Find the Search Terms that Relate to a Search Query with Ahrefs and InfraNodus

Suppose you want to reach customers who look for a certain topic, e.g. "ai tools". In order to do that, you need to learn which other search terms people use to find "ai tools", so that you have a better understanding of the current demand. 

This can be done using the InfraNodus keyword exploration workflow that imports suggested keywords from Google search or AdWords. However, for more detailed analysis you can also use the "matching keywords" data from the popular tools such as Ahrefs or SemRush. The advantage of using these tools is that you get an extended list of search queries: not only the keywords suggested by Google, but also the keywords that websites that rank for the original search query also rank for. 

Here's how you can do that. 

Step 1: Get a list of matching keywords from Ahrefs

In Ahrefs, go to Keywords Explorer, 

then type in a general search phrase, e.g. "ai tools"

then go to Keyword Ideas > Matching Terms in the left menu

additionally, choose to only show keywords with search volume above a certain minimum (e.g. 50)

Once you generate the table, export the the matching keywords list as a CSV file (UTF-8 encoding).

In our case, there are about 2000 keyword combinations. How do we know which ones are the most prominent ones?

Step 2: Import the CSV with the list of keywords to InfraNodus

Now we can import that CSV file to InfraNodus and visualize the results. You can get a sample in our GitHub repo https://github.com/infranodus/datasets 

In order to import the file correctly, we choose the "keyword" column as the column that will be analyzed:

In the next step, we choose the "Difficulty" and "Volume" columns as the columns for filtering. This will enable us to filter the graph by high-volume, low-difficulty keywords to better understand what are the easiest and highest reward keyword combinations to target:

In the next step, InfraNodus will build a graph with the main topical clusters.

Step 3: Visualize a knowledge graph with the topical clusters

InfraNodus will build a knowledge graph where the individual keywords are the nodes and their co-occurrences are the connections. Such representation helps you see which combinations of keywords tend to occur more often together in the same context and, thus, what are the main topical clusters around the "ai tools" search query. 

As we can see, the main topical clusters are:

• AI tools (especially for marketing)
• Image generation (especially "ghiblus")
• Blog insights
• Video creation (especially YouTube shorts)
• Ad research (audience study)

These are the main areas of interests for the users searching for AI tools. 

One thing we can do is to clean the graph by removing "ai" and "tool" nodes from it, so that we see the context around:

The topical clusters are recalculated and we can now see the topical clusters more specifically:

• Video creation
• Content detection
• Data optimization
• Teaching excellence
• Marketing automation

We can click on one of the clusters, e.g. "content detection" to see the specific keyword combinations inside to better understand that cluster. InfraNodus reveals that mainly this topic relates to the tools for AI content writing / generation and detection (the poison and the remedy):

This gives us a clear insight: if we are to make content about AI tools for text, we need to focus on the tools for content generation / writing and AI detection. Additionally, lots of AI tools are for marketing purposes, so that would most likely be the specific niche that we would target.

Step 4: Reveal high-volume, low-competition segments

To refine our analysis, we can now use the built-in filters and see how different the topical clusters are for high-volume, low-competition keywords. 

In order to do that, we will use the filters we created when we imported the CSV file and apply them to only show the keywords that have:

In our example, we select the Difficulty that is relatively low (30 to 59) and Volume that is relatively high but not the highest (to avoid maximum competition).

We also specify that we want to see the overlap of both filters: the keyword combinations that have both Difficulty from 30 to 59 and Volume from 100 to 1200:

The resulting graph reveals additional keyword combinations we should target first to avoid high competition and to still get a decent search volume:

As we're interested in text analysis tools, "data analysis" is definitely the combination we should focus on. Specifically: "ai tools for data analysis". It seems like the volume for this combination is relatively high while the competition is not the highest. 

This is further confirmed by choosing a lower difficulty keyword combination: data analysis and "market research" are the combinations to target:

Interestingly, "ai writing tools" also appears in the low-competition, high-volume graph (1st one), however, when we increase the difficulty, we start seeing "content detection" pattern emerging, which means that it doesn't make sense to target it first as the difficulty there is much higher: