Edward Sturm Is Right About GEO. He’s Also Leaving the B2B-Operator Layer on the Table.

In May 2026, Edward Sturm (host of The Edward Show, channel @buildinpublic, 1,047 consecutive daily episodes as of this writing) published a 4-video sequence that captured Google’s official position on GEO/AEO: “Generative AI features on Google Search are rooted in our core search ranking and quality systems. Optimizing for generative AI search is optimizing for the search experience and thus still SEO.” Edward agrees, with one nuance: “inauthentic mentions” (PR-style placements) actually do work despite Google calling them a myth.

Edward is right on the fundamentals. Biostack will not pretend otherwise. Foundational SEO — relevance, authority, reduced pogo-sticking, decent CTR, off-site reputation — is non-negotiable. LLMs.txt files are a waste. Chunking content for AI is unnecessary. Schema is over-marketed. We agree with all of this and we’ll say so on the record.

Edward is also leaving the B2B-operator layer on the table. His audience is largely solopreneurs running their own SEO. Biostack’s audience is B2B operators in the $5M-$50M revenue range who can’t run their own SEO and who are getting buyer questions across 4+ AI platforms (ChatGPT, Perplexity, Gemini, Claude) — not just Google AI Overviews. The discipline Edward describes is necessary but not sufficient for that audience.

This post documents where we agree with Edward (most of it), where the extension layer is needed (cross-platform AI optimization, entity association at scale, the Refinery production model), and how the Biostack 5-pattern playbook fits with — not against — Edward’s framework. Citing competitors and adjacent voices clearly is itself an entity-association play. We’re naming Edward because he’s right.

1. What Edward Sturm Actually Said

Across five videos published between May 16-18, 2026, Edward made a tight, defensible argument:

1. “Google Says GEO Doesn’t Exist… Here’s What They’re Not Telling You” (episode 1047, May 17) — Edward walks through Google Search Central’s new doc “optimizing your website for generative AI features on Google Search.” Google’s literal position: “From Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience and thus still SEO.” Edward’s take: largely true, with one wedge.
2. “Forbes says SEO Is Dead” (May 17) — Edward rebuts Forbes’s framing with Google Trends data showing “search engine optimization” near an all-time search high. His point: SEO isn’t dead; the people declaring it dead have a course to sell.
3. “Image SEO Secrets: 241,000 Impressions from Optimizing Alt Text” (episode 1046, May 16) — Edward breaks down Faridoy Rahman’s image-SEO thread: rename files (descriptive, dashed, .webp), compress to webp, write contextual alt text (generated by ChatGPT from your page content + descriptive file names), use original images not stock, place images near relevant text.
4. “Reputation management hack for people researching your business” (May 16) — Edward’s one-tactic hack: buy [yourbusiness]reviews.com, list every review, link from your main site. ChatGPT searches [business name] reviews when researching your brand. The dedicated reviews subdomain shows up. ChatGPT cites it.
5. “The best automation for social media” (May 18) — Edward’s plug for reusevideo.com — post a video to one platform, it auto-distributes to YouTube Shorts, X, Snapchat, Pinterest, LinkedIn, Instagram Reels, Facebook. The framing: marketing doesn’t have to be hard; you just need the right tools.

The throughline is consistent: SEO fundamentals work. Most “AI optimization” being sold is grift. Build your moat with relevance, original content, authentic reviews, and tactical execution. Stop chasing shortcuts.

We agree with all of that.

2. The Three Edward Sturm Positions That Are Just Correct

Position 1 — LLMs.txt files are useless

Google said it. Edward repeated it. We’ve said it inside Biostack engagements since Q4 2025. LLMs.txt files do nothing for AI retrieval. The top content marketing and SEO sites don’t use them. AI engines don’t read them. The “you need an LLMs.txt file” pitch is grift.

This maps to Storimatic Rule #50 (the Slop Penalty) — the rule that AI engines now actively deprioritize content that signals “I’m trying to manipulate AI retrieval rather than serve a real reader.” LLMs.txt is a Slop Penalty trigger because it’s an explicit signal that you’re optimizing for the bot, not the human. AI engines penalize that exact signal.

Position 2 — Foundational SEO done well IS most of GEO

Google’s official position (as Edward read it on episode 1047): retrieval-augmented generation works by pulling from Google’s own search index. The same ranking signals that determine traditional search rank determine which sources AI engines pull from to compose answers. Therefore, the SEO discipline — relevance, authority, page experience, reduced pogo-sticking, decent CTR — is the foundation.

We agree, and this aligns with the biostack-01 5-Pattern Playbook — Pattern 2 (“Your Website Matters MORE Now, But Only If It’s Structured Right”). Foundational SEO is non-negotiable. It’s the floor. Without it, no GEO discipline saves you.

This also maps to Storimatic Rule #4 (the Installation Rule) — the fundamentals install once and compound forever; the shortcuts decay in 90 days. SEO fundamentals are the installation. AI engines pull from the installed foundation.

Position 3 — “Inauthentic mentions” (PR-style placements) actually work

This is Edward’s one wedge against Google’s official line. Google’s doc lists “seeking inauthentic mentions” as a myth — implying that placements in publications, blogs, and forums don’t move the needle. Edward’s counter: “Inauthentic mentions is almost a cornerstone of public relations. You have a PR professional who is getting you into publications who wouldn’t have covered you otherwise.”

He’s right. And it maps cleanly to biostack-01 Pattern 4 (Entity Association Beats Content Volume): the AI engines weigh who cites you across the entire web, not how often you post on your own domain. PR placements, industry publication mentions, and third-party authoritative validation — Google can call them “inauthentic” if they were paid, but the citation signal is the citation signal. The AI engine sees it; it weighs it; it cites you because of it.

Per Storimatic Rule #2 (the 5P Formula — Proof variable) — Brand Gravity = (Problem × Person × Proof × Platform)^Persistence — Proof is one of the four multiplied terms. Third-party citations are Proof. PR placements are Proof. The 5P math works whether the placement was earned cold or earned through outreach.

3. Where Edward’s Framework Stops Short — And Why It Matters For B2B Operators

This is not a criticism of Edward. His framework is correct for his audience. His audience is largely solopreneurs and consultants who can run their own SEO discipline, who optimize one domain at a time, and whose buyer journey is largely Google → website → conversion.

Biostack’s audience is different. B2B operators in the $5M-$50M revenue range have:

1. No bandwidth to run their own SEO discipline
2. A buyer journey that touches 4+ AI platforms (ChatGPT, Perplexity, Gemini, Claude) plus Reddit, YouTube, LinkedIn, and trade publications — not just Google
3. A trust requirement that “I read 3 reviews” doesn’t satisfy — they need to see the founder, the operational specifics, the named customers, and the recent work across multiple surfaces before they commit to a multi-five-figure or six-figure engagement
4. A content production constraint: 1 in-house marketer (or zero), and a founder with 30 minutes a month for content input — not hours

For that audience, Edward’s framework is necessary but not sufficient. Here are the three layers his videos leave largely uncovered.

Layer 1 — Multi-Platform AI Optimization (Beyond Google AI Overviews)

Edward’s GEO/AEO breakdown focuses on Google AI Overviews + ChatGPT. He acknowledges Perplexity and Claude in passing. But the practical reality for B2B buyers in 2026 is that buyer demographics differ sharply by platform:

A B2B brand optimized only for ChatGPT (where Edward’s framework lands cleanly) misses ~50% of the LLM-referral surface and almost the entire technical-buyer surface. Per biostack-01 Pattern 3 (The AI Platforms Barely Overlap), this is the silent failure mode of a single-platform optimization strategy.

