Placing users at the heart of your business requires research and lots of it. In the past, most people used focus groups and think tanks. But in the age of digital marketing, we find this is the foundation of SEO: Keyword Research.
Understanding the user and providing them with answers, products and services, is the cornerstone for internet marketers. No form of marketing is quite like it and without keyword planning – SEO wouldn’t exist.
Completing keyword research gives you insight into the users’ intention. This information can be used to provide your customers with helpful content and improve sales. But the most important reason why you should do keyword research is that it works.
What is Keyword Research?
Keyword research is the building block for any SEO campaign. It requires analysts to review the searches performed by users to provide them with better quality content. There are many ways to accomplish this analysis, and no two people do it the same.
The practice has been going on for years, but with all the third-party tools that are available – it’s become a whole new science. One the one hand, keyword research is about finding terms that you can use on your website. On the other, it’s about understanding users.
Below we’re going to cover all aspects. Each section will come with some helpful tips to get started.
How to find Keywords?
Finding great terms that will bring valuable traffic is the first step in keyword research. But to those who haven’t started and those stuck in a rut – it can be hard to know where to start. For me, these are my favourite keyword research tools:
- Adwords Keyword Planner
- Search Predictions
- Suggested Searches
- Google Trends
- LSI Graph
- AHREFS Keyword Explorer
- Moz Keyword Explorer
- SEMRush Organic Research
Many of these tools are keyword research tools are free to use, but some require a subscription. My personal favourite to use is the Ahrefs Keyword Explorer, but a combination of the above tools gives the best results.
Start by searching for your product name, or by searching 2 – 3 words that sum-up your product. For example, Apple Watch is the product name but would fall under searches such as “digital watches” and “smart watches”.
From here you can use the search predictions to get some ideas, and at the bottom of each page are suggested searches. There is some overlap between these two, but you can sometimes find a couple of different keywords.
Then use Google Trends, LSI Graph or one of the keyword research tools to find other terms. This can reveal parent categories that are great for target terms, and long tail keywords that are great for headings in your content.
Choosing the keyword you want to rank for can be difficult and people generally go for the highest search volume. These terms are often the most competitive and the lowest conversions, which makes them poor for SEO.
I find that the best way to categorise keywords is by the following:
- Broad Terms – these are going to be included across your website naturally. They’re huge search volume but ambiguous user intent.
- Target Terms – these are the core keywords you should target on individual pages. They’re around 800 to 2,000 monthly searches and clear intent.
- Long Tail Keywords – these are terms with low search volume and typically 5+ words in length. They’re around 10 to 100 monthly searches and the clearest intent.
Each page should focus on a single target term, with relevant long tail keywords scattered inside the page. But you still need to qualify those terms to see whether it’s realistic to rank for them. The Ahrefs Keyword Explorer gives a lot of insight.
Inside this tool, you will find metrics on difficulty, search volume, click-through rates, and paid vs organic clicks. There are also suggestions on related terms to help plan out your content and get new ideas.
You can find out more about this tool and other great features in my Ahrefs tips and tricks.
Completing keyword research is only valuable if you start to understand your target audience. Understanding the user intent and what they are trying to achieve is essential. But other factors can include where users are searching from and when.
The best tool for this is Google Trends, and it’s a free keyword research tool provided by Google. Below is an excellent example of how Google Trends can show you which countries are searching for your products and it even offers state-by-state or county-by-county details.
There’s a lot more that can be done, so check out my article on How to Use Google Trends.
Related Queries & Topics.
A great way to find more keywords while understanding the user intent is through related queries and topics. These can be found in Google Trends and offer insight into what the user is most interested in for that term.
If you want to improve your market research, this is a vital step to include. It helps you to dig deep into the user intent and understanding what questions these users are looking to answer. The more of their questions you can explain, the longer they’ll stay on your page.
Term Frequency-Inverse Document Frequency.
TF-IDF is an information retrieval algorithm designed to find the most frequent terms in a document and compare them to a corpus. The more frequently a term appears on the page, the higher the value it has, but the more frequently it’s used in a corpus, the lower the value.
There are different ways to calculate TF-IDF, but the most basic way is as follows:
(Term Frequency / Total # of Terms) * log(Total # of Documents / Documents with Term)
Here are some examples of how that weighting might look:
- (13/1500) * log(1,000,000,000/1,000,000) = 0.026
- (22/1000) * log(1,000,000,000/500,000) = 0.073
As you can see the more frequently the term shows up within the document and the fewer words, the higher the weighting. This is offset by Google’s total collection of documents divided by those including that keyword.
You can find out how many documents include your keyword by searching on Google. See below:
Latent Semantic Indexing.
Latent Semantic Indexing is the mathematical analysis on the relationship between words. It seeks to find words that are semantically related and not just the words that have the same meaning. While TF-IDF is often misunderstood, LSI is even more so.
