Keyword research with SEMrush

Published on: Oct 24. 2013

About the author

Why keyword research is so important?

If we compare SEO with gardening, we can think about semantic core as seeds that we will need to harvest. The more carefully you select your seeds, the better your harvest will be. The same goes for keyword research for your site. The more care you take developing the semantic core for your website, the better your results will be.

Mistakes made on keyword research wouldn't be fixed by programing part or design – your website wouldn't be able to get to the top, even if the technical part is done well.

Now we'll talk about most common mistakes young SEO specialists-entrepreneurs make, when they are working on keyword research, and give some tips on how to avoid them. Unfortunately, most guidelines for keyword research are usually quite uncertain: “Try to think as your clients do”, "define some basic keywords that you wish to receive traffic from", "define a number of high frequency requires”. Although this general advices are really useful, if you are not familiar with the SEO topic, the real value in it might be lost.

Types of keywords

In order to get a better understanding, we will start with keywords and search query types. There are several of them and they can be classified by the following:

  • by purpose. When looking for information online, users always have a purpose or intention in doing so. What are your users thinking about?
  • by frequency. Some words are more popular than others are. Does this mean that low frequency searches won't work? Let's see.
  • by competition level. For some keywords, you'll have to fight! But how do you know for which ones?

“Try to think as your clients do”, or why people use Google

Indistinct advice about the transformation of your consciousness into the mind of the customer is actually not as useless as it seems at first glance. Keyword classification by purpose is the key to how your client thinks. What is your client thinking when he enters this type of query into the search line? Let's talk about search queries that can be classified by purpose.

Quite often, users can enter common search queries. Usually they are vague and uncertain and tell us nothing about person's intentions. What did the user mean when he entered the word "flowers"? Flowers delivery? Flower pot cultivation? Book about botany? Search results for basic, non-specific search terminology usually come out equally vague and confusing.

Example 1 — This is Google SERP for “flowers”


After coming up short in the SERPs, the user will generally try to search again, this time getting a little more specific. For example: "How to grow flowers?"

This type of search queries are informative. What is the user want to learn by typing this question into the search line? First of all, he wants to get an answer for his question or maybe he looking for a consultation and advice. So, provide this information to the client! You should start a blog, post some articles and useful tutorials where users can find the information that they are searching for.

But there is another kind of search query — commercial searches. This kind of query immediately reveals the intention of the person trying to buy something. "Buy Flowers", "flowers delivery", "flower delivery".

These requests come from your potential clients, so this kind of requests are very valuable for promotion.

Example 2 — Volume, CPC and a competition rating can tell you if a keyword is valuable enough


It might seem to be an easy choice — compose common keywords for each position of your store or service you provide, adding popular commercial requests. This is a BIG mistake. The rivalry for these words is just too intense. Even by spending a lot of money and effort, it is quite difficult to get to the top of SERPs and increase sales.

"Define a number of high and medium frequency keywords"

Search requests can be of a different frequency. Every SEO specialist sets up a different upper limit in this category. Some might say that low frequency keywords have less than 3,000 requests per month, while others categorize low frequency at less than 1000. In this situation, it is really easy to get confused and make a serious mistake by throwing away a large number of keywords that could work, but you, accidently, have deemed them useless. Therefore, try to define your own average point for each specific category depending on its popularity.

For example, "car rental" is requested by users about 368,000 times per month, but "iPhone 5" — about 2,240,000. The keyword with 1000 requests per month within the topic of car rental can be useful. But for mobile phones, or "iPhone 5" in particular, it will be too small. So just define the upper limit for the most popular searches of your subjects and rank the rest of them focusing on that number.

Again, it seems to be a very easy concept — select the most frequently requested words and work with them. Seemingly, logic dictates that the more people request it, the more traffic you'll get to your website.

It is true, but it does not guarantee you a good conversion. Even though it is true, it doesn’t guarantee you good conversion.

Usually low-frequency search queries have more than one word. And so what? The thing is that every additional word in search queries can tell us a lot about consumers' intensions. And more we know about consumers' intentions — the better chances we have to make them our customers.

Neglecting the low-frequency search queries causes you to lose a lot of customers because such queries remove any ambiguousness of your potential client's consumer goals.

Competitive requests

The competitiveness of a keyword is determined not only by how frequently it is being requested, but also by how many websites use this keyword word in their SEO campaigns. In other words, it will be much harder to promote your website using the keyword, which is already profoundly used by your competitors.

Example 3 — "Tablet" is a very popular request. Thus, don't look at CPC level as an indicator of competitiveness. Instead look at Competition level, which is displayed as 1 — the highest level possible. If you check it on Google, you will see the big companies you will have to compete with for a place at the top


Now let's talk about other types of search queries.

Localization of searches

Searches pointing out location are absolutely essential to any local business whether it is an online store or a company operating on the local market. But geo-referenced requests can be of great help to any large company as well — because a big website can easily get ahead of its smaller local competitors. A person using these type of requests is telling us: “I am here and I am ready to buy!” Although, competition for geo-targeted requests is oftentimes huge, it is worth fighting for.

Brand keywords

It doesn't matter if your business is big or small, you should never forget to pay attention to brand keywords. These type of keywords may either contain your store's name or your product's name.

Composing a perfect semantic core

Only when you have the knowledge about all the types of queries, can you be sure that your keyword research was done perfectly and it's time to begin aggregating a semantic core for your website.

1. First of all, start with common requests — they will be the cornerstone of your campaign. Certainly, the volume of common words will vary from domain to domain, depending on the range of services the business provides. Normally, common requests make up 5-10% of the core. Don't forget to include brand keywords as well. Finally, if it's possible, try choosing words that are less competitive.

2. If you have a blog or somehow provide recommendations on your website, then pay attention to information requests, since 20-30% of the total amount of those will be enough.

3. Take 10-15% of high frequency commercial searchers, but..

...the base for your semantic core should be composed of commercial queries with low frequency, also known as “long-tail keywords”.

Pay attention to variations of the most popular commercial inquiries. For example: "Car rental" is an incredibly popular query; "car rental deals" or "car rental companies", “cheap car rental in Boston” are less popular, but still very useful. A customer who requests these phrases also expresses his intention to make a purchase, and if you come up as the top option in SERP ranking for that particular term, then the possibility to convert this consumer into a buyer increases.

Focus on a narrow segment. Let's say you have either the best rose supplier or roses are the most popular item in your store. Use this knowledge to your advantage! Try all sorts of options: "bouquet of roses in a basket", "buy a bouquet of roses”. No, these requests are not popular, but they will lead you to those consumers that are ready to purchase.

4. Don't forget about geo targeted requests — if needed they can comprise 10-15 % to the total amount of keywords in your semantic core.

Lastly, the most important thing that you need to understand when working with keywords is that work never stops! Don't stop adding to and changing your semantic core as your customer base and competition change.

Keyword research must be done approximately every six months.

Take time to analyze your work. Many services will help you learn the dynamics of your position for each keyword — compare and learn them. This will help you form the aggregate/collection of effective keywords for your site.


Lastest from our blog:

The Power of the Community

In the focus of recent feedback that I got, I think there is a need for me to justify some of the decisions we made while developing Webiny Framework.

Share & subscribe: