The Impact of AI and Machine Learning on the Future of SEO
An important point to address is that while AI changes the way in which we optimize our websites, there will always need to be human beings to ensure that the machines are learning the right things and not engaging in some unsavory web spam.
New findings suggest that machine learning is already being used by at least 1 in 8 search engine results, with this number predicted to double by the year 2020. This is still the early stages of what is likely to be a fast and far-reaching trend. SEO professionals and digital marketers have a good opportunity at this point to keep ahead or at least keep pace with this trend. This could involve further analyzing and interpreting machine learning algorithms to gain an understanding of how websites and web content will be assessed and indexed. A more ambitious approach could be to learn how to deploy machine learning models in some sort of search marketing capacity. This would require learning statistics and machine learning programming or collaboration with those who are familiar with this field.
Artificial intelligence (AI) and machine learning are altering the existing concept of traditional SEO. New resources and discoveries have revealed that the original algorithms that search engines use can now learn and optimize. This is looking to create a new era of search marketing, one where machines will learn how to develop and deliver search results to a human in a way that is similar to the way in which we would have a conversation. This has many implications for the future of SEO.
Overview of AI and Machine Learning
AI and machine learning have become widespread, affecting our everyday lives, be it through virtual personal assistants such as Siri or Alexa, or through the development of industries and automation. The ability to store and interpret large quantities of data through AI has expanded the development of machine learning and has greatly increased the use of machine learning technologies. With AI and machine learning becoming more prevalent, it is important to consider the impact they will have on the world of SEO and its implications.
In the past decade, the use of artificial intelligence has grown and was projected to reach USD 190 billion by 2025. AI technology uses computers or machines to mimic human thought processes such as learning or problem solving, and it is often used to make learning, reasoning, and self-correction possible. Machine learning is a core sub-area of artificial intelligence; it enables computers to enter a mode of self-learning without being explicitly programmed. When exposed to new data, these computer programs are enabled to learn, grow, change, and develop on their own. The term “machine learning” was coined in 1959 by Arthur Samuel.
Importance of SEO in the Digital Landscape
At present, search engine algorithms use specific rules to determine relevance that are subject to change with a new update. An example of such a rule could be the importance of keywords in determining content relevance. If a new update causes changes to such rules, it can have a significant impact on current search results. AI, however, aims to make search algorithms function more like a human, understanding language and intent and learning over time what is most relevant to each unique user. This shift from rules-based algorithms to machine learning will mean less drastic changes to the way that SEO is done, as the learning algorithms will make frequent small changes rather than have occasional large updates. On the downside, this also means that it will be more difficult to discern why a certain site ranks well or badly. Overall, AI and machine learning aim to better understand users and content in order to match them more accurately. This can be seen as synonymous with the aims of SEO to understand and deliver what users are looking for.
SEO stands for ‘search engine optimization’, the process of maximizing the number of visitors to a website by ensuring that the site appears high on the list of results returned by a search engine. As search engines become smarter, the fundamentals of SEO are unlikely to change. It will always be important to have a good, fast, mobile-friendly site with great content and good backlinks. However, it is the way that search engines find and interpret this data that is going to change. Currently, search engines use crawling bots to collect and index data. This data is then retrieved and algorithms are used to determine which is the most relevant content for your search. It is at this last stage that AI is going to have the biggest impact on SEO.
The current digital age is an era brimming with data, and the importance of SEO in such a landscape is immense. The digital landscape is built on data, and it is a key element in directing and connecting prospects to the brands that offer the solutions they are looking for. With the rise of AI and machine learning, this importance is only going to increase as these technologies progress. Search engines are still the primary way for people to find websites, and so it’s essential to understand the changes that AI and machine learning will bring to SEO so that we can make the most of the new opportunities presented, while not losing sight of the importance of the fundamentals in this area.
How AI and Machine Learning are Transforming SEO
At first sight, this would seem to be bad news for SEO. In the AI world, if you can’t determine a causal relationship between your actions and the desired outcome, it’s very difficult to formulate a strategy or plan. There are no certainties to cling to; just probabilities. But all is not lost. In a world where there are no hard and fast rules for determining search rank, what reigns supreme is improving the probabilities. If we can’t reverse-engineer the search algorithms, we can at least try to influence them. The content and links that best match the user’s query and provide the most complete information will have the highest probability of success. It’s still an uphill battle, but in the grand scheme of things a more legitimate way to win quality search traffic.
With the rise of AI and machine learning, search engine algorithms have become more complex and less deterministic. This enables search engines to return results that are better attuned to what the user is really looking for. This is bad news for SEO service, as it becomes impossible to reverse-engineer a particular search ranking. Gone are the glorious days of building an awesome piece of content and seeing it race to the top of the search results. Even analyzing the search results to infer the needs of the user is less effective, because it is no longer possible to reliably make inferences about why the search engine has returned a particular result.
Enhanced Search Algorithms
As it stands now, humans are creating search algorithms, and it is a very expensive and arduous task to organize various forms of data, identify what is relevant to a query, plan how to retrieve it, and make sure that it does in fact retrieve the correct data. An AI can do this semi-autonomously by learning to rank. The computer or machine would learn by examples or experiences to carry out a task, without being explicitly programmed to do so. In the case of search algorithms, the AI would learn which results are best by looking at examples of human satisfaction with a particular result, how long they clicked around, if it was the answer they were searching for, etc., and then use these experiences to build a methodology as to how to order results for a query. This is actually present learning to rank exercises taking place with the use of supervised machine learning, in particular with support vector machines in various academic and research institutions. However, it is important to consider that using such a method can be costly and there is a point where the upkeep of learning is not worth the gained improvement in search algorithm.
A highly significant impact that AI, as well as machine learning, has in regards to the landscape of search engine optimization revolves around the effort to enhance search algorithms. In essence, taking into consideration the massive extent of the internet and the perpetual and constant increase in information and data, traditional SEO that was based around a specific keyword matching a search query, or creating content that will rank for certain keywords, is gradually becoming obsolete. Users, and also targeted demographics, are located in an international aspect, and the internet is a smorgasbord of information with no centralized control. Because of this, search users can be looking for infinite things, and to improve the search experience, engines need to understand the full context of the query as well as all possible topics for a search phrase.
Personalized Search Results
In summary, search is ever-evolving, and personalized search results are an important part of that evolution. In the next 5-10 years, search results will be tailored even more to the individual user. This is very much in line with the trend of AI and machine learning, as these techniques provide the best results with the least possible effort from the user. SEO will have to adapt to these new search environments. It can be expected that traditional ranking factors will become less important, and new methods will need to be devised to get content in front of the right people. How exactly one would go about doing SEO for machine-specific content is unclear, but it is definitely an area that will need more research in the coming years.
On the surface, this sounds great for the user, as they are more likely to find what they are looking for. However, there are implications for companies and SEO. It may be more difficult to target specific demographics using SEO, as different users will see different results. The searchable assets of a company may have to be more diverse and plentiful. A positive is that users are more likely to come across content that they will find useful and engaging.
As machine learning algorithms grow in complexity and efficacy, the search results will better reflect the needs and desires of the user. This applies to all types of searches, whether informational or transactional. Users will see more of what they are interested in and less of what they are not.
At first, search personalization was related to geographic location. Users would get different results based on where they are in the world. This is still an important factor today with mobile computing. In the past decade, personalization has evolved to be much more complex. With continuous advancements in machine learning techniques, the engines are now able to take into account the user’s past search and browsing history, their demographics, social network, and the devices that they are using.
The concept of personalized search results has been around for nearly as long as search engines. Search engines aim to provide the best results for their users, and personalization is an important part of delivering those results. Search engines strive to give the best results for the users’ needs. In order to do that, the engines needed to know more about the users and their habits, which presented a challenge for privacy and user expectations.
Voice Search Optimization
Voice search is having and will surely have an even greater impact on SEO in the near future. Now is definitely the time to start strategizing voice search optimization into your digital marketing campaign.
For best optimization, it might be useful to create an FAQ page to increase the chances of catching a voice search with a specific question. Frequently asked question pages do well in search, and a single FAQ page incorporating all questions and brief answers on a given topic is more likely to be selected by a search algorithm as a direct result of a voice search inquiry.
Marketers and SEO experts will have to consider that a person using voice search is likely to ask a complete question instead of entering a few chosen keywords. The “Who, What, When, Where, Why, and How” questions will be a great place to start in brainstorming how the page would best answer a specific voice search inquiry. This is similar to the shift from short-tail keywords to long-tail keywords, but made even more specific.
Voice search optimization involves adjusting keyword data to reflect the likely spoken phrase someone will use. It entails thinking about the different queries and sentence structures a person might use when verbalizing a search. Ultimately, keywords will need to be more conversational as people tend to pose questions using natural language when using voice search.
Challenges and Opportunities for SEO Professionals
The increasing prevalence of machine learning will also mean big changes in the way that search algorithms are evaluated and tested. In the past, a search engineer might have taken a list of factors or a small set of manually checked results and made changes to the algorithm based on that. It was still a very human process. Now and into the future, any changes to an algorithm or its features will be tested against a control in an offline simulation to try to predict how much of a difference it will make. If the change is an improvement, the new version of the algorithm will run live. If it’s a degradation, it will be scrapped. This might mean that changes which would have previously been A/B tested and then rolled back if unsuccessful are now just deployed in the live system with the expectation that the simulation was a good predictor. An SEO monitoring the changes would see even more rapid fluctuation in the search results for the affected queries and might have a hard time discerning if a change in their own site’s ranking was due to the new algorithm, a change to the competitiveness of the SERP, or just random noise.
A recent case study by Stone Temple Consulting that was designed to test the impact of the Knowledge Graph showed just how volatile search results can be. The study found that within a two-week period, 75% of the test queries saw a change in the number of organic results, and 47.5% lost at least one position in the results.
With the rapid development of AI-driven search algorithms, all SEO professionals will have to adapt to a landscape that is constantly changing. Instead of optimizing for a static set of factors or trying to reverse engineer the ranking system, SEOs will now have to rely on results from tests and correlations to reverse engineer a constantly evolving system. While this may seem like business as usual for some, it’s likely that the dramatic nature of these changes will result in a lot of churn in the organic search results. This means that positions will be lost more frequently, and gains will be slower to realize.
Adapting to Algorithm Updates
Measures to prevent severe drops due to unexpected algorithm changes include diversifying your traffic sources. If you are currently too dependent on Google organic traffic, you can try to get traffic from other sources such as PPC, social media, or offline marketing. Additionally, you can save your extra revenue as a safety net in case there is a significant decline in your traffic.
However, Google has never informed us about what those updates are, when they will be done, and how they will affect the search results. It could even be minor changes with major impacts. This has become a concern for SEO practitioners, who need to keep learning, monitoring, and testing on a repeated basis in order to understand the algorithm pattern and the impacts on their SEO work.
Algorithm changes and updates may be good or bad for you, depending on your expectations or readiness to cope with the changes. The good news is that these frequent tweaks and updates mean that Google is still concerned about the quality of the internet and its contents. For those who are doing clean and quality SEO work, there should not be too much worry.
