Machine learning technology are the secret SEO tools of 2016. As Google and other search engines increasingly incorporate artificial intelligence into their algorithms, SEO practitioners will need to keep pace to rank well.
Until now, SEO practitioners have relied on basic keywords and phrases to focus their efforts. We know from previous patents and code that search engines like Google rely on artificial intelligence and deep neural networks to learn how people think and deliver optimized search results. If we don’t understand and use machine learning tools to evaluate our SEO efforts, we risk an algorithmic ambush.
Consider Google’s major open-source announcement in November 2015, when the company released TensorFlow to the public as open-source software. TensorFlow is a library of software for artificial intelligence; Google has used it and components of it to understand natural language in search. If we can develop similar (albeit smaller-scale) tools or even leverage TensorFlow itself, we could begin to understand how Google sees the language we use in our content.
What can we do to focus our SEO efforts with machine learning? We must learn how machines see the words we write in relation to each other; that’s what Google does. Machine learning algorithms like Latent Dirichlet Allocation (LDA) identify words that cluster together. Once we know what words naturally associate together, we can refine our SEO and content marketing efforts.
Let’s look at an example of how this might work. Suppose we work at a gin company like Nolet Spirits.
Perhaps we have a page on our site about cocktail recipes which use gin, but we’re not ranking well in search for this phrase. What might we want to do with our page to help it rank better? What content might make the most sense for us to write?
We’d start by using an SEO ranking tool to identify what pages and sites currently hold the top spots. I’m partial to SpyFu, but use whatever tool you’re most comfortable with.
Once we know what the top 10-20 pages are, we download the text from those pages to a machine learning tool. Again, this is your choice; use whatever you’re most comfortable with. If you can make TensorFlow work for you, use that. If you prefer Python and NLTK, go with what you know.
Above, we see the broad topics the LDA algorithm has identified. Note how spread out the topics are. This indicates a lot of diversity in the content we downloaded about gin. However, we see a tight cluster in the lower left-hand side; if we dig in, we find these topics all center around tonic:
How does this help our SEO efforts? Topic 12 discusses simple syrups; many of the pages we downloaded share recipes for a simple syrup for use in cocktails with tonic, or as part of making our own tonic water with cinchona bark.
If we’re in charge of Nolet Spirits’ content marketing strategy, we have an entirely new line of content we can create which is closely related to tonic water – which pairs with our gin – but isn’t directly about gin recipes per se. From our analysis, we can draw the insight that we can attract additional search traffic about tonics based on the content from top ranking sites.
Compare this to our old way of doing SEO. We’d write up pages and pages of content optimized for our product names and related generics: Nolet gin, best gin, gin recipes, gin and tonic, etc. Would we know to create content solely about tonic water? Not through this method. Machine learning identified a clustered, closely related topic for us.
Machine learning tools focused around natural language processing are the secret tools of SEO for 2016. Learn the tools. Learn how they work. Become proficient with them. Measure your SEO program by how well your topic model matches the top ranking sites in your industry. You are practicing content innovation – taking old content from other places and remixing it with your own insights to create new, intelligently optimized content.
Disclosure: At the time of this writing, I have no affiliation with Nolet Spirits; I was not asked to write about them in any capacity. I just like their gin.
You might also enjoy:
- Cómo decide Google Analytics el seguimiento de atribuciones - Christopher S. Penn - Orador principal de ciencia de datos de marketing
- AI for Marketers, Third Edition, Available Now!
- Google Analytics: A Content Marketing Engagement Test
- Unsolicited "Embargoed" Press Releases Are Absurd
- Best Practices for Public Speaking Pages
Want to read more like this from Christopher Penn? Get updates here:
Get your copy of AI For Marketers