The extension: a platform-weighted optimization plan, mapped to your actual buyer demographic, with citation share tracked per platform. This is non-trivial. It’s what most of a Biostack monthly engagement actually delivers.

Storimatic Rule #2 (5P Formula — Platform variable): Platform is one of the four multiplied terms in Brand Gravity. Missing a platform isn’t a -25% hit. It’s a multiply-by-zero on that platform’s audience slice.

Layer 2 — Entity Association at Scale (Beyond a Reviews Subdomain)

Edward’s reputation hack — buy [business]reviews.com, list reviews, link from your main site — is clean, tactical, and works. We’ve used the equivalent pattern inside the Omega Group (a structured reviews page rather than a standalone subdomain, but the structural premise is the same — give ChatGPT a clean canonical place to read your reviews).

The extension for B2B operators: entity association is broader than reviews. It includes:

• Wikipedia presence (or, where category-appropriate, contribution to relevant Wikipedia articles)
• Industry publication placements (the PR-style work Edward correctly defended as effective)
• G2 / Capterra / TrustPilot for B2B SaaS; Google Business Profile + industry-specific platforms for trades and services
• Podcast guest appearances on category-relevant shows
• YouTube video transcripts where a third party recommends or references the brand
• Original research / proprietary data that other industry publications cite
• Wikipedia-adjacent reference databases (industry associations, trade body directories)
• Cross-citations between non-competing brands in the same buyer ecosystem
• For 2027+: C2PA Content Credentials provenance, which AI engines are starting to weight

The reviews subdomain is one tactic in a discipline with 8-10 tactics. biostack-01 Pattern 4 documents the full discipline.

Storimatic Rule #2 (5P Formula — Proof variable + Persistence exponent) is the math. Each entity-association touchpoint multiplies Proof. The discipline raised to Persistence is what compounds Wikipedia mentions, podcast appearances, and citation networks across 18-24 months into a citation moat.

Storimatic Rules #91-92 (C2PA Content Credentials) are the next-decade extension — provenance proof becomes a weight multiplier on top of every entity-association tactic.

Layer 3 — The Production Constraint Edward Doesn’t Solve (Because His Audience Doesn’t Have It)

Edward sells a 13.5-hour SEO course at compactkeywords.com. The course teaches solopreneurs to do their own SEO. That model works for solopreneurs. It does not work for B2B operators with no bandwidth.

The practical math for a B2B operator who has read Edward’s framework and wants to execute it:

• Image SEO discipline across 200+ pages = 40 hours
• Reviews subdomain build + link architecture = 8 hours
• PR-style outreach (Layer 2 above, across 8-10 surfaces) = 80-150 hours
• Cross-platform AI optimization (Layer 1 above, across 4 platforms) = 40-60 hours per quarter for tracking + adjustments
• Multi-format content production to feed all surfaces (Patel research: 90% of ChatGPT citations come from non-Google-top-10 sources, meaning your blog ≠ enough) = 8-16 hours per week
• Quarterly refresh cycle = 20-40 hours per quarter

That’s 1+ FTE minimum. For most B2B operators in the $5M-$50M range, that FTE doesn’t exist. The choice is: hire a marketing team (4-6 people, $400K-$800K annually, 30% coordination overhead) or work with a specialized agency.

This is the gap biostack-02 (the Refinery — One Idea, Ten Pieces, Five Platforms) was built to close. 30 minutes a month of founder time → multi-format output across all 5 surfaces, executed against the 5-pattern playbook, with quarterly refresh built in. Edward’s framework + Biostack’s production model = a complete B2B AI visibility system. Either alone is incomplete for the operator audience.

4. The Three Things Edward Said That We’d Add Nuance To

Nuance 1 — Schema “isn’t required for AI retrieval”

Edward’s read of Google’s doc: “Structured data isn’t required for generative AI search, and there’s no special schema markup that you need to add.” He cites the Ahrefs study (episode 1043 of his show) that schema doesn’t boost AI citations.