Leveraging AI for Competitive Advantage
In a recent survey, nearly 80% of search advertisers expressed plans to increase usage of location targeting in the next 12 months. Marketers have long understood that motivation is often implicitly conveyed in consumer location. Machine learning can help identify complex patterns in search queries to automatically improve bids for better performance, but true location targeting requires an understanding of the location itself and a real-time connection to the user. For example, a person in a different mood or with different intent may search for the same thing at home and at work, yet marketers would want to serve this consumer with different ads. This could be facilitated by understanding the implicit context of the different locations and serving ads only when the location fits the desired context. Static location data and simple bid modifiers are insufficient to accomplish this. Advertisers should expect continued improvement of location context targeting technology, while also being aware of the implications of increased data usage on consumer privacy and regulatory restrictions.
Keyword-driven strategies are no longer sufficient to support the breadth of activity search marketers take in the current environment. Consumers continuously shift from one application or device to another across the journey. While it is important to use keywords that are consistent with consumer search queries, it is also necessary to align messaging and content strategies with the consumer’s state of mind during the search process. Increasingly, this will require the use of semantics and machine learning. In order to appeal to consumers in different states of mind, search marketers must natively adapt their content and messages to fit the unique motivations implicit in the varied search contexts.
Balancing Automation and Human Expertise
High-quality automated recommendations such as this are the result of current developments in machine learning, which enable software to adapt to specific business goals and use live or historic data to make decisions and predictions. This, in turn, can lead to less need for manual trial and error testing of different SEO strategies. Meticulous planning of experiments and proper data collection and analysis will enable pre- and post-tests of software-run SEO changes to evaluate their impact and ROI. Control and evaluation of an automation system is essential to ensuring that it supports rather than hinders the effect of human expertise.
The danger is that too much reliance on automation can result in search campaigns that are of a “one size fits all” approach dictated by software rather than the unique characteristics of a business. According to research carried out by Benjamin Elad and Itai Himelboim, businesses that rely on SEO seek to differentiate themselves in marketing their products and services through creating a unique and valuable online consumer experience. This may take the form of improving brand visibility, targeting a specific audience, or emphasizing product innovation. Businesses may decide that targeting a new audience is a better idea in order to be better differentiated from other businesses. However, the automation tool may recommend that they focus on improving rank for keywords that are more relevant to products offered to their current audience based on an analysis of keyword search volumes and the current state of on-site optimization.
By facilitating the automation of rote tasks and providing analysis and recommendations, AI and machine learning have already had a positive impact on SEO. Companies such as BrightEdge and Conductor are already providing SEO platforms that deliver enterprise-level analysis and recommendations that in the past could only have been done by a very experienced SEO. Their effectiveness can be seen by the increase in organic search to clients’ websites. Indeed, Bruce Clay attributes a fall in costs per client for SEO since the 2008 recession to the greater availability and effectiveness of SEO automation tools.
The Future of SEO in the Age of AI and Machine Learning
Those with an understanding of SEO and digital marketing will know that keywords are a vital tool to increase visibility for websites and increase revenue. Machine learning has enabled a shift in focus from individual keywords to topics. This is due to Google’s algorithm update, BERT, which will prioritize understanding the context of a word within a sentence. This will mean sites with content that truly matches a user’s search will rank better, which is a positive change. However, this could potentially mean lesser visibility for sites with content that is not fully relevant to their desired audience. A recent development in NLP is its capabilities to generate written content. An example is OpenAI’s GPT 3.0. This is an area of machine learning that may be beneficial for a content-driven site looking to decrease costs. However, it poses the question of whether written content now needs to compete with machines.
Search engine optimization (SEO) is a familiar concept to those working in marketing or business operations. The purpose of SEO is to increase the quantity and quality of traffic to your website via organic search engine results, thereby increasing your site’s visibility to users. Presumably, a concept that solely relies on the visibility of websites on search engine results is vulnerable to being heavily impacted by an industry that embodies change, specifically the IT sector. It is indeed the case that AI and machine learning are already changing the landscape of SEO, and there is much more to come. Some of the changes will be positive for digital marketers and businesses, while others may not favor certain industries and will require a new way of thinking for the SEO process.
