The Question Is the Terrain: From SEO to Discoverability
TL;DR: A widely shared thread by Apoorv Sharma opens with a question that should retire your entire organic dashboard: your buyer can discover you through Google, ChatGPT, Gemini, Reddit, G2, LinkedIn, YouTube, or a private Slack group — so why is your "organic strategy" still a spreadsheet of Google keywords? His answer — stop building an SEO strategy, build an AI-era discoverability system — is correct as far as it goes. But it's one layer short. Presence on surfaces isn't something you post your way into; it's something you train the surfaces to do for you. The unit of strategy is no longer the keyword or even the channel. It's the question. Own fifty of them everywhere they get asked, and you stop being a search result. You become the path of least resistance.
James here, CEO of Mercury Technology Solutions.
Hong Kong — September 2026
The SEO Map Expired
Somewhere in your company there is a spreadsheet of Google keywords. It is a map of a country that no longer exists.
Not because Google died — it didn't; one of our properties still pulls 333,000 referrals a week from it — but because the map's borders stopped matching the territory. The old journey was a pipeline: keyword → ranking → click → website → conversion. Clean, linear, ownable. Entire teams were built around owning the ranking.
The new journey is a swarm. The buyer discovers the problem on LinkedIn, asks ChatGPT how to solve it, googles the category, reads Reddit opinions, compares vendors on G2, watches a YouTube demo, consults a private community — and then, maybe, books the demo. Sharma's question cuts deep: which one of those was the "organic channel"? None of them. All of them. The category itself dissolved. And increasingly the swarm compresses: question → AI answer → shortlist, with the middle steps eaten alive.
The Alliance fortified Iserlohn because that's where the enemy was supposed to come through. The Empire walked through Fezzan. Your Google rankings are Iserlohn Corridor — magnificent, expensive, and defending a front that no longer decides the war. The buying journey just went through Fezzan, and Fezzan is a Reddit thread.
The Five Layers of AI-Era Discoverability
Sharma's thread is the best mainstream framing I've seen of what we've called SEVO — Search Everywhere Optimization — for years, and what the industry is now rushing to rebrand as GEO, generative engine optimization, as if the insight were new: stop treating SEO as the map, start building a discoverability system. His architecture, compressed:
DEMAND — start with buyer questions, not keywords. Mine call transcripts, support tickets, Reddit, comparison searches. "What should we use instead of Vimeo?" "Do we really need SOC 2?" "Why is Datadog so expensive?" His line deserves framing: keywords are one expression of demand, not demand itself.
SURFACES — each question has a natural habitat. "CRM pricing comparison" lives on Google + ChatGPT + G2. "How are RevOps leaders fixing attribution?" lives on LinkedIn + Slack + Reddit. Match the question to the surface; don't spray.
EVIDENCE — different questions require different proof. "Is this secure?" wants certifications, architecture docs, and customer proof — not another 1,500-word essay. "Which performs better?" wants benchmarks and data. Most content strategies die here, serving essays to questions that demand receipts.
PRESENCE — the matrix: your fifty highest-value questions on one axis, every surface on the other, each cell marked Strong / Weak / Absent. The gaps jump off the page. Dominating Google while absent from AI recommendations isn't a strategy; it's an Iserlohn.
CONVERSION — connect discovery to pipeline: where did you first hear, where did you research, what shortlisted you, what nearly stopped you. Not perfect attribution — understanding how discovery becomes preference becomes pipeline.
That's the diagnosis, and it's right. Now the layer he left out — the one that separates a deck about discoverability from a discoverability system.
The Missing GEO Layer: Surfaces Are Algorithms, and Algorithms Get Trained
The thread treats presence as something you build — show up on the surfaces, fill the matrix. Operators know it's something you train. Every surface that shapes your buyer's decision is a ranking algorithm with a memory:
Reddit threads don't just persuade humans — they're training corpus for the LLMs your buyers ask. Answer honestly there and you've planted a citation inside every shortlist prompt for the next two years.
G2 grids train the procurement comparison. YouTube demos are evidence the eval team finds without you in the room.
And the answer engines follow the rule I wrote down nine days ago — the Citation Rule: an answer engine will either cite you or replace you, and which one it does is decided by what your footprint *is*, not how well you rank.
Your homepage cannot outvote the internet, as Sharma puts it. Correct — so stop campaigning for your homepage and start feeding the voters. Every third-party surface that mentions you, benchmarks you, or answers a question with your product in it is a ranking factor you don't own but can influence. That is surface training, and it compounds: we've watched client LLM visibility lift 300%-plus and qualified consultations run four times baseline — not from posting more, but from making sure the surfaces buyers actually consult had no choice but to mention them.
Stop chasing the algorithm. Start training it — all of them.
The Fifty-Question War Game
Run it as a standing operation, not an annual offsite:
Inventory the fifty questions that actually shape your buyer's decision. Not five hundred keywords. Fifty questions, classified Sharma's way: problem, category, comparison, objection, implementation, pricing, recommendation.
Map presence honestly — Strong / Weak / Absent per surface. Include the surfaces that hurt: if Reddit hates you or G2 ranks you fourth, that's not a PR problem, that's a discovery problem.
Assign the right weapon per question, not the default blog: a landing page, a benchmark, a public calculator, an engineer answering on GitHub or Stack Overflow, a YouTube teardown, honest participation in the Reddit thread, original research that other people cite. These are Answer Assets — built once, cited everywhere.
Wire the feedback. New customers get four questions — first heard, researched, shortlisted, nearly stopped. Watch branded search, direct traffic, AI referrals, review-site referrals. The goal isn't attribution perfection; it's knowing which surfaces create qualified demand and where buyers vanish.
Quarterly, rerun the loop. Surfaces shift — today's Fezzan becomes tomorrow's Iserlohn. The matrix is the map; the cadence is the OODA loop.
Report It Like a CEO Cares
Kill the old dashboard metrics. "We published twelve articles" and "organic traffic +17%" measure effort, not position. Even "AI visibility 64%" is a vanity number until it's tied to questions. The report that matters answers six things, for your fifty highest-value questions: Where are we discoverable? Where are competitors stronger? Where are we recommended? What evidence wins the recommendation? Which surfaces create qualified demand? Where do buyers disappear?
That's an organic strategy a CEO can actually care about — because it reads like a battle map, because it is one.
The New Reality for SEO, GEO, and SEVO
SEO isn't disappearing. GEO isn't a fad. AI didn't kill organic — it multiplied the places it lives. Each is becoming one component — a corps in a larger army, still vital, no longer the war. The companies that internalize this early won't just rank in more places. Sharma's closing line is the whole thesis: they'll be harder to avoid when someone is deciding what to buy.
The Chinese has a term for the end-state: 天羅地網 — the net of heaven and earth, the inescapable lattice. That's what a discoverability system actually builds: fifty questions, every surface, every answer — and no matter where the buyer looks, you're already there, cited by someone they trust more than you.
Stop owning keywords. Start owning questions. The terrain moved. Meet it there.
Mercury Technology Solutions: Accelerate Digitality.
