How to Use Machine Learning for SEO

Machine Learning, deep learning, and neural network technologies are undoubtedly the wave of the future.  Machine learning allows computer systems to perform specific tasks without using explicit instructions, rather than relying on patterns and inference instead.  These advanced technologies are a subsets of artificial intelligence (AI) and are becoming an increasingly important aspect of digital advertising.  Machine learning algorithms use large data sets to train predictions and decisions without specifically programmed instructions, thus advertising companies with large sets of data are positioned to apply machine learning algorithms to their advertising products. We can use machine learning in multiple different digital channels.

Google is at the forefront of Machine Learning

Google’s Deepmind program AlphaGo made history when it successfully beat a world champion at the profoundly complex game of Go.  Go is known as the most challenging classical game for artificial intelligence because of its complexity and large number of possible positions which exceed 10 to the 172nd power  (that number is larger than googol which is 10 to the 100th power).

Google continues to advance artificial intelligence across several sectors including, advertising, image analysis, video analysis, computer systems, controls and robotics, neuroscience and safety.

How Google Ads is Currently Using Machine Learning

Responsive Search Ads. Google’s responsive search ads allow you to add up to 15 headlines and 4 descriptions. Google’s machine learning will automatically test different combinations of your headlines and descriptions to learn which combinations perform best.

Smart Bidding.  Google has a set of automated bid strategies that help you achieve specific goals for your business.  All of these automated bidding strategies use Google’s machine learning to adjust your bids for each auction in real time. “In bidding, machine learning algorithms train on data at a vast scale to help you make more accurate predictions across your account about how different bid amounts might impact conversions or conversion value. These algorithms factor in a wider range of parameters that impact performance than a single person or team could compute.”

“With auction-time bidding, you can factor in a wide range of signals into your bid optimizations. Signals are identifiable attributes about a person or their context at the time of a particular auction. This includes attributes like device and location, which are available as manual bid adjustments, plus additional signals and signal combinations exclusive to Smart Bidding.”

A list of signals Google’s Machine learning uses to adjust bids:

  1. Device
  2. Physical Location
  3. Location Intent
  4. Weekday & Time of Day
  5. Remarketing List
  6. Ad Characteristics
  7. Interface Language
  8. Browser
  9. Operating System
  10. Demographics
  11. Search Query
  12. Search Network Partner
  13. Web Placement
  14. Site Behavior
  15. Product Attributes (Shopping)
  16. Mobile App Rating
  17. Price Competitiveness (Shopping)
  18. Seasonality

How to Use Google’s Current Machine Learning for Search Engine Optimization

To apply Google’s machine learning to your SEO efforts, you will need to run Google Ad products that currently use machine learning like responsive search ads and smart bidding.  This will allow you to collect data that has been optimized by machine learning technologies and apply that data to your SEO efforts.

Integrate SEO Language Proven by Machine Learning. 

  1. Create a search campaign that uses responsive search ads.
  2. Maximize responsive search ads by adding 15 headlines and 4 descriptions to 3 separate ads for a total of 45 headlines & 16 descriptions
  3. Over time, Google will apply its machine learning to determine which headlines perform the best.
  4. Use the Ads & Extensions Asset Details report to see a list of all of your headlines.  This asset detail report will score your headlines and descriptions in the performance column and display the number of impressions in the impression column.  Note: Responsive search ads usually need around 5000 impressions over 30 days to display a rating in the performance column, however, the impression column is the primary indicator to how Google’s machine learning determines the best headline & description language.
  5. Integrate the best performing headline language proven through machine learning into key SEO elements on your website.

Machine Learning Responsive Search Ads

The key SEO elements where proven machine learning language can be integrated and optimized for SEO are:

  1. Title Tags
  2. Meta Descriptions
  3. Headers
  4. Paragraph Headers
  5. Text Content
  6. Blog Content & Subjects

Integrate SEO Keywords Proven by Machine Learning

  1. Create a search campaign that uses smart bidding strategies that meet your business’ primary objectives.
  2. Create an Ad Group with broad keywords or modified broad keywords. This is an important aspect of the process.  The broader your ad group keywords are, the wider variation of keyword phrases Google’s machine learning will sift through and assign bids to.   This gives you a much larger set of keywords to work with when performing SEO.  However, it’s always best to follow best practices and use negative keywords and other targeting options that narrow your ads for relevancy toward your advertising objectives.
  3. Over time, Google’s machine learning will assign bids to all of the keyword queries based on the probability of achieving your objectives outlined by the bid strategy type.
  4. Use the keywords search terms report to see a list of all the search terms and associated bids created by machine learning for each search term.
  5. Use the columns to determine how you want to sort and value your keywords.  Note:  The importance of the data in the columns may vary based on the bid strategy type you used.  For example, if you used a conversion focused smart bidding strategy like Target CPA, Maximize Conversions, Maximize Conversion Value or Target ROAS, the Average CPC, Conversions and Conversion Rate, columns will be important to determining the most valuable keywords according to Google’s machine learning.  On the other hand, if your smart bidding strategy was Maximize Clicks or Target Impression Share, you might want to focus on the value of non-conversion-focused column like clicks, impression, CTR and more.
  6. Integrate the most valuable keywords proven through machine learning into your overall SEO strategy.

Keywords determined to be the most valuable through machine learning can be integrated and optimized into various aspects of your overall SEO strategy including:

  1. Keyword Research – look for relevant variations of your most valuable keywords proven by the machine learning process.
  2. Content Strategy – integrate the most valuable keywords into your content strategy.
  3. On-Page Content Integration
    1. Title Tags
    2. Meta Descriptions
    3. Headers
    4. Paragraph Headers
    5. Text Content
    6. Blog Content & Subjects
  4. Image Data
  5. Video Data
  6. Link Building

Unfortunately, there no SaaS or plugins to enhance your SEO with machine learning technology.  Leveraging machine learning for SEO requires you to use current technologies that have already implemented machine learning.  Companies that already use Google Ads will find tremendous benefits to enhancing their SEO keyword research, understanding and integrating ideas by leveraging the machine learning data out of Google Search Ads.  It may take several months to get a good set of SEO keywords and SEO language that has been proven through the machine learning process for your SEO efforts, but we believe that data accumulated using machine learning campaign settings in Google Ads is invaluable to your SEO efforts.  Give it a try!