Last Updated on November 2, 2026
I’ve spent nearly nine years working in marketing at Udacity, and I’ve watched search evolve through AI Overviews, the rise of video, and more algorithm updates than I can count.
Generative AI has changed search more dramatically than any of those developments.
Not because everything we’ve learned about SEO is suddenly wrong. Technical foundations still matter. Helpful content still matters. Authority still matters. But AI tools have changed how people research products, compare options, and make decisions. That’s changing how brands get discovered.
The clearest picture of what’s actually shifting comes from practitioners doing this work every day: Katie Hale, who built Udacity’s course on AI search and discovery, and Megha Jain, a Udacity mentor and performance marketer. The biggest takeaway from both is reassuring: you don’t need to abandon your SEO strategy, but you definitely need to expand it.
GEO vs SEO: what’s the difference?
“I like to call them sisters. You know, we’ve got AEO, GEO, SEO. They all came from the same root concept.” — Katie Hale
Traditional search engine optimization (SEO) focuses on helping your content appear in search engine results. You identify relevant queries, create useful pages, establish authority, and make sure search engines can crawl and understand your website.
Generative engine optimization focuses on making your brand and content visible in AI-generated answers.
When someone asks ChatGPT or Perplexity to recommend a product, explain a concept, or compare providers, the response may draw on information from multiple websites. Depending on the platform and query, it may mention brands, summarize their content, or link to supporting sources.
GEO aims to increase the likelihood that your business appears in those answers, ideally when someone is asking a question relevant to what you offer.
There’s an important distinction here. Google AI Overviews appear within traditional Google search results. ChatGPT, Perplexity, and other standalone AI assistants provide a different search experience. Their underlying systems and methods for selecting sources aren’t identical, so a tactic that helps on one platform won’t necessarily have the same effect on another.
Still, the marketing challenge is similar: how do you make sure your brand shows up when people are making a decision?
That’s a better way to think about it than treating GEO as a replacement for SEO. The disciplines overlap considerably, even as the places where your content can appear continue to multiply.
Why your search traffic might be falling even when your rankings aren’t
One of the more frustrating patterns in organic search right now is something I’ve seen described as the crocodile: impressions climb while clicks decline, creating a widening gap between the two lines in Google Search Console that resembles the open jaws of a crocodile.
Katie sees this with her B2B clients. Pages can maintain strong rankings and continue appearing in search results while attracting fewer visitors.
One possible contributor is Google AI Overviews. When Google answers a question directly in the results, some users no longer need to click through to the original website.
Of course, not every decline in click-through rate is caused by AI Overviews. Changes in rankings, query mix, search features, and user behavior can all affect performance. But the broader concern is real: a brand can remain visible in search without receiving the traffic it once would have.
In Katie’s experience, some visitors who arrive after researching a product through AI are further along in their decision-making process. They may already understand their options and have specific reasons for visiting a company’s website.
She believes those visitors can be more likely to convert, particularly in the B2B environments where she works. That raises the question: are we evaluating organic performance solely by traffic, or are we also looking at how effectively we’re reaching potential customers?
GEO is about more than what’s on your website
One of the biggest differences Katie emphasized is the scope of the work.
Traditional SEO already extends beyond your own website through backlinks, reputation, and other authority signals. But GEO makes the importance of your broader brand presence especially difficult to ignore.
Someone asking an AI assistant for a recommendation may encounter information drawn from your website, customer reviews, third-party publications, community discussions, and comparison pages.
The precise ways AI systems select and weigh those sources aren’t fully transparent, and they vary by platform. That’s why Katie emphasizes that marketers need to think beyond optimizing individual landing pages.
Consider a B2B software company. Its website might explain its features beautifully, but a prospective buyer could also ask an AI assistant how it compares with competitors, whether customers find it reliable, or what limitations people have encountered.
Those answers may depend partly on information the company doesn’t publish itself.
Katie shared an example from her agency work: a client improved its G2 profile by adding more complete information and encouraging customers to leave reviews. She subsequently observed the brand appearing in relevant AI answers.
That doesn’t establish a universal causal relationship between G2 reviews and AI visibility. But it illustrates an opportunity that a website-only optimization strategy would miss.