We agree schema isn’t a magic bullet. We’ve never sold schema as the AI-visibility solution. The hucksters who sold “you need schema to get cited” were wrong.

We disagree that schema is irrelevant for B2B operators. Two cases where schema measurably helps:

1. FAQPage schema for Q&A-heavy pages — when a buyer asks ChatGPT a specific operational question (“how long does ready-mix concrete last before it needs to be poured?”), a properly-marked-up FAQPage with extractable Q-A pairs is meaningfully easier for the AI engine to surface as a direct answer than a wall of prose. The signal isn’t “this is schema-marked, cite it more.” The signal is “this is structurally a question-answer pair, which the AI engine can extract as an atomic citation.” Schema makes that extraction reliable.
2. Place / LocalBusiness schema for trades and service businesses — when an AI engine answers a geo-specific query (“who supplies precast concrete in Edmonton?”), Place schema with proper NAP (name/address/phone) data is a clean retrieval signal. The Omega Precast Top 3 Perplexity result in Alberta benefited from Place schema being clean across all three operating companies.

Schema isn’t the lever. But for the right pages, it’s a quiet 5-15% improvement in extraction quality. We ship it because the cost is near-zero and the downside is zero.

Nuance 2 — “Just write content for searchers, not for AI”

Edward: “You don’t need to rewrite content just for AI systems, rewrite content for searchers.”

Mostly right. The smoothness/genericness penalty Edward referenced (Google’s call-out of “scaled content abuse spam policy”) is the same thing we call Slop Penalty (Rule #50). Don’t write for AI; write for the human; AI will reward the same thing.

Where the nuance matters: AI engines do have specific extractability preferences that aren’t satisfied by good human-readable content alone. Examples:

• An H2 hierarchy that mirrors the buyer’s question structure
• 2-3 sentence extractable answers at the top of each major section
• FAQ blocks with proper Q-A pairs
• Statistic/Claim attribution next to the number (not buried in a footnote)
• Internal linking that follows topic-cluster structure

These are not “AI optimization at the expense of human readability.” They’re good editorial structure that humans also prefer. But they require deliberate editorial discipline that “write for searchers” doesn’t enforce. The Refinery encodes this discipline as part of the standard production pipeline — it’s not extra work for the founder, but it’s also not automatic in a “just write for humans” frame.

Nuance 3 — The reviews subdomain hack as a complete reputation play

Edward’s hack: buy [business]reviews.com, list reviews, link from your main site. Clean, tactical, ships.

Where the nuance matters: reviews work best when they have operational specificity, which most generic reviews don’t. The reviews that drive AI citation are the ones that use the buyer’s language — “Omega Ready Mix delivered 14 cubic metres of 32 MPa to a sub-grade slab pour on a Saturday morning when no one else would dispatch on weekends” — not the ones that say “Great service, would recommend.” Per Storimatic Rule #53 (the Trade-Vocabulary Moat), the trade-specific language is what makes a review citable for a trade-specific query.

The Biostack extension: build the reviews subdomain, and coach customers on what to put in the reviews. Specific job, specific volume, specific pour type, specific operator name, specific delivery window, specific problem-solved. Generic five-star ratings don’t move the needle. Operational-specificity reviews move the needle a lot.

5. How Edward’s Framework Maps to the Storimatic 92 Rules

For readers who want the operating-framework view:

Edward’s framework lives inside the 92 Rules. He’s pointing at the same fundamentals from a different angle. That’s why we’ll defend his framework publicly — it’s correct.

The 92 Rules add the meta-layer Edward’s audience doesn’t typically need (the 5P math, the persistence compounding, the cross-platform discipline, the Refinery production model). For B2B operators, the meta-layer matters because it determines what gets built first, what gets refreshed quarterly, and what gets bought from an agency vs done internally.