Predictive Analytics and SEO
On the micro level, it will become more practical to determine which single actions are most effective in moving towards a desired outcome. Should machine learning establish a way of comparing data across a wide range of SEO tactics, it could spell the end for sifting through large amounts of data to determine the best course of action. A process where the AI can simply state “based on the pattern of your search rankings, creating more content like X has a Y% probability of increasing to position Z over W weeks” will greatly simplify the jobs of SEOs.
Transitioning from AI to machine learning, the growing capability of computers to predict future outcomes will have a dramatic impact on SEO. In the past, Google’s Hummingbird update and the subsequent years of refinements placed a huge emphasis on understanding the intent behind search queries. Technologies such as predictive keyword analysis looked to forecast which keywords were most likely to lead to a conversion or the acquisition of a long-term customer. Machine learning goes a step further, enabling computers to learn from previous data inputs and use this to predict outcomes on new data. For SEO, this will take the form of predicting how well a piece of content will rank and being able to forecast outcomes of SEO campaigns.
Natural Language Processing and SEO
NLP has revolved around linguistic data (words and grammar), but as NLP has evolved, it has been more focused on statistical data from examples of natural language and then processing the language through machine learning techniques to extract meaning and have the machine respond with appropriate information. There are many SEO applications for NLP. Firstly, spoken language data is used to improve search engine potential for image and video search optimization. This is particularly good news for content producers who have a visual media focus. As an example, podcasts are now being turned into textual data that can be indexed, providing new search opportunities for content owners. The same goes for video media. YouTube is a major success story, but video search is still quite limited. Many content producers have found success in video optimization, especially with local search. NLP will further increase the potential of combined video/local search marketing. Newspapers and TV guides are set to benefit as well. Google revealed a new feature at its Searchology event on TV show search. This is said to eventually merge TV and the web for one interactive entertainment experience. This has huge marketing implications for TV shows and networks who want to link ratings to web traffic.
Natural Language Processing Natural Language Processing (NLP) is an area of computer science and AI concerned with the interactions between computers and natural human languages. In a recent interview I had with Steve Papa, founder of the NLP company Endeca and now founder of the parallel-processing NLP and machine-learning startup, Solariat, he defined it even more precisely. “It’s not about understanding every bit of human language, it’s about understanding enough of the language to respond in an appropriate way.” This area of science has been around since the 1950s, but has made enormous advances in the last few years due to the large amounts of digital text data being generated, the increase in computing power, and the emergence of improved algorithms. Since search engines primarily revolve around people searching for information through textual means, NLP has been an important concept in improving search engine capabilities. NLP involves several different processes that interact together: lexical analysis, syntactic analysis, semantic analysis, and discourse integration. Traditional search engines have generally done the first three processes, which involve identifying and classifying elements of the input text (i.e. words), to then organize those elements into structural components. This, in turn, allows the search engine to understand the meaning and draw inferences from the input data. Search engines are, in fact, now capable of producing search results where meaning is taken into account and not just keyword matches.
Ethical Considerations in AI-Driven SEO
As such, it would be a safer bet to assume that AI development will continue and that at some point, there will be a point of no-return where AI technologies are too widespread and advanced to not use in the SEO industry. At this stage, developers and proponents of AI technology for SEO may point to the successes of technologies in other industries and argue that AI is a natural progression and have potential to greatly increase the efficiency of SEO practices.
While it would be ideal to predict and control the progress of AI technology, history has generally shown us that technological progress cannot be restrained and that once started, technological development is difficult to stop. Even if it were possible for the current SEO community to agree on the direction for development of AI for SEO, it is almost certain that outsiders (for example, those in other advertising industries who stand to lose out in the event of AI success in SEO) would still continue AI development and use behind closed doors.
The implementation of AI and machine learning technologies in SEO will inevitably provide an increase in functionality and efficiency of current practices. Given the power and inevitability of progress in these technologies, it is essential to consider the ramifications AI development will have in the evolution of SEO.
