SEO and GenAI
I wrote this article while I was the teaching assistant a core marketing class at USC Marshall’s MBA program. Generative AI has been a focus of this class. Classroom exercises and assignments introduce students to a variety of marketing use cases and tools. Material such as this is meant to make students aware of how industry dynamics and capabilities could change when they enter the workforce.
Table of Contents
Generative Search. What is it?
“AI-powered snapshot of key information to consider, with links to dig deeper”. Thus, as a marketer who wants to be discovered, you want the generative search model to mention your product/service either or provide your property as a reference.
How SEO will change
The power is shifting even more into the hands of big tech companies through which we access the internet. Five years ago, almost half of all google searches were zero click searches i.e. the search ended on the search page itself, without users clicking into any links. Users could find answers to their questions in the meta descriptions or in rich results. That number has likely crossed 50% a while ago with text summarization maturing even before ChatGPT took the world by storm. Bad news for digital display advertising. But for search engine marketing, the need still remains, just with different rules.
Most search results do not need you to leave the search engine!
Getting traditional SEO right involves optimizing the content you want to be discovered for as well as making sure this content is discoverable by crawlers of search engines. These are referred to as content and technical SEO. This article focuses on Content SEO.
To provide ads you need ‘property’, whether that’s a billboard by the freeway or a sidebar on a website. The value of such a property is proportional to the eyeballs that it receives. In digital properties today, consumers scroll by and click on multiple links in an attempt to find personalized or/and comprehensive answers. If AI can meaningfully summarize multiple sources for you, there will be a drastic reduction in the visibility these spaces receive and thus their value.
Generative AI has the potential to streamline the consumer journey by enabling direct conversions from the search engine results page. Instead of navigating through a website, users can complete purchases right from the search engine. This might lead to fewer website impressions, but if executed effectively, it can transform initial consumer engagement into a complete sales funnel process in a single step.
However an SEO change is coming, and it’s a massive one. A study by Insight Partners found that only 57% of links cited by generative search are from the first page of organic search results.
“A task force at the Atlantic modeled what could happen if Google integrated AI into search. It found that 75% of the time, the AI-powered search would likely provide a full answer to a user’s query and the Atlantic’s site would miss out on traffic it otherwise would have gotten.”
~ Wall Street Journal
Through an analysis of Google’s classic search guidelines, emerging trends in generative AI and expert interviews, I have identified six developments that are likely to shape the future of generative AI assisted SEO .
Conversational Search
As AI becomes more integrated into search engines, the way users search will change. Queries will be longer and more specific, often initiated through voice. SEO strategies must adapt to these nuances by optimizing content for more conversational, natural language and incorporate long-tail keywords that mimic how people naturally speak.
Most voice searches are phrased as questions. Content can optimize for this by including who, what, when, where, why, and how in your content and developing FAQ sections that directly answer common questions.
AI-driven search engines are better at understanding the intent behind a query, not just the literal keywords. This shift towards semantic search means that SEO strategies need to focus on topics and concepts related to keywords, rather than just the keywords themselves.
Added Emphasis on Quality and Originality
No, the solution to creating more content for GenAI is not to create it using GenAI! The research and industry to be developed around attribution for GenAI is large - academic integrity, journalism, intellectual and legal rights, social media and of course search engines.
With AI able to generate large amounts of content quickly, search engines might prioritize content that is original and provides real value to users. This could mean a greater focus on unique insights, deep analysis, and creative approaches that AI cannot easily replicate.
Search engine optimization already has the bad rap of having created content that is built first for the search engine and second for the user. Buried deep in the google search documentation are the E-E-A-T trust guidelines. It stands for Experience, Expertise, Authoritativeness, and Trust and determines the page quality ranking. EEAT is is currently a form of reinforcement learning from human feedback (RLHF) and is likely to be an inspiration for GenAI content quality guidelines too.
| To what extent… | |
|---|---|
| Experience | …does the content creator have the necessary experience for the topic? |
| Expertise | …does the content creator have the necessary knowledge or skill for the topic? Different topics require different levels and types of expertise to be trustworthy |
| Authoritativeness | …is the content creator or the website is known as a go-to source for the topic? |
| Trust | …is the page is accurate, honest, safe, and reliable? |
Enhanced Personalization:
AI-driven search engines could offer more personalized search results based on user behavior, preferences, and history. SEO strategies might need to focus more on catering to specific audience segments rather than a general audience. Consider the amount of information that google has about you across your search, browsing, watch (YouTube), location (GMaps) and communication (GMail). Using this information to serve personalized ads had a lot of value, but imagine if those could be used for prompt engineering to generate the perfect answer for you.
A snapshot of what Google knows about me. All accurate!
Due to personalized search results, there’s a need for more diverse content that can cater to different user preferences and intents. This means creating a variety of content types (articles, videos, infographics) and covering a range of topics within your niche to appeal to different segments of your audience.
Understanding Intent
As a marketer you’d need to understand customer journey to understand what users ask before they arrive at the transactional stage, or what else does a user ask who might use your product/service. For instance Amtrak might want to purchase visibility for “Los Angeles - San Diego flights” searches by recognizing that non-Californians probably would default to a flight and not realize that they could consider a train journey that takes the same time and costs lesser.
This idea is not new, but its value will increase in the future for the reasons mentioned earlier - voice based and long-tailed queries since seach engines will be capable of better interpreting them.
Multi-modality of Sources
Multi modality refers to the ability of models to blends different types of data—such as images, text, video, and speech. Google’s upcoming Gemini Ultra model is leading the charge with demonstrations (of concerning authenticity) of its ability to infer images and videos with a very high level of reasoning.
Hands-on with Gemini: Interacting with multimodal AI
Thus, when AI models can search for information across internet videos, podcasts and more, marketers will not be vying only for discovery on each platform as traditional multi-channel marketing processes, but across content formats. The tactics of this will play out as industry partnerships evolve. Claude + Apple Music? Gemini + YouTube, duh!
Rebalancing the Marketing Funnel
Traditional Steps in the Marketing Funnel
Broadly, user queries can be categorized in four ways - Informational, Navigational, Investigational, and Transactional. Looking at them from a marketing funnel viewpoint:
- Top of the funnel - SEO to answer informational and navigational queries are usually for brand awareness. User might be looking for a specific webpage or source of information.
- Bottom of the funnel - Investigational and transactional queries where the user is considering and comparing options for a purchase decision.
Expect more transactional questions than informational compared to the past. As the search bar becomes capable of understanding long questions instead of matching web pages by relevance, the marketing funnel should become bottom heavy.
| More searches like | Rather than |
|---|---|
| What car should I drive given that I have a family of three that might grow and I am price conscious? | What to look for in a new car? Family car necessary features? Cars with best mileage? |
References
- How Google is improving Search with Generative AI (blog.google)
- “It is probable that targeting capabilities would increase as we will use not only the specific search term and keywords but an entire conversation as a context.” - Google’s Search Generative Experience: Shaking Up the SEO Game (cmswire.com)
- 3 new things you can do with generative AI in Search (blog.google)
- How to Future-Proof Your Marketing for an AI-Centric Search World?
- How Much of Google’s Search Traffic is Left for Anyone But Themselves? - SparkToro
- Search Quality Rater Guidelines: An Overview (google.com)
- searchqualityevaluatorguidelines.pdf (googleusercontent.com)
- News Publishers See Google’s AI Search Tool as a Traffic-Destroying Nightmare - WSJ
- Web Vitals
Published: 21 December, 2023