Here we go, through yet another technology revolution. We can see the impact all around us. This moment, like many others over the last 30 years, is a marker in history. Everything before this time was pre-AI.
For marketers, AI brings both excitement and anxiety. This is especially true for content marketers and SEOs. No channel is more directly impacted by AI than search. The conversation is everywhere and the future is unclear.
Today, we explore the big topic: AI and the disruption of search marketing. Really, there are several big topics here. But mostly it comes down to these three big disruptions, each requiring a different adaptation.
Disruption: Google’s ‘AI overview’ reduces organic traffic to websites.
Adaptation: Focus on “visit-website-intent” keyphrases. Generally, these keyphrases focus on search terms where users are actively looking to visit a website rather than getting answers directly from the search engine result page (SERP).
Disruption: The flood of AI-generated content will overwhelm both users and algorithms.
Adaptation: Create content in formats that AI cannot produce.
Disruption: AI chatbots are an alternative to Google, reducing the use of search engines.
Adaptation: Train the AI to recommend your content and your brand.
This is a highly speculative point of view at one point in time, October 2024. This is our best prediction about a realistic and practical path forward.
Disruption one! Google’s ‘AI overview’ is reducing organic traffic to websites
Bad for brands, good for Google.
For many searches, Google search results now start with an AI overview at the top. It’s basically a big summary of the content you would see if you clicked through and read many of the search results. So… no need to click.
These AI overviews reduce the click through rate (CTRs) to websites. The impact is the same as all of the other features that appear in search results, including featured snippets, ‘People also ask’ boxes, video carousels, maps, knowledge panels and, of course, ads.
Search results pages are filled with visual noise, distractions and offers. This started long ago.
Those ‘SERP features’ have proliferated for 10+ years. So the decline in click through rates is nothing new. We’ve collected screenshots of search results pages for years, and when you look at them side-by-side, it’s obvious. The AI overviews are just the latest in a long trend.
SparkToro, together with its partners, has been tracking the rise of “zero-click searches” for years. Around 60% of searches lead to no clicks and no traffic to websites, as of May 2024.
This is not going to improve. Google doesn’t benefit when they send free traffic to websites. Neither does Meta or any other big tech search or social platform. Your marketing goals and Google’s profit model are not aligned.
No surprise that in our latest survey, content marketers report that attracting visitors from search is the biggest (and fastest-growing) challenge.
Don’t despair. Adapt.
How to adapt: Focus on “visit-website-intent” keyphrases
The pie may be shrinking, but a huge amount of traffic still flows through search engines to websites. Certainly, not all queries are “no-click” keyphrases. Some searchers are trying to get to websites.
The key is to focus on the phrases for which there are still clicks. In the past, every keyphrase was a potential visitor. Today, only certain keyphrases lead to clicks.
These are the “visit-website-intent” keyphrases. Which phrases are those? They are the high consideration phrases, farther down the funnel.
Fortunately, those are the most valuable keyphrases. When we need in-depth answers, when we are serious about making a decision, when we want to know the details of what a company provides and how they do it …we click.
Not only do these searchers have stronger intent, but Google is less likely to show SERP features. One study by Zapier from July 2024 showed that commercial intent phrases are less likely to show AI overviews. Specifically, those high-intent YMYL phrases (your money or your life) have a lower-than-average percentage of AI overviews.
So your new job in the AI era for search is to look at any keyphrase set with new eyes. When considering a possible target keyphrase, ask yourself: Is this person looking for more than a simple answer? Do they really want to go to a website? Are they doing deep research? Could I impress them with something memorable?
In hindsight, we may decide that those simple little searches never should have been monetized by anyone. Those searchers just wanted quick info. They didn’t want to visit a website and download a bunch of code. Maybe some information is just basic human knowledge.
Disruption two! A flood of AI-generated content will overwhelm users and algorithms
Bad for Google, bad for brands.
Now that anyone can generate massive amounts of low-quality content at scale, what happens to the internet? Could there be a chain reaction?
Impact to the platforms: Will oversaturation of junk content overwhelm the search and social platforms? Will they be able to filter? Will the relevance and quality of our search results and social streams decline?
