The real shift: you’re not just ranking in Google anymore

Look at those headlines again and a pattern jumps out: everyone is quietly admitting that “SEO” is no longer just about blue links on a Google SERP.

You’re now competing for:

• ChatGPT answers and “shopping” suggestions
• AI overviews and “LLM-optimized” snippets
• Google Discover and other feed-style surfaces
• Social feeds and visual search (Pinterest, Instagram, TikTok)
• AI-written summaries of your own site and reviews

The common thread: AI and algorithmic feeds are becoming the new “homepage” for discovery. Search is still huge, but it’s no longer the only, or even the first, place people encounter your brand.

For performance marketers and media buyers, this isn’t a philosophical problem. It’s a planning problem:

• Where do you actually optimize?
• What still moves the needle?
• How do you avoid chasing every shiny new acronym (AEO, GEO, LLMO) and still not fall behind?

From SEO to “surface optimization”

Let’s give this a name that’s actually useful: you’re doing surface optimization.

A “surface” is any place an algorithm decides what to show:

• Google SERPs and Discover
• ChatGPT / Claude / Perplexity answers
• Amazon, Pinterest, TikTok, Instagram feeds and search
• News and content recommendations

The old game: optimize for one dominant surface (Google search) and let everything else trail behind.

The new game: pick the 2-3 surfaces that actually drive revenue for you and deliberately optimize for how their algorithms see and summarize your brand.

What’s actually changing under the buzzwords

Strip away the acronyms and think pieces and three real shifts matter:

1. LLMs are new “meta search engines”

Tools like ChatGPT, Claude, and Perplexity:

• Read and summarize the web
• Blend organic content, reviews, and sometimes ads
• Act as a “first pass” filter before users click anything

That means:

• Your product may be “chosen” or ignored before the user ever hits a SERP or marketplace.
• Brand, authority, and clarity of positioning matter more, because LLMs compress options.

2. Feeds are decoupling from search rankings

Google Discover, Pinterest, TikTok, Instagram, YouTube, even email inboxes – they all run their own recommendation logic.

Recent coverage about Google Discover being less aligned with search rankings is a polite way of saying:

Your #1 ranking blog post might get zero Discover exposure, and a random mid-performer might catch fire.

Feeds care about:

• Engagement signals (click, dwell, save, share)
• Freshness and recency
• Format and creative (thumbnails, titles, visuals)
• User behavior and interest clusters

3. Automation is opinionated – and not always in your favor

Google Ads auto-apply, “recommendations,” smart bidding, and now AI assistants inside ad platforms are not neutral. They’re trained to:

• Spend your budget
• Hit the platform’s default metrics
• Favor broad, easy-to-measure outcomes

At the same time, custom GPTs and workflow tools are giving you more control if you know what you want them to do.

So you’re stuck between:

• Platform automation that optimizes for the platform
• Your own automation that’s only as good as your prompts and data

A practical framework: AOI (Algorithm, Object, Intent)

To make this usable in a media plan, stop thinking “SEO vs social vs email vs AI” and start thinking:

Algorithm → Object → Intent

For any surface you care about, ask three questions.

1. Algorithm: what is this system trying to maximize?

Examples:

• Google Search: Relevance + authority + satisfaction (no pogo-sticking).
• Google Discover: Clicks and engagement from users with certain interest profiles.
• ChatGPT / Claude: Helpful, non-controversial, “balanced” answers that cite credible sources.
• Pinterest: Saves, clicks, and downstream engagement on visually appealing content.
• Meta / TikTok feeds: Time spent, repeat visits, and content that keeps you scrolling.

If you don’t know what the algorithm wants, you’re just throwing content at it and hoping.

2. Object: what is the unit that actually ranks?

This is where a lot of teams get sloppy.

• On Google Search: the URL (page) is the object.
• On Discover: the article with its title, image, and topic cluster.
• On ChatGPT: the source document or brand that gets cited or summarized.
• On Pinterest: the pin, not your homepage.
• On Instagram: the post or reel, not your profile.
• In email: the send (subject line + audience + content).

Once you know the object, you can design it for that surface instead of just recycling whatever you had lying around.

3. Intent: what is the user trying to do on this surface?

This is where performance lives.

• On ChatGPT: “Compare options for me and give me a short list.”
• On Discover: “Show me interesting stuff related to what I’ve been into lately.”
• On Pinterest: “Give me visual ideas I can save and maybe buy later.”
• On TikTok: “Entertain me, but I might impulse-buy if it feels right.”
• On email: “Don’t waste my time. Make this feel relevant or I ignore you.”