Most SEO hobbyists are repeating articles they’ve previously read. They view LSI as just related queries or synonyms, but those are separate. This is why tools like LSI Graph are so popular because they feed into this concept.
Google has a tool called word2vec, and it’s used to find relationships between words. For example, an article discussing the UK’s National Health Service may include the following words that are semantically different and yet related:
Tools that merely give you a list of related queries and synonyms are useful, but they’re not the same thing. In my opinion, the best way to find relevant questions is to read your competitors content. To find out what you should look for, keep reading below.
A synonym is another word that is semantically similar but may have subtle differences. For example, research implies the discovery of something new while analysis is the reflection of something found. However, both are similar in meaning and could be used interchangeably.
While I do recommend that you use synonyms in your content, it’s important not to go overboard. Too many synonyms might make your content unclear, while too few may make it repetitive and boring. Try to be natural, yet creative.
One of the most overlooked ways to improve content quality is through proper usage of verbs. The following words are not synonyms they’re the same word in different tenses: happen, happening, happened; choose, choosing, chosen.
Create inspiring content when you choose your words carefully. Avoid choosing words that were already chosen and focus on rhythm and coherence of sentences. Carefully select which tense is appropriate for each sentence:
- Past – an event that has happened, or a previous state.
- Present – an event that is happening, or a present state.
- Future – an event that will happen, or a future state.
- Past Perfect – something that occurs in the past, but before the subject’s event.
- Present Perfect – something that began in the past and is ongoing.
- Future Perfect – something that will happen but is yet to be completed.
The best way to find a variety of ways to use each word is through Google’s online dictionary. Just search for “define word” or visit Dictionary.com for the same data in a less user-friendly format.
An NGram is a series of words that often come paired together. It’s based on earlier works by a man named Claude Shannon, who wanted to find the probability of the next letter with any given sample.
Google has kindly provided an NGram Viewer for all the books they analysed and offer some great examples. The name Albert is often referenced alongside the surname Einstein, but this started around 1920. This is shortly after he graduated.
If you check your competitors content, certain NGrams frequently show up. Within the health industry, you will almost always find “loved ones” and “highly-trained nurses” in each page. Consider including these to solidify content relevance.
Other Information Retrieval Models.
There are alternative ways to retrieve information from a page, some provide more personalised results but others provide aggregated results. These are helpful for different types of queries and I wouldn’t be surprised if Google used them.
Here are some examples of the ones that I’ve already reviewed:
- TF-PDF is used to aggregate similar content and return important topics, great for news syndication and highlighting top stories.
- TF-IDuF is similar to TF-IDF but also takes into account previously viewed documents to add emphasis, great for personalised searches.
What is Keyword Stuffing?
Keyword stuffing is the overuse of a specific phrase within your content, specifically to rank highly in Google. It delivers low-quality content to the user because it becomes tedious and difficult to read.
This can be done on individual pages, but can also be done site-wide. It is common for beginners to use a single keyword across multiple pages to try and make the whole site relevant. This creates an issue called keyword cannibalisation that causes harm.
Below, you’ll see an example of content that has stuffed ‘personal injury’ and ‘injury lawyers’ into the content as much as possible. There are many ways they could have written this content – but they chose keyword stuffing.
What is Keyword Cannibalisation?
Keyword cannibalisation is the name for two pages competing for a single keyword. This results in neither page ranking long-term and unable to reach the top of the search engine results pages.
If you read my article on Google Ranking Factors you will find that one of the algorithm rules is domain diversity. This means that Google will only serve a single page from your domain for each keyword. If Google cannot detect which page to rank, it will flop between the two and result in keyword cannibalisation.
To fix this, you will need to figure out what caused the cannibalisation in the first place.
Hummingbird Algorithm Update.
The Hummingbird update was designed to improve conversational queries. To do this, Google wanted to understand the nature of a query, as opposed to the specific words typed by each user.
This means that factors such as Latent Semantic Indexing became more important, whilst the traditional TF-IDF approach became less effective. Including specific terms is no longer as important as answering specific queries.
Hummingbird likely also decides when to use SERP features such as a map pack. If there’s a query never searched before but includes ‘near me’, then it will likely try to find locations near you that are relevant to the rest of the query.
The Future of Keyword Research.
Whilst keyword research has dropped in value for short-term marketers, it’s becoming increasingly valuable to those with long-term strategies. Understanding your target audience and their questions are the most important part of a search.
Tools such as Keyword Explorer offer great insight into user behaviour and helping choose the right focus keywords. As our perception of keyword research moves away from stuffing strategies and towards focus group data – we’ll find it continues to deliver value.
However, for the moment, this continues to be one of the most powerful ways to improve relevance signals for each page.