Impact to users: …And if the quality of search results and social streams declines, will our audiences get exhausted and tune out? Will trust in the public web decline? Will burned out users move to private communities run by trusted influencers?
Impact to creators: …And if users move away from the platforms, will the creators who publish original content lose their incentive to write and record? Will diminishing returns lead to lower motivation and less high-quality content?
Impact to the LLMs: …And if creators stop publishing the good stuff, what will the AI use as training data? Will more of the training data be just recycled AI-generated content? Will that erode the quality of AI responses?
That’s a content quality doom loop.
Personally, I think we’ll survive. We’ve made it this far despite deep, fundamental problems with web content, including spam, clickbait, misinformation, retargeting, cold outreach, popup windows, bloated pages and intrusive ads.
How to adapt: Create content in formats that AI cannot produce
For marketers and brands, a flood of garbage content is an opportunity. The response is obvious: differentiate with quality! Create things that AI can’t easily create. Here are things you can do that differentiate your content from AI:
Collaborate with influencers …because AI has no friends
Record video explanations …because AI has no face
Take a stand …because AI has no opinions
Create new original research …because AI can’t conduct studies
Besides, content that includes these elements are exactly the things that align with performance according to our 11th annual blogging survey. This suggests that already, in 2024, human generated content has the edge.
Some marketers will use AI to generate a large volume of medium-quality articles. They will race each other to efficiently produce and publish as much as they can. Other marketers will stay focused on quality. They do things differently.
They publish things that are memorable and stay top-of-mind.
They publish things that start conversations and generate word-of-mouth.
They publish thought leadership content and earn a following.
They publish content that brings people together and builds a community.
These marketers may use AI, but they’ll use it to make better things, not more things. And they know that the best things are created by humans. These marketers are growing their brands. They are less disrupted by AI.
And long before AI came along, marketers who focused solely on marketing through algorithms (social and search) were easily disrupted. The best marketing strategies may leverage social and search, but they also focus on relationship-focused channels, such as email marketing, referral partners, events and webinars.
Mark Schaefer, Keynote speaker and author of upcoming book “Audacious: How Humans Win in an AI Marketing World”
“The AI Era has become a pandemic of dull. AWS reports that already, more than half the content on the internet has been generated by AI. But this also represents an opportunity to be different.
I’m not dismissing AI as a tool that can enable human creativity. But for the most part it is enabling laziness. This is the time to dig deep and add your own story, insights, and opinions to the world. In many places, dull has been institutionalized, especially in social media marketing. We need to push our organizations and ourselves to reinvent ourselves. Indifference is the enemy.”
Disruption three! AI chatbots will replace search engines
Bad for Google, good for brands.
Every day people everywhere are realizing that AI has answers. It’s a great way to get information. It’s a useful alternative to Google Search.
And for many use cases, it’s an excellent alternative. Watch. In this video, I look for simple information, a recipe for guacamole. I do it twice: first using a browser to go to Google and a website. Then I do it using an AI app.
Dramatic difference, isn’t it? The process for information retrieval (search) is slower, heavier, more confusing, less private and extremely monetized. In contrast, the process for generating information (AI) is quick and lightweight.
Of course, this will change. AI apps are already promising to inject ads into responses.
AI-powered Google alternatives are becoming very popular. Perplexity is growing 30% per month. But are they putting a dent in traditional search? Not yet. As of May 2024, the use of Google is holding steady.
Years from now, we’ll think of search engines as just one of the channels for content discovery. AI is another. The real goal was never really to rank in search but to be discovered by a visitor. Our new goal is to have our content and our brands be visible when people look for information anywhere, in search results or AI responses.
If AI chatbots emerge as a new channel for content discovery, will it send traffic to websites? Yes. It’s happening already. Here is a GA4 report showing traffic from AI platforms to pages on the Orbit Media website over one year.