It also explains why GEO can’t sit entirely with an SEO specialist. Content marketing, PR, product marketing, customer advocacy, and community management all contribute to the information available about a brand.
That’s why Katie calls this broader approach “search everywhere optimization”.
How to make your content more useful for AI search
So what should marketers actually change?
In a recent Udacity workshop, Megha Jain outlined five principles for making content easier for AI systems to interpret and reference.
They’re not a proven set of ranking factors. Think of them as practical guidelines for making content clearer, more informative, and more useful.
1. Structure content around questions and clear answers
People often use conversational prompts when searching with AI. Rather than entering a short keyword phrase, they may describe their circumstances, constraints, and desired outcome.
Your content should address those questions directly.
That might mean turning a vague heading like “Our approach” into a more specific one, such as “How does this service compare with an in-house solution?”
Then answer the question immediately, before expanding with supporting detail.
Megha also cautions against stuffing FAQs with keyword variations. The goal isn’t to manufacture dozens of questions. It’s to answer the ones customers genuinely ask.
2. Prioritize depth over length
More words don’t necessarily make a page more useful.
Megha’s point is that a strong answer provides enough context for someone to understand the recommendation. That might include specific examples, comparisons, limitations, or supporting data.
If you’re comparing products, for instance, don’t simply list their features. Explain which differences matter, who each product is best suited for, and why.
A comparison table can help readers understand those distinctions quickly. A clear conclusion can be equally valuable.
The objective is to give someone enough information to make a decision, not to hit an arbitrary word count.
3. Support claims with credible evidence
Generic claims are easy to make and difficult to evaluate.
If you say your product saves customers time, explain how. If you cite a performance improvement, provide the underlying data. If you’re making a broader industry claim, link to credible research.
Megha emphasizes the difference between asserting something and supporting it with evidence.
Original research, documented customer results, and firsthand expertise can be particularly valuable because they contribute information that isn’t simply repeated across dozens of other websites.
4. Make relationships between topics and entities explicit
AI systems need to interpret what your content is actually about.
If you’re comparing several products, explain how they relate to one another. If you’re discussing a technical concept, clarify its connection to the broader topic and the problem it solves.
Megha illustrates this with running shoes: mentioning Nike, Brooks, and Asics on the same page isn’t especially informative by itself. Explaining how particular models compare for different running needs is.
This is useful content design regardless of whether an AI assistant ever references the page.
5. Keep important information accurate and current
Outdated prices, discontinued products, old policies, and stale comparisons can make a page less useful.
Megha recommends reviewing important content regularly, particularly when the subject changes quickly.
That doesn’t mean every article needs constant rewriting.
Katie offered an interesting counterexample: she has content written more than a decade ago that still ranks well because it answers a question people continue to ask.
The lesson isn’t that newer content always wins. It’s that accuracy and relevance matter more than the appearance of freshness.
Where I’d start: optimize what’s already working
Knowing the principles is one thing. Deciding where to spend time is another.
When I asked Katie how she’d approach GEO for a new client, she didn’t recommend creating a huge library of AI-focused content or immediately investing in every emerging platform. She started with existing pages.
Her first recommendation was to examine pages already ranking on the first two pages of Google. Those pages have demonstrated some relevance for their target queries, making them sensible candidates for improvement.
I’d turn that into a fairly straightforward workflow:
- Identify promising pages. Use Google Search Console to find pages with meaningful impressions and relevant queries, particularly those already ranking reasonably well.
- Find the questions behind the queries. Look for searches beginning with who, what, when, where, why, how, can, or should. Review Google’s People Also Ask results and the questions customers bring to your sales or support teams.
- Improve the answers. Add clear explanations, relevant comparisons, supporting evidence, and question-and-answer sections where they genuinely help.
- Check technical fundamentals. Make sure the content is accessible, properly structured, and supported by appropriate structured data where relevant.
- Test your visibility. Ask relevant, non-branded questions across the AI platforms your audience uses. Document whether your brand appears, what sources are cited, and how the answers describe you.
Once you’ve addressed the most obvious opportunities on your own site, Katie recommends looking more closely at your competitors’ broader presence.
Where do they have reviews? Which publications mention them? What questions are customers asking about their products? Are there relevant communities where your brand has little presence?
