Traditional Search vs. AI Search: The Side-by-Side Comparison (diagram)

Search is changing. This is the biggest story in digital marketing.

The way we use the internet is in flux. Billions of users, millions of marketers and thousands of SEOs are adapting fast. So let’s pause, step back and think about the moment in internet history we are living through.

Today, we’ll summarize the big change in a small chart.

We’ll take three points of view: the technology, the user behavior and the marketing implications. As a bonus, we’ve added a prediction for the future at the bottom. Because times are changing fast.

Here’s a simple way to understand the difference for lead generation marketing.

• Search is about getting ranked and clicked. The website then answers their questions, shows supportive evidence and converts the visitor into a lead.
• AI is about getting recommended based on the user’s context. Responses include comparisons and trust signals (reviews, testimonials, awards, case studies)

In other words, AI reaches farther down into the lead generation funnel.

This has implications for marketing.

1. Content needs to align closely with the decision criteria of buyers. Copywriters need to be specific about who gets what outcomes from which services.
2. Proof points need to be crawlable by AIs. Those award logos and testimonial videos aren’t easily ingested.
3. PR, review sites, directories, industry associations and influencers all become more important. Brands with big targeted digital footprints have an edge.

It’s all about getting the right answers and evidence into the AI training data.

You’ll find some very specific recommendations below, but first let’s look closer at the items from the infographic and share the related research for each. We’ve also included input from experts who watch this space closely.

The technology

Information retrieval is library science. Users query and get results. It’s how the web has been organized and accessed since the dawn of http. Each major category has a few dominant search engines (travel, jobs, real estate, music) and Google rules for general inquiry.

The underlying tech is algorithmic. Content gets crawled, indexed and ranked. Results are determined by ranking factors. Traditional search is deterministic.

Generative AI is a pre-trained transformer. It first ingests huge amounts of training data (mostly internet content), then uses billions of parameters to generate language (or code) in response to prompts. There are six major foundational models and millions of tools and apps built on top of them.

The underlying tech is predictive. Responses are simply predictions of the next token of language in a sequence. Generative AI is probabilistic.

User behavior, user experience

Here’s the simplest way to understand the difference in how your future prospects uses traditional search and AI search: Google gives options. AI gives recommendations.

That insight alone may give you plenty of ideas for adapting your website and content strategy. We’ll share practical approaches below. But first, we’ll look at how people use AI, because every lead begins and ends with a user action.

Open AI recently published data showing how people use ChatGPT. If you’d prefer not to read the 60-page report, I’ll summarize it here. People use generative AI in all kinds of ways that have nothing to do with search or marketing.

The report categorizes AI use into 25 use cases across seven categories. I count just four that relate to how users find content and make buying decisions. Generative AI isn’t mostly about search for answers and marketing.

But times change fast. Deeper in the report, they show the trends. “Seeking information” recently surpassed “Writing” as the most popular category.

But it’s still not mostly a search tool. Crunch the numbers for the latest months, and it looks like around one third of prompts are searches or buying decision research.

This is less than data from the research by Profound, which uses far fewer categories, and suggests that nearly half of ChatGPT prompts show search intent.

Either way, people use AI to get information. But when we do, we are looking for more than just options. We are looking to get recommendations. We are doing research and asking for help finding a fit for our specific needs. This is obvious in the research. The average prompt is 7x longer than the average search query.

• The average search query is 3.4 words (source)
• The average AI prompt is 23 words (source)

Those longer prompts are about getting more personalized responses. In a survey we conducted with QuestionPro, we found that the bigger the decision or more complex the answer, the more likely people are to get help from AI.

This suggests that users are asking AI for help with research. AI is a decision support tool. It’s about getting recommendations, not just options. So the user is more confident and informed when they reach a website. They are ready to act. They are more likely to become a lead.

You can find evidence of this in your own Analytics. Use this report to check the conversion rate from AI sources. You’ll likely see that visitors from AI sources convert at higher rates than visitors from traditional search. In this account, it’s double.

With that in mind, let’s explore the implications for marketing.

The marketing differences

Let’s start with the training data. AI ingested the internet and it responds based on what it read. So the marketers job is to add to the LLM’s training data by publishing content on the web, both on the company website and in other places related to the industry.

There’s lots of research showing that AI trains on our websites. One a study conducted by MuckRack of a million AI citations (those are the links in responses) showed that nearly half of all citations were to company websites and coporate blogs.

The takeaway is clear: AI trains on company websites. Make your website the best possible resource on your brand. Make your content program a credible resource for your industry.

But remember that every LLM needs to retrieve information to provide good responses. For for many prompts, ChatGPT will immediately search the web, often using Google.

A clever study by Backlinko proved this by creating content about nonsense phrase, blocking access to the content from everything but Google, then asking ChatGPT about the phrase. It searched and found it, therefore it searched with Google.

Again, the takeaway is clear: AIs search the web with search engines. So continue to optimize your content for traditional search.