When you line up Algorithm → Object → Intent, you can finally make tradeoffs that aren’t random.

How to adapt your strategy in the next 6-12 months

Here’s how to turn this into an actual plan instead of another “AI is changing everything” rant.

1. Pick your critical surfaces

Don’t try to optimize for every algorithm at once. Choose:

• 1-2 discovery surfaces (e.g., Google Search + TikTok, or Pinterest + Instagram)
• 1 intent-heavy surface (e.g., ChatGPT / Perplexity, Amazon, or your own site search)
• 1 retention surface (e.g., email, SMS, app push)

Tie each surface to a specific business metric:

• Discovery: assisted conversions, new users, view-through revenue
• Intent: last-click revenue, lead volume, CAC
• Retention: repeat purchase rate, LTV, churn

2. Make your brand “LLM-readable”

If you assume an LLM will summarize you to a stranger, what does it see?

Practical steps:

• Clarify your positioning on your homepage and key category pages in plain language. LLMs love clear, unambiguous statements.
• Strengthen third-party proof: reviews, testimonials, case studies, and press on sites that are likely to be crawled and trusted.
• Publish comparison content (“us vs alternatives”, “best X for Y use case”) that’s honest and specific. These often end up as source docs for “best” lists in AI answers.
• Use structured data (schema) where it actually helps: products, FAQs, reviews, organization info.

Then literally test it: ask ChatGPT, Claude, and Perplexity:

• “What are the best
  options for [use case]?”
• “Who are the main companies that do [your category]?”
• “Compare [you] and [competitor] for [audience].”

Track how often you appear, how you’re described, and what sources are cited. That’s your new “LLM share of voice.”

3. Treat feeds like paid channels, even when they’re organic

Discover, Pinterest, TikTok, Instagram – they all respond well to the same discipline you already use in paid:

• Creative testing: Test hooks, thumbnails, and formats like you test ad creatives.
• Audience hypotheses: Build content for specific interest clusters, not “everyone.”
• Frequency and recency: Plan content calendars around recency decay, not random bursts.
• Measurement: Use UTM parameters, view-through logic, and simple MMM-style analysis to estimate impact.

The mindset shift: “organic” doesn’t mean “free.” It means you’re paying with time, creative, and opportunity cost instead of media dollars.

4. Put guardrails on platform automation

With Google Ads recommendations, auto-apply, and “helpful” AI suggestions, your job is to:

• Define the sandbox: Clear budgets, target ROAS / CPA, and excluded placements or audiences.
• Lock in non-negotiables: Brand terms strategy, negative keywords, geo constraints, and compliance rules.
• Audit weekly: New keywords added, bids changed, budgets shifted. Roll back what doesn’t align with your goals.
• Use custom GPTs or scripts to monitor changes and flag anomalies, instead of trusting the platform’s summary.

Automation is fine. Unsupervised automation is a tax.

5. Use AI for production, not strategy

The temptation with “AI SEO tools,” auto-title rewrites, and content generators is to let them decide what you should care about.

Flip it:

• You decide the surfaces, objects, and intents that matter.
• You use AI to speed up execution against that plan.

Concrete uses that actually help:

• Generate 20 variations of titles and hooks based on your brief, then test them.
• Summarize long research or transcripts into outlines for content aimed at specific surfaces.
• Draft first-pass ad copy, email variants, and meta descriptions that you then edit with a performance lens.
• Cluster keywords or topics into intent-driven groups for specific surfaces (e.g., “Discover-friendly”, “LLM-friendly”, “bottom-of-funnel”).

What to stop doing

To make room for this, some habits need to die.

• Stop treating “SEO traffic” as one thing. Break it into search, Discover, and AI-assisted visits.
• Stop chasing every new acronym. If it doesn’t map to a surface you care about, ignore it.
• Stop publishing content that only exists to rank. If it’s not useful to a human or a summarizer, it’s dead weight.
• Stop letting platforms quietly auto-apply changes without a human in the loop.

The operators who win this wave

The teams that win the next 2-3 years won’t be the ones with the fanciest AI stack. They’ll be the ones who:

• Know exactly which surfaces matter for their funnel
• Understand how those algorithms think and what objects they rank
• Use AI as a force multiplier on a clear plan, not as a substitute for one
• Measure impact across surfaces instead of obsessing over one channel’s vanity metrics

You don’t need a new buzzword for that. You just need to accept that “ranking” now happens in more places than a SERP – and plan like it.