Join Eric Van Buskirk for his SEMrush webinar on July 8, "Beyond Keyword Research: Deep Data Dives with SEMrush API Modules." Sign up today!
If you are optimizing a web page for one or two keywords, you missed the news about the Hummingbird algorithm that affected 90% of all web pages. Furthermore, compared to three years ago, Google now ranks far more keywords from one contextually strong page than from many smaller pages versus.
We need to optimize more for well-constructed content than for keywords in isolation. What is GOOD content? And who says it’s good? Your aunt, or next door neighbor, or a professor in Zurich?
For ranking purposes, Google makes these judgments based largely on looking at your related keywords and topics by using semantic learning, aka latent semantic indexing (LSI). So, if you can’t track down the expert professor in Zurich to ask him if your content rocks in a niche field, you’re next best choice is to find sophisticated ways to examine what Google knows about associated topics.
Enter SEMrush’s API and the free API modules I created for making multiple-variable comparisons. The related keywords tool in SEMrush is useful for long-tail phrases, but for more popular terms, interpreting the results is like boiling the ocean.
The photo above shows the results for the search term “USA.” They are listed in descending order by search volume, but you could also order by other metrics. The problem is strength of associations is not part of the data.
This is why savvy SEMrush users use the ranking keywords by URL report, not by domain, for much of their research. If you know a page/URL ranks top for a topic you want to write about, than other words on that page from the URL tool on SEMrush will be related. Google deems the page super authoritative on that subject. That one page may be looking at the topic or keyword phrase from only one of several angles. This is why the API bulk URL module is valuable: it lets you find 15 or even 30 authoritative pages quickly and see what words co-appear across these pages. I’ve discussed these techniques in a recent post on the SEMrush blog.
The challenge for SEOs and content strategists is after we do this best-practice research for relatedness. We need to know that if we include topics — for example, in sub-headings on a page — they can actually rank for us. To do this, in our final table of 100s of related keywords we can add an additional score: likelihood our page can rank.
To do this, we examine only the ranking; authoritative pages that ranked on domains we can compete with. So, we can look at each ranked page and evaluate them based on metrics like Page Authority and the SEMrush Rank.
In the above example, I’ve sorted pages that were part of an authoritative list. I’ve circled in blue the domains and pages I’m confident my hypothetical website can compete with. On the sheet of related phrases and topics, I select those that appeared from these four pages as a starter.
Alternatively, try taking an a list of all ranking keywords in SEMrush from a few competitor domains with similar or lower domain authority. If they can rank, you can rank.
To figure out how strongly related a particular keyword is AND if we can rank on it, we can match our list from the top ranking authorities to keyword phrases and see where our competitors also rank on the same terms. These are phrases and topics to write about and we have confidence in knowing which of these are truly on topic and to what extent.
Keep in mind a site with similar authority may rank, but we want to compare only those in our niche: ranking has much to do with the holistic theme of a website’s content. If an urban newspaper ranks for a keyword but has stronger domain authority, that may not provide a good comparison if our website is directly about the niche related to the keyword.
I am doing a webinar on July 8 with SEMrush. In addition to covering the above techniques, please join to also learn how bulk search with SEMrush lets you examine content strategy for an entire industry, which I’ve written about for Search Engine Journal. We work in “bulk” so we can find trends based on comparing metrics from many domains in the same industry side-by-side.