Is machine learning going to take over SEO professionals?
by Jabir Tomás del Río & Manuel Anduquia
During the past few years, search engines have been using machine learning(ML) algorithms to improve their understanding of content and the results they deliver to users. And we mean search engines as plural because, with the integration of search boxes in different apps and software, engines like Bing, Yahoo or DuckDuckGo are gaining popularity. So you’ll need to optimize your website for each one of those.
Remember that opinions about machine learning and automation have been divided between SEO professionals. Some affirm that this advance in machine learning will progressively erase the SEO role, and those who, on the contrary, think that this is an opportunity to improve the quality of our job.
Well, in this article, we’ll explain why these technologies can make your job easier and improve the quality of your insights.
Why is machine learning an ally for SEO professionals?
To get to the point, we need a little history about Google. In 2015, the engine released a new system to better understand the user intent of a search query called RankBrain.
RankBrain is a system by which Google can better understand the likely user intent of a search query. At its core, is an update that took from a “strings” to “things” understanding, the algorithm started to read characters and their context instead of just seeing text strings in the code.
Start Ahead & Stay Ahead
In marketing and SEO, we are creating and analyzing data all the time to understand user behavior and trends and determine the best way of action. The amount of data used increases exponentially every week, month, and year. You’re already behind if you’re waiting to see what customers think and only then getting ready to respond.
That’s why Artificial Intelligence (AI) must be considered an ally of marketers, and specifically of SEO role. Today’s AI- and ML-powered tools enable real-time insights to personalize and optimize content at the moment for each user’s individual needs. According to a Brightedge study, 31% of the time, tools help to understand customers’ needs better, and 27% improve productivity.
Considering that platforms are using AI and ML technologies to empower their capabilities, it’s time to rethink the role of marketers. With these technologies, repetitive tasks like data analysis and predictions can be automated, letting professionals more time to generate quality insights of high value and impact on business.
ML in Search & Ranking: can you predict organic traffic?
Typically one of the main objectives of an SEO role is to define a strategy based on relevant keywords that help generate and position quality content. This is usually done traditionally and meticulously using tools that allow us to see how to do keyword research with approximate search volumes, competition/difficulty, and CPC, among others.
It is essential to mention that the ecosystem of disruptive technologies integrated with Business Intelligence and Big Data adds more value to the predictive capacity of ML in ranking positions, as data lakes can provide trends and particular behaviors of keywords and search terms. This data can be analyzed for better decision-making in generating quality and valuable content that helps organic positioning.
But, what if, through AI and ML, you could improve, automate or even predict the ranking position of a particular keyword over time? Let’s see what ML offers us today related to the estimation or prediction of keyword ranking positions.
There are probably many methods for predicting organic traffic, but let’s focus on two in particular:
1. The Holt-Winters Method
Holt-Winters is a model that is based on the behavior of time series. Forecasting data usually requires a model, and Holt-Winters is a way of modeling three aspects of time series: a typical value (mean), a slope (trend) over time, and a cyclical repeating pattern (seasonality).
The R language and R Studio are among the most popular tools to handle ML with a statistical approach. That said, you can use R along with the Google Analytics view ID and date range to get the data of the organic sessions you want to know their prediction.
2. The CTR Method Using Search Console
The CTR method has a shorter analysis approach in projecting keywords in the next 12 months. However, it has the advantage of being able to target specific URLs based on additional, customized criteria.
We will need crawling software such as OnCrawl, SEMrush, or Search Console that can connect to other data sources and any other tool that provides keyword data. With this, we will be able to create positive and negative projections based on the pages whose CTR is higher or lower than the average CTR of the whole site.
Conclusions
These are just a few methods of how AI & ML-powered tools can improve repetitive tasks and data analysis in the marketing field. In terms of SEO, it’s interesting to think about how this technology could be combined with already existing tools to automate website implementations of features like titles and descriptions based on validated keywords.
Or, for example, since SEO analysis of keywords is a data-based task, why not program and train an ML algorithm capable of predicting the keywords' values for our business?
The usage of ML to predict possible scenarios, such as impressions or traffic impact, and content planning, would facilitate better decision-making through data
All this data processing and task automation would be a powerful tool to leverage the value of SEO inputs and their outcomes. Still, it is essential to note that data without context could also drive wrong decisions. At this point, a specialist perspective is critical to unleashing the power of these technologies.
