It is generally agreed that, with each new new algorithm update, Google gets cleverer. “Things not strings” people cry. “Keywords are dead” people exclaim. At least I’ve heard at least one person say this in the last few conferences I’ve been to. Sure – there have been plenty of experiments to try and prove this point and largely, taken at face value, I’d agree. Google knows when you type “feet fingers” you mean “toes” and it’ll show results accordingly:
It knows “best restaurant on Christmas Island” is the same query as “Christmas Island best restaurants”. The recent release of BERT makes their ability to parse queries even better as it can extract meaning from subtle sentence modifiers like “to” which can actually have a profound effect on a sentence. In short, there is no need, and no value, in creating landing pages targeting every single keyword permutation any more. This isn’t a post to prove that Google still shows different results based on slightly different keyword variations, which it actually still does based on my experience and observations (the last legitimate experiment I can find was in 2016 which is far too long ago so I’ll be conducting my own soon). This post explores the other benefits and insights keyword research can bring, providing it’s been done properly.
Keyword Research and Site Taxonomy
I don’t like bitching about other people’s work. It’s bitter and childish. You can have 50 people in the same room and they’d all do a task differently with varying results. That said, whenever I meet a new client and ask to see the keyword research their previous agency did for them (to get a sense of opportunity), my heart drops when they present me with a document of 100 keywords. If you ran a search query report from their Google ads on one campaign alone you can see there are literally thousands of ways people can type a query in. A proper keyword research document should contain thousands of keywords – mine often in the region of between 5k-50k depending on the business. You should be aiming to rank a “cohort”, or “category”, of keywords rather than just one individual keyword. When you have a significant volume of keywords, you can actually see patterns in how people search. You start to see which parameters are important for people and what characteristics they are interested in. This is especially prudent for e-commerce sites. Imagine a clothing retailer selling shirts. Your site taxonomy might prioritise shirts based on purpose over colour because that’s how you, the business owner, shops. So in your website menu you have a situation where a user hovers over shirts and the options are “shirts for work” and “shirts for school” and “shirts for nights out”. You, the business owner, have fallen foul to your own shopping bias. Had you done proper keyword research you might have discovered that actually, people search for “white shirts” and “black shirts” far more than “shirts for business”. In that instance, it would have been more important to categorise your shirts by colour over purpose. Now I have no idea if this is true, and the reality is a lot more complex. In reality there are probably a number of factors people search for shirts for – likely by colour, purpose, size and gender (to name a few). You’ll want to be visible for all of these and need to have a decent faceted navigation to account for this. Your shopping filters will need to be set up in a way where you have static pages covering the bases of all major “facets”. On the other hand, you’ll also want to set up a situation whereby you noindex facets that aren’t important to save on URL bloat. My point is, you won’t truly know what facets are important to include, exclude and what you should have in your website navigation without having conducted proper keyword research. A basic example of product taxonomy can be seen below:
Keyword Research and Product Opportunities
Another advantage to a business that’s conducted a thorough keyword research is that of product opportunities. If the person carrying out your keyword research is even slightly capable, they’ll look at “keywords” indirectly related to your products and topics. When this happens, you may discover products or offerings that you don’t currently provide, but which wouldn’t be too complex to add to your portfolio. For example, you may sell “car insurance”, “bike insurance”, “van insurance” and “taxi insurance”. Having conducted keyword research you might find a demand for “limo insurance”. Again, I don’t know the “ins and outs” of commercial insurance – it’s likely that “taxi insurance” is legally all you need for “limo insurance”. But what do you think will convert better? If you, the user, are searching for “limo insurance” you’d expect to land on a page which explicitly mentioned this right? Thorough keyword research will identify the key terms and pain points users have which business owners should be made aware of, and capitalise on, by creating dedicated pages or new products altogether. Consequently, conversion rates will go up, and you’ll add more money to your bottom line.
