How to Feed a Hummingbird: Improve Your On-Page SEO with Related Topics in Moz Pro

[Estimated read time: 7 minutes]

SEO is changing. We can no longer rely on keyword targeting alone to optimize our content. Whether we should focus on topics or keywords is a debate in progress. But figuring out which topics can influence the SERP is, at best, a manual process; at worst, it's a timesuck that can take hours out of your day.

TL;DR

Today we've launched a new feature in Moz Pro that can help you make sense of how search engines understand topics and phrases.

You can use this data to build deeper content, improve your topical authority, find keyword ideas, and generally better understand the SERP. It uses machine learning and topic modeling to mine related topics from the SERPs.

We see this as another step on the journey to help marketers better understand the complex world of SEO in 2016. As of this moment, Moz Pro is one of the only places you can get this kind of data.

Want the quick run-down? Tori explains it all in this brief 1:39 minute video, complete with snazzy music.

Can't wait to dive in? Already an avid Moz Pro user? Head to the Keyword Rankings section of any campaign and get started. And if you're not a Moz Pro subscriber, you can satisfy your curiosity with a free trial, too:

Try it for 30 days!

Pandas, Hummingbirds, and the relationship between keywords and pages

We've all noticed that SEO has become a lot more complex in the last few years. When Google started to figure out the meaning of words and phrases, simple keyword usage alone no longer guaranteed us results. Then Hummingbird spread its wings, and now in some cases, pages in the SERP don’t contain the keyword at all. Utter chaos, right?

Panda made sure we put effort and research into our content. And while it’s still a good idea to ensure your target keywords appear in key parts of your page, the simple on-page optimization of the past can no longer move the rankings needle on its own.

Related Topics is a new feature in Moz Pro that helps you understand how phrases and topics influence the SERP, allowing you to broaden your content and build out pages instead of devoting yourself to time-consuming (and let’s be real, sort of boring) research. As of today, Moz Pro is one of the few places you can get this kind of data.

That all sounds well and good. But how do we get insight into how Google understands the relationship between topics? Well, it turns out they give us a handy clue: the SERP itself.

Related Topics examines all of the pages that rank in the top 20 for a given keyword. Using machine learning and topic modeling, it figures out which unique sets of terms and phrases those pages include. It then removes the topics that your page already talks about and presents the resulting list, along with the ranking URLs. Armed with this mighty list, you can now understand which topics have influence in the SERP and decide whether to integrate these into your own pages and content. It lives within the Page Optimization feature in Moz Pro, which you can now get to by clicking the "Optimize" next to any keyword in the ranking table.

While it’s impossible to say for sure that including topics in your page will result in a higher ranking (that ol’ correlation versus causation thing), we do know that pages that rank well are already including these topics in their content. If you’re looking to diversify and broaden your page’s subject coverage to try and win more authority, Related Topics is the place to start. Bonus points: it’s also quite likely that including coverage of these topics will improve the user’s experience of your content.

How can I use this data to get ahead?

1. Experiment with including different topics and content to build authority

Adding topically similar content to your page can help Google understand what that page is about, establishing yourself as an authority on those topics.

I'm a fan of Tim Ferris and his productivity hack blog, fourhourworkweek.com. Let’s take this article on speed reading. Looks like the page is optimized pretty well for the target keyword and has a decent link profile and PA. Now, let’s look at some other topics that have influence on the SERP.

Here I can see a couple of variations I might want to play around with, but a couple in particular catch my eye. I notice the topic “reading comprehension” seems influential (it's included in 3 of the top 5 ranking pages), and it’s not syntactically related — this is a topic I might not have discovered manually by looking at variations of the target keyword. I also see “subvocalization” being influential. This is a term I might not be familiar with, but using Related Topics, I can drill into the actual URLs mentioning that topic, learn about it, and get some inspiration for how I could build out my content to include it.

This is a particularly interesting case, as “speed reading” has a somewhat reasonable search volume of 9,900 (from Keyword Planner). In contrast, “reading comprehension” has a search volume of 18,100. If I can integrate it well, I have an opportunity to broaden my audience.

2) Avoid thin content and go deeper

You’ve got to pacify the Panda. If you’re looking for ways to expand on thin content, go deeper or broader on an existing page, or convert shorter content to long-form, using Related Topics suggestions can give you inspiration for subject-matter expansion. Multiple studies have shown that deeper and more topically relevant content correlates with better ranking performance.

In the example below, I have a page about Product Management Events, if I wanted to make it broader I might do a deep dive on the subject of Product Design, or even talk about some of the branded topics that were discovered.

3) Save time on topical, competitive, and SERP research

This can be especially helpful when you’re wearing many hats, and tackling a new domain you're not as familiar with. Using Related Topics — and especially researching the ranking pages they appear on — can give you a head start for topic-appropriate language to use, or inspiration for areas to research.

At Moz, we all think we're experts on the housing market since we watched "The Big Short." But challenge us to write about the more technical terms and we might struggle! Here's another example using a US real estate blog recommended by our own in-house real estate guru Tim Ellis.

Let’s say we want to understand a bit more about the SERP for the keyword "real estate forecast," and perform some industry research on terminology. Here are some topics that have influence:

I notice there are a few technical terms in here that I'm not familiar with, and if I want to learn more I can jump right into the ranking URLs that contain the topic and research them instead of trying to manually pull them out of the SERP.

4) Keyword Ideas

The list of topical suggestions also double as suggestions for other keywords to target, or as seed keywords for keyword research (we have some new keyword research tools coming very soon).

How does it actually work? (Tech jargon alert!)

Wondering how Related Topics knows just which content is on the page? Well, we use Moz’s proprietary Context API, which also powers other tools around here (such as Moz Content). Here are a few words from Dr. Matt Peters (Moz’s Chief Data Scientist) on how it works:

Moz's topic modeling algorithm extracts relevant keyword phrases from English language web pages. We use natural language processing algorithms to analyze the page content and create a list of candidate topics. Then, a machine learning model assigns each candidate phrase a relevance score and ranks them from most-to-least relevant. The relevance score is a combination of traditional information retrieval techniques like term frequency–inverse document frequency (TF-IDF) and language modeling, syntactic and semantic signals such as part of speech tags, and graph-based features. The resulting lists of highly relevant topics and relevance scores are used in both Moz Pro and Moz Content.

As mentioned above, Related Topics takes the top 20 ranking pages on the SERP, extracts topics from them using the Context API, and then applies a series of filters and rules to show topics that we think are relevant. We exclude topics we find on any URLs that you rank with for the keyword. During feature development, we were faced with a choice: show topics that occur more frequently, but show less of them; or show more topics with varying ranges of frequency. We decided that our customers prefer having more data, and often we find gems near the bottom of the list. For this reason we went with the “more data” option. You might find the odd strange suggestion in there, but we think that’s outweighed by having more data to choose from.

See it in action!

Want to take it for a spin? If you're already a Moz Pro subscriber (hey, pal!), head to your Keyword Rankings section in any Moz Pro campaign and hit the "Optimize Keyword" button.

Curious but not ready to commit? Check it out with a 30-day free trial:

Try it for 30 days!

As always, we want your feedback / comments / experiences in the comments below!