Editor’s note: This post continues our weekly primer in SEO, touching on all of the foundational aspects. In the end, you’ll be able to practice SEO more confidently and converse about its challenges and opportunities.
Keyword research is the data that informs content strategy and optimization in search engine optimization. In this post, I’ll address how to collect and analyze keyword data. For the conceptual information needed to understand what keyword research is, how search engines use keywords and more information on keyword tools, read my previous post, “SEO How-to, Part 4: Keyword Research Concepts.”
This is the fifth installment in my “SEO How-to” series. Previous installments are:
- “SEO How-to, Part 1: Why Do You Need It?”;
- “SEO How-to, Part 2: Understanding Search Engines”;
- “SEO How-to, Part 3: Staffing and Planning for SEO”;
- “SEO How-to, Part 4: Keyword Research Concepts.”
Seeding Keyword Research
Researching keywords is similar to other forms of marketing research in that you get out what you put in: Garbage in, garbage out.
Keyword research tools build on the choice of words that you input to request data from the tool. I think of this as seeding. When you input a thorough list of seed words and phrases into the keyword research tool, it yields the greatest amount of data to analyze and inform your content plans. If you input just a few keywords or an incomplete set of keywords, you’ll get a small number or incomplete set of data back.
To brainstorm a good list of seeds, start with your own site’s navigation. Because your site is organized in a logical, hierarchical manner, it hopefully contains navigational options for pages that mirror what people search for. A men’s apparel site sells outerwear, coats, and vests as one category and set of subcategories. That’s a good start.
Now add in the synonyms and the types and styles of those items that the site sells. Coats are also called jackets. Add “jackets” to the seed list. And the coats and jackets sold on the site come in various types and styles that can also serve as seeds, like blazers, winter coats, raincoats, windbreakers, down coats, puffer coats, etc.
Keyword suggestion tools are also good for seeding keyword ideas. In this example, tools like Keyword.io and Ubersuggest might give help you add keywords you’ve overlooked like “jackets for men,” brands of jackets that the site carries like “Nike jackets,” and “bomber jackets.”
Don’t be afraid to generate a long keyword seed list. It will look daunting, but the process of collecting the keyword data can go surprisingly quickly once you get started and the resulting data is worth it’s length in gold.
Collecting and Organizing Keyword Data
The actual process of collecting keyword data is simple, if a bit tedious. All you’re doing is copying a set of seeds from your list and pasting them into the keyword tool, waiting for it to finish processing, exporting the resulting data into a spreadsheet, and repeating until all of the seeds have been entered into the tool.
When you finish passing the seeds through the keyword research tool and exporting the data for each, you’ll have a mass of .csv or .xls files. On a PC, use the command line, an archaic-looking bit of software built into your operating system, to merge files into a single file. On the Mac, the same thing can be done in Terminal.
Clean up your keyword research spreadsheet by removing duplicate rows. Don’t do this step manually – your spreadsheet program will have a command for it. On Excel on a PC, go to the Data tab and choose “Remove Duplicates.” Check the help file if you’re not sure where to find it or how to use it.
Your keyword data will have irrelevant words and phrases in it – probably a lot of them. This is natural because the keyword research tool doesn’t know your business model or the context in which you’re using the keywords. It is simply programmed to return keywords like the keywords similar to the keyword you entered.
Sort the keyword data by the number of searches, from highest to lowest. Review the top several hundred, or until the number of searches drops below a level you find valuable, and remove the irrelevant rows. Make sure to filter out other keywords like that one as well. For example, if your keyword data contains “women’s jackets but you only sell jackets for men, filter on “women” to isolate and remove all of the keywords containing “women.”
If some of the keywords that drive larger numbers of searches are relevant to your product offering but weren’t on your seed list, consider running those back through the keyword tool, as well.
Analyzing Keyword Data
Right away you’ll notice patterns in how people search, but unless you can quantify them en masse and compare them to each other, you won’t know their true value. Take the keyword data example below.
For example, do people search for “jackets” or “coats”? You can see that the keyword for “jacket” and its plural and misspellings are searched for approximately 74,000 times in an average month in the U.S., while “coat” is searched for 60,500 times.
Taken in isolation, that can lead us to believe that jacket keywords are a larger keyword opportunity. But taken in context with the other jacket-oriented keywords, across a complete set of well researched keyword data, a different picture may emerge. It’s possible for one keyword with very large search volume, like “bomber jacket” in this example, to give the impression that one type of keywords is the best opportunity. But another type of keyword, though small individually, when added up can actually present a larger aggregate opportunity. You only know by doing the analysis.
Start by making columns in your spreadsheet for the different types of keywords you see.
In the example above, I found that searchers enter queries for men’s coats around gender, type, style, material, color and brand. I added a note in each column that a keyword applied to. “Bomber jacket” is both a type of jacket and the style “bomber.”
This is a fairly subjective process, and 10 people would do the analysis 10 different ways. The important thing is that the organization makes sense to you so you can apply it in your content planning and optimization.
Don’t worry, you can always add more columns later or remove some, if the ones you start with don’t seem to be working. You’ll know you have it right when the keywords fit easily into one column or another.
Labeling a large keyword dataset can be even more daunting than collecting the data in the first place. Use the same process as you did in removing irrelevant keywords. Sort by the largest number of searches, start at the top, and work your way down while filtering for like keywords to label many at once.
Once the keywords are labeled, you can filter on the different labels or do various formulas in Excel to sum up the search value of the different keyword types. A simple example would be identifying how many people search for “jacket” keywords, as shown below.
In this truncated example, 960,900 searches for “jacket”-related keywords happen in the U.S. in Google in an average month. More complex analysis can be applied using multiple VLOOKUP formulas in your spreadsheet to identify which combinations of keyword labels are more valuable. Do people want to see jackets more by color, brand or material? Do you have filters for the most valuable ones?
Valuable analysis and results relies heavily on several factors in this process:
- Thorough seed lists to ensure relevant and complete data is collected;
- Patience in working the seeds through the keyword tool;
- Removal of irrelevant keywords;
- Thorough labeling of the different types of keywords.
Skipping or short-changing any of these steps will introduce bias in your keyword research, which will damage your content planning and optimization processes, and ultimately your natural search performance.
Read the next installment of our “SEO How-to” series: “Part 6: Optimizing On-page Elements.”