The best practices for SEO and GEO (generative engine optimization) or if you prefer, AEO (answer engine optimization), are very closely aligned. In this study by AirOps, you can see the overlap. Content should be highly structured with clear H1s, bullets and schema markup.

When it comes to structure, optimize for traditional search and you’ve optimized for AI. 

But this may not be sufficient if the goal is to be the recommended brand. Remember, prompts are longer than queries. The user may be doing serious research. Prompts often include role and context. The user is asking for specific, personalized recommendations. They want to see evidence that the brand is a strong fit.

That’s why we need to get more into the training data. Here are 10 types of content that may help you get recommended by AI, but often are not on the checklists for traditional SEO copywriters:

1. Case studies with outcome data within your prospects’ industry
2. Comparisons between your services and alternatives
3. Pricing information presented in plain text
4. Testimonials and quotes from real customers and partners
5. Clear support policies, guarantees and service level agreements
6. Detailed product or service specs written in concise language
7. Step-by-step explanations of your processes
8. Awards, certifications and memberships listed as AI-readable text
9. The job titles of the professionals your company helps
10. Leadership bios with relevant qualifications

Ideally, these kinds of specific answers and supportive evidence appear in many places across the web. But without question, they should appear on your website. All of the answers to all of the questions they ask on sales calls and all of the proof points from your pitch deck should be on your pages.

Lidia Infante, SurveyMonkey “The nature of the training corpus of LLMS has made a holistic view of a brand’s online presence more important than ever. While users used to mostly get information about you from you, now LLMs can surface sources that we would have never paid attention to before. This brings up a super interesting challenge surrounding online reputation management and digital PR: having to switch gears from what’s important today to the bigger picture.”

In a way, the user is having a sales call, but your sales reps aren’t on it. They are having the initial call with the LLM. If we accept this is the new paradigm, then digital marketers have a new job: train the AI to be a sales rep for your brand.

Mike King, iPullRank “Search has shifted from being a pure performance channel to being a branding channel. AI isn’t just surfacing links, it’s making recommendations and that means your brand equity decides whether you show up. Most marketers aren’t ready for that reality, but it’s an enormous opportunity. If you build a brand that AIs trust, you don’t just compete for clicks, you dominate the conversation.“

The future of finding information

The line between traditional search and AI search has been blurred for years. LLMs search the web; search engines summarize results and generate answers. All the tools are hybrid tools. All the models are multi-modal.

The tools are still separate and distinct. Users choose whichever suits their needs in that moment. Looking for something specific? Go to Google. Need help with a task? Pop open ChatGPT.

But the technology and user behaviors are changing. Here’s our predictions for the near future, middle distance and farther out.

Coming next…

In the next era, the tools will converge and our experience as users will change. Here’s what the convergence may look like:

• Search engines will feel more conversational. We’ll talk to them more about what we’re looking for.
• AI responses will look for more search results. We’ll see ads, maps and videos.

What tends to work will become clear and marketers will settle on some best practices. Incumbents, influencers and companies with better websites will have advantages.

Soon thereafter…

In the next era, the major tools will simply give us answers and options anytime, based on who we are. We won’t think about the difference between retrieving information and generating answers. We’ll open our super agent of choice and expect it help with both finding things and making things.

Smart marketers will adapt with clever campaigns, better stories and content tuned specifically to train LLMs and carefully optimized websites.

Mark Schaefer, Author of How AI Changes Your Customers: The Marketing Guide to Humanity’s Next Act “When AI recommends, we ‘consider.’ When humans recommend, we act. The next evolution of search must involve the human override, and this will occur in two ways. First, branding will be more important than ever. If we develop true loyalty, the brand name will be embedded in an AI search (“Please plan this trip for me, and make sure I stay at Hilton Hotels.”). This not only creates a direct referral, it “trains” your AI companion to steer you toward a brand in the future. Second, word-of-mouth marketing has always been the most trusted and effective referral engine. And yet, it has been overlooked by almost every organization. Can you create meaningful, helpful, conversation-worthy brand buzz that overrides the need for search? Almost every significant purchase I make comes with advice from trusted friends and experts.”

Eventually…

New wearable devices will reduce the friction between humans and AI tools. We’ll all be whispering to our glasses and listening to responses through earbuds. A new generation will come to expect on-demand personalized interactions with the web.

We’ll have AI agents that we trust make decisions for us and complete the related tasks by talking to other AI agents directly.

Potential buyers will ask AI to build cases for various options. User experiences will become more custom. Trust signals will dominate. Ads, brand marketing and offline efforts will drive impact. So will deep-niche PR, influencer collaborations and content marketing.

That’s what we see when we peer into the crystal ball of marketing. What do you see coming next?

Scott Brinker, chiefmartech “The biggest disruption ahead is the use of more agentic AI by buyers. If the Web’s big go-to-market revolution was a disintermediation between buyers and sellers, AI is poised to bring whole new layers of intermediation. The more buyers delegate to their own AI agents, the more marketers need to develop the capabilities to engage and delight these agents as much as the humans they’re operating in service of.”