"Optimizing for Google will basically Optimize for Generative Engines, like Perplexity, Claude, Llama, ChatGPT", not so fast SEO's.
Question: Optimizing for Google is primarily the way to optimize in one of the above Generative AI options. It will help you for 20-80% of the queries in another (made up stat), and will not help you very much in 2 other ones. Do you know which is which?
Let me explain why I think that phrase while rooted in truth, shows a lack of nuance about where people might be going to get answers and how those answers are generated.
Search marketers know that optimizing for local search, is a different algorithm than image search, paid search, and organic search, yet we’re not seeing the same distinctions in AI and I think that we need to take a step back and define what exactly we’re optimizing for, and how to optimize for each type of AI search “engine”. Yet, when it comes to influencing AI or large language model (LLM) responses, the internal chatter is heavy on ideas which is a great start, but I noticed how rarely people would tie the tactics to the type of Generative Engines they were optimizing for. This lack of alignment is a problem. So, let’s get aligned.
Optimizing for Google, gets you most of the way in Bing...that doesn't apply for Perplexity vs Claude.
AI systems aren’t a one-size-fits-all game. Today, I want to walk you through what I see as the three key categories of AI and how we, as digital marketers, can position ourselves to influence them effectively.
Training-First Systems (ex: Claude, Llama)
These systems generate responses based solely on their training data—snapshots frozen in time until the next model update. If you want to influence these systems, you’re playing the long game: updating your digital footprint, generating PR buzz, and waiting for the next training data refresh.
Search-First Systems (ex: AI Overviews, Perplexity)
These rely heavily on real-time indexing of web pages. Influence here is more immediate—optimize your visibility in search indexes and ensure your content is among the sources being referenced. It’s like SEO on steroids, with a dash of content curation.
Hybrid Systems (ex. Gemini, ChatGPT)
These systems blur the lines by deciding, often in real time, whether to answer based on their training data or fetch updated information from the web. For example, if you ask ChatGPT about digital marketing basics, it might lean on its training data. But if you ask about breaking news, it’s likely to pull from live web sources.
Question: Do you know which types of questions trigger search results (more real time) vs trained answers (frozen in time)? Knowing that split helps you determine your generative engine optimization strategy, what you can win on in the next 6 months vs what might take you 6 years.
1 min: Optimizing for each of these Generative AI Engines, the high level:
Training-First Systems:
It’s all about building a strong, evergreen foundation, this is about BRAND belief systems. Think thought leadership pieces, case studies, and authoritative articles that establish your brand over time.
Search-First Systems:
Hone your SEO chops and focus on being indexed by relevant search engines. The goal is to show up tomorrow, not next quarter.
Hybrid Systems:
Strike a balance. Make sure your foundational content is solid while staying agile enough to provide real-time updates that keep you relevant.
Deep dive: Marathoners, Sprinters, and Decathletes Define Your Generative Optimization Strategy:
The Marathoner: Volvo + Training-First Systems
When you think of Volvo, one thing comes to mind: safety. It’s a brand identity that wasn’t built from “content marketing” or “social” — it’s the result of decades of consistent messaging, innovation, and delivering on their promise to protect lives. You can not win here with backlinks, you win with beliefs.
Volvo’s commitment is embedded into their products, from their pioneering seatbelt designs to modern crash-avoidance systems.
Look at these 1970's ads.
For Training-First AI systems like Claude or Llama, Volvo’s approach is the perfect analogy.
Influencing Training-First Systems is about playing the long game.
Just as Volvo has spent years embedding safety into their DNA, and taking that message to market, your strategy here needs to focus on building a digital presence that’s authoritative, consistent, and valuable. It’s not flashy or quick, but when these systems update their training data, your brand becomes part of their core knowledge.
This will be a challenge in a world of hitting the quarterly numbers and CMOs getting fired every 18 months, could this be an advantage for “steady” companies with strong belief systems, we’ll see.
How would you as a Generative Engine Optimizer beat Volvo?
You don’t.
Honesty. A generative engine optimization consultant should remind their client Volvo has a 98 year head start, and realistically our best case scenario is moving into #2-3-4 for “safety” - but we’re not taking #1.
Control what you can: Develop high-quality, evergreen content that positions you as an industry leader in gaps where we can, queries for cars “like volvo” where maybe the searcher doesn’t want one could be a place to start.
Break the silo: Partner with Advertising, PR and other long term reputation-building campaigns that solidify your brand in the public consciousness. This is a LONG term play so consulting wise you have to ask questions around the long term commitment to this effort, and be honest with your leadership.
The Sprinter: Search-First Systems
These systems operate in real-time, relying on web indexes to generate answers. If Training-First systems are a marathon, these are a 100-meter dash.
The question isn’t how you’ll show up eventually; it’s how you can show up quickly and what queries are best answered this way?.
For sprinters, let’s look at a hypothetical example of a brand launching a a pop-up shop or new product drop. Think Nike releasing new running sneakers.
The goal is immediate visibility for customers questions, and success hinges on being visible.
Influencing Search-First systems, mining the clues:
Organic clue #1: 2024 Queries. Best running shoes 2024 might be better than best running shoes for wide feet (New Balance is embedded long term). Look for the ngrams and other groupings where you see a lot of 2024 or "year" style queries.
Organic clue #2: Publish dates. You could also look at title tags for the top 10 that contain years, or publish dates by query…what kinds of queries have more recent dates as answers, maybe they present a better opportunity to win today.
This type of "engine" is most optimized by traditional SEO tactics, so if your customers are here in a major way, you are good to use the SEO-ing = Influencing Generative Engines approach.
- Leveraging SEO tactics and ensuring your content is highly relevant and optimized.
- Creating timely, newsworthy content that feeds the system what it needs, new content, as users are looking for it that way.
- Indexability & indexability reputation is KEY here.
The Decathlete: Hybrid Systems
Finally, we have the decathletes—the Hybrid AI systems like ChatGPT and Google’s Gemini. These systems are constantly switching between pulling from training data and accessing real-time information. To succeed here, you need the versatility of an all-around athlete.
To win in Hybrid systems, you must do the innovative work to separate your customers potential queries into which will trigger "search results" in ChatGPT vs which ones ChatGPT will answer based on legacy training data.
Tell me about Seer Interactive was my prompt here, traditional SEO FTW:
Tell me about Volvo's safety track record was my prompt here, traditional SEO ain't gonna cut it (for my old school rap fans).
Then I switched my prompt to "What about Tesla". The result? Pulled from training data. But then... I said tell me about their most recent updates, web results...
If I’m Tesla then I’m already synonymous with Electric cars in the training data.
Our newest software update won’t be part of the training data, though. If I want to be visible in chats about our bleeding edge features, what is my strategy?
For marketers, the Hybrid playbook includes both versions of what we heard above:
- Evergreen FTW: Ensuring your evergreen content is robust enough to provide answers from training data.
- Updating Cadence: Sometimes evergreen content gets stale, or out of date. Put a system in place to catch it, fix it and make it relevant in today’s context. Self driving was new when they launched, but do you want to just leave your visibility up to your historical context, or do you want to keep reinforcing it as others have started to play catch up?
- Staying agile: Be ready to address trending topics with up-to-date content (look for new themes in your branded search from paid search to guide you)
Choose Your Lane—Or Master Them All
I hope I made the case that there’s no single approach that fits every scenario.
Your success will depends on understanding the strengths and weaknesses of each category and how your customers search for answers.
Whether you’re building a Volvo-level legacy with Training-First systems, sprinting to the top with Search-First systems, or balancing both worlds as a Hybrid decathlete, the future is wide open for those willing to adapt.
Yes, the road ahead might feel uncharted, but the fundamentals remain the same: understand the customer, follow them wherever they are, stay curious,, deliver value, stay relevant, and keep evolving.
Whether you’re influencing a training-first AI with slow-burn content, mastering the fast-paced world of search-first systems, or juggling both with a hybrid, there’s room to win—if you’re ready to adapt.
So, what’s your play? Are you doubling down on your long game, sprinting for quick wins, or aiming for all-around mastery? The answer might just define your success in the AI-driven future.