Those findings can help you decide whether to invest in customer reviews, third-party content, PR, or other forms of brand visibility.
The right priority will depend on your audience and where they actually look for information.
How to measure GEO when the data is incomplete
This is the part I find most challenging. In traditional SEO, Google Search Console gives us a reasonably detailed view of the queries generating impressions and clicks. We can see which pages are performing, how rankings change, and where opportunities exist.
With standalone AI search, we don’t have an equivalent view of all the questions people ask, how often our brand is considered, or how frequently it appears in answers. That makes measurement less straightforward, and it’s why Katie recommends using three complementary sources of information.
Google Search Console: understand traditional search performance.
Continue tracking impressions, clicks, click-through rates, and the queries associated with your important pages.
Look beyond aggregate traffic. Are you losing visibility for queries that matter to your customers, or for broad informational searches that rarely converted?
Katie described a client whose number of top-ranking keywords had declined, initially making performance look worse. But a closer look showed that some of the lost rankings were for less relevant consumer-oriented searches, while more valuable B2B queries were improving.
The headline metric didn’t tell the whole story.
GSC can also help you understand performance in Google Search, where AI Overviews may influence user behavior. It does not, however, provide a complete, separate reporting system for every AI Overview appearance or for standalone AI assistants.
Google Analytics: monitor traffic from AI platforms.
GA4 can identify some visits referred by platforms such as ChatGPT and Perplexity, depending on how referral information is passed and how your reporting is configured.
You can build a reporting segment for known AI referral sources and compare engagement, conversions, and other outcomes with traditional organic search.
But this is only part of the picture. Some visits won’t be attributed correctly, and many interactions with AI-generated answers never produce a website visit at all.
Katie emphasized that this tracking isn’t completely reliable.
I’d treat referral traffic as a useful directional measure, not a complete representation of AI search visibility.
AI visibility tools: monitor relevant questions and brand appearances.
A growing number of tools aim to help brands measure their visibility in AI-generated answers.
The process starts with defining the questions or topics you want your business to be associated with. A tool can then test those prompts across supported platforms and monitor whether your brand is mentioned, whether sources link to your site, and how the answer presents your business.
Katie made an important distinction here: focus on non-branded questions.
If you ask an AI assistant about your own company, it’s not particularly surprising when the answer mentions you.
A more meaningful test is whether your brand appears when someone asks about the problem your product solves, without naming you first.
These tools still rely on sampled prompts rather than a complete view of real user demand. But used consistently, they can help you track changes in visibility that ordinary web analytics won’t capture.
Ultimately, I’d want to look at these signals together: search performance, AI referral traffic, brand visibility in relevant answers, and downstream conversions. None is sufficient on its own.
Don’t lose sight of who you’re trying to reach
There’s a temptation with every major change in search to chase the latest tactic.
One month everyone is talking about Reddit. The next month it’s Wikipedia citations. Then it’s a particular schema implementation or content format.
Some of those developments deserve attention. But they can also distract us from a much more basic question: are we creating information that helps the people we actually want to reach?
Katie gave me a memorable example from her time working with a university.
One academic program used the abbreviation “MLB” internally. Keyword research showed substantial search volume for MLB, which might have looked attractive on paper.
The problem was that most people searching for MLB were interested in Major League Baseball, not the university’s laboratory sciences program.
Ranking for that term wasn’t going to bring the right students.
It’s a simple example, but I think it captures something important about GEO.
The questions people ask AI assistants may be longer and more conversational than traditional search queries. We have less visibility into their volume. And we don’t fully control how those systems choose which brands to recommend.
But the underlying marketing problem hasn’t changed.
We need to understand who our customers are, what they’re trying to accomplish, and where our product genuinely fits.
That should guide which questions we answer, which pages we improve, where we build our presence, and how we measure success.
GEO introduces new surfaces and new constraints. It doesn’t eliminate the need for sound marketing judgment.
If you want to put these ideas into practice, Katie developed the AI search and discovery course in Udacity’s AI Marketer Nanodegree program. It covers practical GEO techniques and culminates in a website audit project where you’ll identify optimization opportunities and present recommendations a stakeholder can act on.
