If your website traffic has felt like it fell off a cliff lately, you’re not alone. The shift isn’t just “SEO is getting harder.” It’s that buyer behavior is moving from clicking links to consuming answers directly inside AI interfaces—Google AI Overviews, ChatGPT, Perplexity, Copilot, and whatever comes next.
Here’s the uncomfortable truth: in 2026, you can have the #1 SERP for a high-volume keyword and still lose traffic because the answer is served before someone ever reaches your site. So the question becomes: how do people buy in an AI-powered age, and how do you show up in the places where decisions are being shaped?
The answer is much bigger than SEO; this is the new Generative Engine Optimization (GEO) world, and if you’re still ignoring it, you’re in trouble. Here’s how GNW is helping clients show up in LLMs ahead of their competitors.
This post recaps (and extends) the key ideas from two of our recent videos:
The problem: “Impressions are steady, clicks are down.”
A pattern many teams are seeing: impressions hold steady, but clicks decline sharply. That’s exactly what happens when an AI Overview answers the question up top—people get what they need without visiting the source pages.
Two implications follow:
- Most queries show a 1/3 traffic decline when AI summaries start taking over the keyword (this is increasingly true for long-tail keywords as well).
- Some publishers and sites may experience an outsized drop—the kind of drop that becomes an organizational crisis, not just a marketing problem, as the Wall Street Journal reported about Business Insider in 2025.
This isn’t a reason to abandon SEO. Google still matters. People still search. But it is a reason to stop treating a #1 SERP as the gateway to your website. If you aren’t hedging other paths to growth, you will get left behind.
GEO vs SEO: same universe, different physics
A lot of people (namely traditional SEO agencies) want GEO to be “just like SEO.” It’s not.
Here’s the cleanest way to think about the difference:
- SEO optimizes for text-based queries and aims to drive clicks to your web pages.
- GEO (Generative Engine Optimization) optimizes for AI-generated answers and summaries and aims to earn citations/mentions….even when users don’t click.
That no-click element matters: in GEO, being cited can be “good enough” because brand recognition and consideration can happen without a session on your site. Marketers are increasingly being pushed to embrace the “no-click” world. From social, to now GEO, we must give up the notion that we control the buyer’s journey. We don’t; they do.
Even more importantly: GEO expands the playing field.
SEO mostly focuses on your web properties (technical + content improvements). GEO still cares about your site, but it also cares about everywhere an LLM might learn about you, including what you don’t own: YouTube, Reddit, LinkedIn, forums, G2, analyst reports, and more (“rented land”).
Step one: get your benchmarks right (or you’re flying blind)
Before you fix anything, you need to know what’s actually happening and where it’s happening.
A practical starting point and call to arms for all marketing teams: take GA4 seriously. Confirm goals, conversions, and that you’re not double-counting. Then run consistent reporting (YoY/ QoQ/monthly, ideally by page) so you can identify which pages are shrinking at a “normal” rate versus which pages are collapsing.
Why it matters: GEO strategy is a prioritization strategy. You can’t optimize everything at once—especially if you’re an SMB or mid-market team without endless budget.
The “Protect, Pivot, Promote” framework (plus “Monitor” when you’re swamped)
Once you have baseline visibility, classify pages:
1) Protect
These are pages that directly drive pipeline, conversions, or revenue intent–think homepage, product pages, pricing, case studies, and high-intent landing pages. Watch them like a hawk.
Expansion: Protect pages should also be the pages you harden against AI misunderstanding:
- clarify your positioning fast (what you do, for whom, and why you’re different)
- remove ambiguity (especially around geography, industries served, and “who it’s for”)
- ensure canonical sources (pricing, product claims, differentiators) exist on pages you control
2) Pivot
These pages used to perform well, but are now getting hit disproportionately. Instead of tweaking keywords, restructure the content so AI engines are more likely to cite it—more contributor-style and evidence-based.
Expansion: Pivot pages are where you test new page patterns (see SCQA below), add structured Q&A blocks, and build explainers that map to how people ask LLMs questions.
3) Promote
These are strong pages that deserve more visibility, so you amplify with link strategy and distribution.
Expansion: Promotion is no longer only about backlinks. It’s also about distribution signals across the wider ecosystem: credible mentions, community participation, and content formats (video, especially) that are being pulled into AI results.