Keyword Research and Seasonality
One of the most important aspects of keyword research should be categorisation. I believe in this so much that I created a keyword categorisation tool. The keyword research documents I produce categorises keywords into at least 3 categories, but sometimes as many as 5. When keywords are categorised properly, providing you also have search volume metrics pulled in as part of your keyword research, you should be able to pivot the data to get seasonal volumes for each “topic cluster” and optimise for that. What does that ridiculously long sentence mean? Well it means this. Say you have 3000 keywords to do with “t-shirts”. Your keyword research should cover everything including “white t-shirt”, “where can I buy a black t-shirt” and “how to make a tie dye t-shirt” etc.
Your keyword research will sum the volume of search volumes across each month (providing you’ve pulled it in) and you should be able to easily produce a graph so that you can apply a filter and see that people looking for t-shirt related queries normally do so in June (for example). With this information, you now know to create content at least 3 months prior to the peak month in search volume (to give it time to be indexed by Google) or allocate more budget in Google ads to your “t-shirt campaign”. Of course you can do this with tools like Google trends, but you can only really compare a maximum of 6 keywords at a time and it uses a relative index as opposed to a proper volume:
Whereas a proper keyword research document will allow you to view seasonality with a lot more accuracy as you are comparing many (often 500+) semantically related keywords:
In this example, I’ve filtered for all keywords to do with mens shirts. We can see it peaks in July, so content should be created in April (for SEO) and more budget should be set aside in July (for PPC). In short, your keyword research can literally plan your content calendar for SEO, and your campaign budgets for PPC providing it’s done correctly.
Keyword Research and Topic Feasibility
Building on from the above point, if keywords are categorised correctly and organic difficulty data is pulled through, your business should be able to pivot the date to understand which “topics” should be focused on for quick results. It is rare you will be working for a company which has such high DR ( or DA, trust flow + any other arbitrary metric) that you can just post content and it’ll rank. And yet the pressure will be on for you to prove yourself quickly for your client/employer to keep you on. If you’ve conducted proper keyword research and pulled in all the relevant metrics you should be able to produce bubble diagrams like this:
This graph tells me, for this particular client, “suits” as a category should be our short term priority as we’ll see organic gains quickest based on search volume vs difficulty.
Making individual pages for each keyword is pointless. That I agree with. If you’re making different pages targeting the following keywords – “shoes in Birmingham”, “Birmingham shoes”, “where can I buy shoes in Birmingham” then you are doing it wrong. To that extent, keyword research is dead. In fact, it’s likely you’ll be penalised for creating “gateway pages” (That said, I will be doing an experiment on this shortly as slight deviations in keyword searches still seem to produce different results in the SERPs.) However, if you treat keyword research as more of a “market research” task, you’ll be leagues above your competitors. Keyword research can identify missed product opportunities. It can predict seasonality trends for a given product. It can guide your site taxonomy. You can use it to produce relevant landing pages, content calendars and structure your Google ads campaigns. So keyword research isn’t dead, but my takeaways are this:
- Keyword categorisation is incredibly important – more so than ever to find content topics and clusters (let’s be honest ahrefs does a good job but it’s not there yet). Your aim shouldn’t be to rank for high volume individual keywords anymore, but a cohort of keywords. This can only be done with effective categorisation. To truly excel, you need to track your successes against a category of keywords rather than an individual keyword. If you put all your eggs in one keyword, one algo update might lose you 50% of your traffic.
- Keyword seasonality is incredibly important – Google trends is good, but to be accurate you need to know the month by month volume of a series of related keywords. This will help you forecast your business’s “busy periods” as well as planning content/ads budgets that will land with customers when they are most prone to searching for it. To be successful you need to be visible when your customer needs you.
- Keyword breadth is incredibly important. You can’t make decisions based on one or two keywords. You need hundreds, if not thousands, of keywords around a given topic to make conclusions that will benefit your business.
- Above all – keyword research is not dead. The goal may have changed; it’s not to create individual landing pages targeting each query any more but rather to create “sections” of a site optimised towards “cohorts” of keywords. Keyword research needs to be a lot more than a list of 100 keywords that you fob into your keyword tracking software. It needs to be approached (and delivered) in a way which is given as much kudos as a “market opportunity” document produced by a business analyst. I say this because, if done properly, it can be just as valuable.
Comments? Let me know. Maybe I rambled too much. Cheers, Special thanks to Suganthan for getting me off my arse and pushing me to start writing.