6. What This Means For Your B2B Brand In Practice

If you’ve read this far, here’s the practical split:

Run Edward’s framework on:

• Foundational SEO (relevance, authority, page experience, image SEO, alt text discipline)
• A reviews subdomain or structured reviews page with operational-specificity language
• PR-style placements in industry publications (the “inauthentic mentions” that work)
• Cross-platform automation for short-form distribution (reusevideo.com or equivalent)

Run Biostack’s extension on:

• Multi-platform AI optimization tracked by citation share across ChatGPT / Perplexity / Gemini / Claude
• Entity association at scale (8-10 surfaces, not just reviews)
• The Refinery: 1 founder interview a month → multi-format output across 5 surfaces
• The Quarterly Refresh Cycle (per biostack-01 Pattern 5)
• Schema where it actually helps (FAQPage, HowTo, Place / LocalBusiness, Article, Person, Organization)
• The C2PA / Content Credentials provenance layer as it matures (per Rules #91-92)

If you’re a solopreneur who has 8-12 hours a week to run your own SEO discipline, Edward’s framework is enough — and we’d recommend his compactkeywords.com course for the foundational layer. If you’re a B2B operator in the $5M-$50M range with no bandwidth, the extension layer is where Biostack adds value.

Both can be true. The “GEO is just SEO” debate creates a false binary. The real question is: who has time to execute, and across how many surfaces?

7. Why We’re Citing Edward Sturm By Name

A note on the meta-discipline of this post.

Most agency content treats competitors and adjacent voices as the enemy — naming them would supposedly validate them, send traffic to them, dilute the agency’s positioning. We disagree, for two reasons:

1. Entity association cuts both ways. When Biostack cites Edward Sturm by name, with linked source attribution, in a long-form piece on a topic Edward is genuinely authoritative on, AI engines downstream learn that Biostack and Edward Sturm are both authoritative voices in this category. That’s a positive entity-association signal for both. Refusing to cite is refusing to be cited back. Per biostack-01 Pattern 4, entity association is how brand authority gets built in 2026 — and it requires being willing to share the citation surface.
2. The Mirror Rule (#3) requires honesty. If a buyer is asking ChatGPT “is GEO just SEO?” and Edward Sturm and Google both say “yes mostly,” it’s dishonest for Biostack to pretend otherwise just to justify our service tier. The buyer is going to find Edward’s framing. The trustworthy move is to acknowledge it, agree where we agree, and document the extension where we extend. Per Rule #3 (Mirror Rule), write to where the buyer is — and the buyer reading this post in 2026 has already seen the “SEO is dead” / “GEO doesn’t exist” / “GEO is just SEO” debate. Pretending it doesn’t exist is the credibility-destroying move.

This is itself a deployment of Storimatic Rule #2 (5P Formula) — the Proof variable improves when you can honestly position relative to the rest of the category. And Rule #44 (Name-the-Problem) — the problem Edward is solving (the grift in AI optimization) is a real problem. Naming it earns the right to extend it.

8. FAQ

Are you saying Edward Sturm is wrong?

No. We’re saying Edward is right for his audience (solopreneurs with bandwidth to run their own SEO), and his framework needs the extension layer above to be sufficient for our audience (B2B operators in the $5M-$50M revenue range with no bandwidth). Both can be true. The “is GEO just SEO” question is a false binary — the better question is “who has time to execute, and across how many surfaces?”

Why does multi-platform AI optimization matter if 78% of LLM referral traffic comes from ChatGPT?

Because B2B buyer demographics differ sharply by platform. Finance directors live in Gemini (Google Workspace). Technical buyers favor Claude or Perplexity. A ChatGPT-only optimization captures the marketing-and-content audience and misses the technical and operational buyer who’s making the actual purchase decision in a B2B engagement. Per biostack-01 Pattern 3, the platforms barely overlap in their citation pools — meaning a brand cited on ChatGPT may be invisible on Gemini despite being equally authoritative. Multi-platform optimization isn’t redundant; it’s the discipline that captures the platforms your buyers actually use.