4) Monitor (the reality-based option)
If you don’t have the capacity to do a full classification right now, create a “monitor” bucket and start with your top intent pages first.
The SCQA model: write for how LLMs read, not just how humans skim
One of the most tactical ideas in the walkthrough is using the SCQA model: Situation, Complication, Question, and Answer, packed into the first ~200–300 words. The logic: LLMs overweight the beginning and end of documents, so you structure the page to feed the model what it needs immediately.
How to expand this into a repeatable template:
In the first 200–300 words, include:
- the problem context (what’s happening in the market)
- the tension (why it’s hard / what’s changing)
- the question buyers ask in plain language
- the answer (your stance) + what makes it credible
Then reinforce with:
- structured sections with question-based H2s
- how-to steps and concrete definitions
- citations to high-authority sources where appropriate (e.g., analyst research)
- your own first-party data when you have it (gold for credibility)
This is the bridge between SEO and GEO: SEO taught us structure matters; GEO raises the stakes because structure affects whether you’re summarized and cited.
You’ll note that we do all this in this very article.
Multi-channel discoverability: stop betting everything on one gatekeeper
A core GEO idea is redundancy: be discoverable in multiple channels so you’re not fully dependent on clicks from traditional search.
Here are some source pools that LLMs commonly draw from:
- Reddit and communities (messy but influential)
- LinkedIn for B2B credibility and distribution. This is mostly from Pulse articles right now, not regular posts
- Analyst relations (Gartner/Forrester) for software categories
- Peer review sites (e.g., G2) as a structured corpus of buyer voice
- Video platforms (YouTube, TikTok), because video is increasingly embedded in results
A smart twist: “differentiation-shaped” review strategy
Instead of asking customers for generic 5-star reviews, build campaigns that prompt customers to describe specific differentiators (especially against a known incumbent). LLMs can synthesize themes across reviews, so consistent, authentic messaging patterns can become a competitive edge.
Reverse engineer AI: test how models answer buyer questions (and patch the gaps)
Another highly practical move: prompt the model like a buyer would and observe what it uses as decision criteria. Sometimes hidden factors (like user location) meaningfully change the answers, revealing missing context on your site and in your ecosystem.
Expansion: a simple reverse-engineering sprint
Run a 60–90 minute workshop and capture:
- the 10 questions buyers ask before talking to sales
- the AI’s answer (and what sources it cites)
- where the AI is wrong, vague, or missing your POV
- what content asset would fix that (page update, FAQ block, comparison page, video, community post, review campaign, etc.)
One especially high-stakes gap from the transcript: pricing. If buyers ask an LLM about pricing and your site is vague, the model may fill in blanks unpredictably. Decide intentionally how much to disclose and where to anchor the narrative.
Don’t forget the oldest strategy in the book: own your audience
The most timeless point in this whole thing reiterates that anything old is new again. There’s some comfort in that stability! When you own the audience relationship, you reduce dependence on external platforms that can change the rules overnight.
Practical options, scaled by effort:
- a newsletter (now easier to sustain with AI assistance)
- a community hub (including a strategically managed subreddit, if relevant)
- events and real human connection (especially valuable as digital spaces become more AI-mediated)
- serialized content (weekly digests, resource kits)
What to do this week: a quick-start GEO checklist
If you want a non-overwhelming starting plan:
- Audit traffic + conversion integrity in GA4 (fix goals, dedupe conversions).
- Pick 10 pages: 3 protect, 5 pivot, 2 promote.
- Rewrite the first 200–300 words on pivot pages using SCQA.
- Add question-driven H2s + short FAQ blocks that map to LLM phrasing.
- Run reverse-engineering prompts and document the gaps (especially pricing + geography + comparisons).
Choose one “rented land” channel to commit to for 30 days (LinkedIn, YouTube, Reddit, reviews, analyst motions—pick one).
The mindset shift: SEO to GEO
SEO isn’t dead. Google isn’t dead. But the monopoly on attention is weakening, and “click-to-learn” is being replaced by “summary-to-decide.” GEO is what you do when your job isn’t just to rank. It’s to become the source AI systems trust enough to cite across the entire digital ecosystem.