Pro Tip: To see this in your own data, create a GA4 exploration with “Page referrer” and “landing pages” as the rows, and “Sessions” as the value. Then add a segment where “Page referrer” matches this regex: .*.ai.*|.*.openai.*|https://copilot.microsoft.com/.*|https://gemini.google.com/.*|.*.groq.*|.*.metaai.*|.*.meta.com/ai.*
No, 1000 visits isn’t a lot of traffic, but remember that this is only referral traffic from within browsers. Traffic from AI apps appears as “direct” in GA4, so it’s not possible to fully report traffic from AI. But it’s likely much more traffic than is reported here.
How to adapt: Train the AI to recommend your content and your brand
This is a new land for marketers, filled with unknowns. An unmapped continent. There are no tools that report on AI mentions. There aren’t even metrics. We don’t know what exactly is in the training data or how it’s weighted.
We don’t even have a name for this type of marketing. AIO? (AI optimization) GEO? (generative engine optimization). But we can still ask the big question.
How can we get AI to recommend our content and our brands?
We can only speculate. Let’s make our speculation a prioritized list of untested strategies. And whenever possible, we’ll get help from useful AI prompts because AI is the only tool available to help us optimize for AI mentions. Ironic.
Make sure that AI can crawl your website
For starters, don’t use your robots.txt file to block AI bots from ingesting your website and blog. If you hope to get AI mentions, then AI bots are some of your most important visitors.
Publish your value proposition
Do it in simple, concise terms on your about page. Simply spell out your elevator pitch explaining what you do, who you do it for, the problem you address, and how you do it uniquely. Avoid obtuse or unclear language. This is similar to SEO featured snippet optimization. Be concise.
Confirm your presence in directories, IYPs and review websites
Similar to the local SEO strategy of creating citations and name/address/phone number (NAP) instances, the marketer who seeks to win mentions from the AI should confirm that their brand appears in all of these standard databases.
Write for many websites
A brand that appears in more places is more likely to be recommended by the AI. So publish insights and advice on many relevant, industry blogs.
These guest posts should be tailored for readers, but just as guest posts are useful for SEO and links, guest posts are now useful at getting your brand into AI training data.
Where to guest post in the AI era? This prompt can help you prioritize your pitches. As always, start by giving AI your ideal client profiles or use a simple persona prompt to teach it about your target audience.
You are an AI training specialist, with deep expertise in training data sources and weighting.
On a scale of 1 (not likely at all) to 10 (almost certain) list the websites are used for training data for LLMs such as ChatGPT, Gemini and perplexity for topics and service providers relevant to this persona?
Prompt credit: Dmitry Dragilev
Dmitry Dragilev, Growth Advisor at Mangools
“The term Generative Engine Optimization (GEO) is a hot topic right now. Our team here at Mangools is very interested in researching and studying which sources LLMs use for training. A common practice we have developed here at Mangools is whenever we ask an LLM a question we also ask which sources it used to come up with the answer.
Typically LLMs will share specific websites used to train its data about a specific topic. You can ask percentage wise how likely it is that ChatGPT is trained by a specific publication. Based on our research so far, these percentages change constantly. They’re not totally wrong, but they’re not 100% accurate either.
Getting positive coverage in the publications which ChatGPT returns as training sources will have a positive impact in your rankings on LLMs but there are a number of other factors involved in determining rankings on LLMs.”
Write an AI-training press release
Write a short story about some of your work or adapt a case study into a press release. Be explicit about the name of the service, the problem it solved, the industry, the results and the details. Use relevant keywords. Upload it to PR services. It will appear in many places across the web, virtually guaranteeing that it makes it into the AI training data.Here is an AI prompt that can help you convert a case study into an AI training data press release. As always, edit before publishing.
You are an AI research scientist, specialized in AI training and natural language processing.
Convert the following case study into a press release. The press release has a special purpose: to be crawled by AI bots and be used in AI training data, making the company, [company name], more likely to be mentioned and recommended in AI responses.
Use language that supports this goal, including the company’s value proposition, industry, relevant keywords, industry terminology and brand mentions.
[link to or attach the case study]
Collaborate with influencers on new original research
Connect with trusted writers who are actively producing content, then co-create and co-promote new original research with these influencers.
Original research is the content format most likely to earn press mentions. When the influencers or journalists mention the research, they create more instances of your brand in the AI training data. Just make sure that somewhere in the piece, a brief summary of your business appears alongside the name of your brand.