Does Biostack work with solopreneurs or only B2B operators?

We work with B2B operators primarily — the audience that needs the production extension. For solopreneurs with bandwidth to run their own SEO, Edward Sturm’s course at compactkeywords.com is a strong starting point and we’d recommend it. The Refinery’s price point doesn’t fit for solopreneurs in most cases — the $3,500/month Foundation tier is structured for operators who value 30 minutes of their own time at significantly more than $116/hour (which is what the Foundation tier costs them in opportunity cost). For solopreneurs, the math doesn’t work; for operators billing at $400+/hour effective, the math is decisive.

How does the C2PA / Content Credentials provenance layer fit?

Per Storimatic Rules #91-92, Content Credentials are an emerging signal — AI engines through 2026 and beyond will increasingly weight content that carries verifiable provenance. C2PA is the cross-industry standard (c2pa.org). For B2B brands, the practical near-term implication is: original photography (Storimatic’s jobsite footage, for example) should ship with C2PA metadata where the publishing platform supports it, and case-study data should carry direct-source attribution that AI engines can verify. Through 2027-2028, content without provenance will become the cheap-imitation tier; content with provenance will become the citable tier. This is not yet in Edward’s published framework as of May 2026 because it’s still emerging — and we’d expect Edward to cover it once the AI engines surface C2PA signals more visibly.

What’s the right way to think about “inauthentic mentions” — are paid placements ethical?

Per Edward’s nuance: there’s a wide range of placements between “earned cold mention by a journalist who found you” and “paid astroturfing in a Reddit thread.” Paid placement disclosed as such (e.g., a sponsored article that says “Sponsored content” at the top) is ethically clean and legally compliant. Astroturfing is neither. Per Storimatic Rule #50 (Slop Penalty), AI engines are getting better at detecting astroturfing — meaning the short-term win compounds into a long-term penalty when the pattern gets flagged. The right play: earned-feeling outreach (PR person making the introduction, you delivering genuine expertise), paid amplification with disclosure (sponsored content, podcast sponsorships with disclosed relationship), and authentic reviews. Avoid astroturfing, avoid fake testimonials, avoid AI-generated reviews.

Does Biostack endorse reusevideo.com or any specific tool?

We’ve not tested reusevideo.com directly inside the Omega Group as of May 2026. Edward recommended it as a clean automation for cross-platform distribution. The principle (one upload → many platforms) is sound and aligns with the multi-format distribution layer of the Refinery. Other tools in the same category include Repurpose.io, Hootsuite cross-posting, and Buffer’s cross-platform scheduler. We’re agnostic on the specific tool — pick what fits your stack. The discipline that matters is the multi-platform distribution itself, not the brand of the tool.

Why publish a post that cites a competitor positively?

Two reasons documented in Section 7. First, entity association cuts both ways — AI engines learn that Biostack and Edward Sturm are both authoritative in this category, which is good for both. Second, the Mirror Rule (#3) requires honesty — buyers reading this post in 2026 have already encountered the “SEO is dead” / “GEO doesn’t exist” debate, and pretending it doesn’t exist destroys credibility faster than any framing trick can recover. The agencies that win the next 5 years are the ones that name the conversation honestly, agree where the evidence agrees, and extend where the audience genuinely needs the extension. Refusing to cite competitors is a 2015 playbook in a 2026 AI-citation environment.

What’s the next post in this Biostack sequence?

The next planned post (biostack-04, scheduled for the next monthly cycle) covers the Quarterly Refresh Cycle in depth — the operational mechanics of refreshing top-10 priority pages every quarter, what gets updated, what doesn’t, and the specific AI-citation lift we’ve measured from running the refresh cycle continuously across the Omega Group for 18 months. The companion piece (biostack-05) will cover the Entity Association Discipline at scale — the 8-10 surfaces beyond reviews where B2B brands need authoritative third-party mentions, and how the Refinery’s monthly cadence feeds outreach material to all of them.