had another string to their bow, too \u2013 along with M&S, they were one of only two sites mentioning free delivery in their title.<\/p>\n\n\n\n

But again, I\u2019m not trying to convince you that the right websites won, or work out what Google is looking for here. The point is more simple than that: Evidently, when this keyword became high volume and big money, the game changed completely. Again, this fits nicely with my hypothesis of Google using user signals to continuously shuffle its own results.<\/p>\n\n\n\n

Evidence 3: Ranking changes often relate more to Google re-assessing intent than Google re-assessing ranking factors<\/h2>\n\n\n\n

My last piece of evidence is very recent \u2013 it relates to the so-called \u201cMedic\u201d update<\/a> on August 1st. We work with a site that was heavily affected by this update \u2013 they sell cosmetic treatments and products in the UK. That makes them a highly commercial site, and yet, here\u2019s who won for their core keywords when Medic hit:<\/p>\n\n\n\n

Site<\/strong><\/td>Visibility<\/strong><\/td>Type<\/strong><\/td><\/tr>
WebMB<\/td>+6.5%<\/td>Medical encyclopedia<\/td><\/tr>
Bupa<\/td>+4.9%<\/td>Healthcare<\/td><\/tr>
NHS<\/td>+4.6%<\/td>Healthcare \/ Medical encyclopedia<\/td><\/tr>
Cosmopolitan<\/td>+4.6%<\/td>Magazine<\/td><\/tr>
Elle<\/td>+3.6%<\/td>Magazine<\/td><\/tr>
Healthline<\/td>+3.5%<\/td>Medical encyclopedia<\/td><\/tr><\/tbody><\/table>
Data courtesy of SEOmonitor.<\/sub><\/em><\/figcaption><\/figure>\n\n\n\n

<\/p>\n\n\n\n

So that\u2019s two magazines, two medical encyclopedia-style sites, and two household name general medical info\/treatment sites (as opposed to cosmetics). Zero direct competitors \u2013 and it\u2019s not like there\u2019s a lack of direct competitors, for what it\u2019s worth.<\/p>\n\n\n\n

And this isn\u2019t an isolated trend \u2013 it wasn\u2019t for this site, and it\u2019s not for many others I\u2019ve worked with in recent years. Transactional terms are, in large numbers, going informational.<\/p>\n\n\n\n

The interesting thing about this update for this client, is that although they\u2019ve now regained their rankings, even at its worst, this never really hit their revenue figures. It\u2019s almost like Google knew exactly what it was doing, and was testing whether people would prefer an informational result.<\/p>\n\n\n\n

And again, this reinforces the picture I\u2019ve been building over the last couple of years \u2013 this change is nothing to do with \u201cranking factors\u201d. Ranking factors being re-weighted, which is what we normally think of with algorithm updates, would have only reshuffled the competitors, not boosted a load of sites with a completely different intent. Sure enough, most of the advice I see around Medic involves making your pages resemble informational pages.<\/p>\n\n\n\n

Explanation: Why is this happening?<\/h2>\n\n\n\n

If I\u2019ve not sold you yet on my world-view, perhaps this CNBC interview with Google<\/a> will be the silver bullet.<\/p>\n\n\n\n

This is a great article in many ways \u2013 its intentions are nothing to do with SEO, but rather politically motivated, after Trump called Google biased in September of this year. Nonetheless, it affords us a level of insight form the proverbial horse\u2019s mouth that we\u2019d never normally receive. My main takeaways are these:<\/p>\n\n\n\n