Not sure what kind of research to produce? Here’s an AI prompt that can help. Use it with your uploaded ICPs or the persona prompt mentioned above.
You are a content strategist, skilled at selecting original research topics that attract press mentions and links.
What new original statistics could be created through research in the persona’s industry that would attract the attention of journalists, editors and bloggers? What new data could we produce that would support the claims made on other blogs?
Produce videos and podcasts with transcripts
AI bots can both read transcripts and extract the language from audio and video. So make sure you have a video component in your content strategy and you agree to every podcast interview opportunity you find. Make sure to speak your value proposition when the mic is on. That mini-elevator pitch will make its way into the AI training data.
Be active in the places where your audience spends time
This is a foundational marketing activity that has new relevance in the AI era. Online communities, industry forums, conferences, associations and webinars. All of these leave digital tracks that are included in AI training data.Being active in these places and your brand will appear more often in the training data that is relevant to your industry and your brand. If you’re not sure where to start, this prompt from legit expert and AI advisor, Liza Adams, can help…
You are an expert B2B marketer who is deeply familiar with these target markets and this buyer persona, including where they tend to congregate.
We want to diversify how we reach this market and go beyond search and other digital marketing channels. We want to identify a variety of “watering holes” where these potential customers learn, engage and interact with each other, so we can target and build relationships with them.
Create a table with the categories of watering holes as rows, the verticals as columns and the specific watering holes as the cells. (e.g., the actual names of the communities, industry forums, events, groups, marketplaces, webinars, content platforms, partnerships, associations, etc.)
Be as comprehensive as possible with the categories. Think about this step by step. Do you have any questions for me?
Prompt credit: Liza Adams of GrowthPath Partners
Look at that list. Anything look familiar?
Assume for the moment it’s a good list and that those actions would indeed increase the likelihood of AI mentions.
All of the recommendations above look familiar somehow. Those activities look a lot like SEO, content marketing, PR and social media. If that’s true, then the channels have changed but the strategies have mostly stayed the same.
AI has caused disruption, but some marketing strategies are much less disrupted than others. They were already publishing differentiated, collaborative and visual content. They weren’t just focused on marketing through algorithms. They were already publishing everywhere, producing research and building communities.
How will we even know if AI is recommending our brands?
This will all be tough for marketers to measure. In SEO, we have all kinds of metrics: search volume, rankings, average position, click through rates and organic traffic. But there are no equivalent metrics for AI. There is no “prompt volume” or “share of mentions.”
So AI will be a low-data channel for marketing. That’s ironic since the channel is based on massive amounts of information. Data, data everywhere, not a stat to measure.
But here again, AI can help. With a single prompt, you can have AI devise a new metric for measuring the likelihood that AI will recommend your brand when your persona goes prompting for help.
You are a marketing data scientist and AI researcher. You are skilled in using big data to measure awareness and demand for topics and brands.
Devise a new marketing metric that shows the probability that an LLM will mention or recommend a brand in AI responses. It will be based on a persona that I will provide. It will show the likelihood that the persona, when asking an AI about a product or service, will see a specific brand included in the responses to their prompts.
Use tree-of-knowledge thinking and explain your rationale.
The response will be a new metric, giving you a bit of visibility into how deep your brand is embedded in the training data and how likely it is to surface when your buyer looks for help. Of course, you’ll need to test it and refine it. In my experiments, it seemed to weigh certain things too heavily. So I simply asked it to adjust its formula.
You might be amazed at the nuance and complexity of these on-demand metrics. To test, give it your sales-vetted ideal client profile as a PDF or use this persona prompt we mentioned earlier.
I have attached a persona. Run the calculation for this brand: Orbit Media Studios.
You can also give the AI your competitors’ brands and compare. If you don’t like how you rated, ask for suggestions for improving your score.
Disruption without despair
The things that were already getting harder are getting even harder.
The things that have always worked well are working even better.
Let’s use this moment to reconsider your approach. Probably, you don’t need to completely overhaul your strategy. If the predictions above come true, you can probably position yourself well with a few strategic shifts. Then relax, keep learning and stay focused on the human side of marketing.