The rules changed. Here’s the new playbook. 30 principles for building startups in an AI world — what worked before, what broke, and what founders need to do now. Not theory. Tactics that ship.

New principle every Monday

AI commoditised point solutions — founders need a specific high-value wedge that earns the right to expand, rather than building horizontal AI tools that incumbents bundle as features within a product cycle. You were told to start with a tiny niche, a single feature done ten times better. Turns out, that advice is now a trap. The defensible wedge isn’t a feature anymore; it’s the entire workflow you make obsolete.

Why This Mattered Before

The conventional wisdom was to start narrow and deep. Paul Graham’s essays hammered this home: own a tiny piece of the market completely before you even think about expanding. And it was right. Building anything, even a single, focused tool, required serious engineering effort. That effort was your moat.

A startup could build a defensible business by solving one problem exceptionally well because incumbents were too slow or unfocused to bother competing on such a small feature. You could earn passionate users, build from a position of strength, and expand from there. This playbook worked for years. Until it didn’t.

Subscribe now

The Graveyard

I haven’t seen a single, high-profile company vaporised overnight because its wedge became an AI feature. It’s not that dramatic. It’s a slower, more painful death.

In the cohorts I’ve worked with since 2023, I’ve watched dozens of founders pitch some variation of ‘AI for X’—AI for summarising legal documents, AI for writing marketing copy, AI for generating social media posts. They build a slick UI over a foundation model. They get some early interest. Then, within six months, Google, Microsoft, or the dominant SaaS player in their market announces the exact same capability, bundled for free into a platform with a million existing users. The startup’s traction stalls, the pipeline dries up, and they quietly pivot or wind down.

What AI Actually Changed

Here’s the thing. The barrier to building a point solution has collapsed. What used to take a small engineering team a quarter to build can now be replicated by a single developer using an API in a weekend. Incumbents are not asleep at the wheel; they are weaponising this.

The wedge is no longer about doing one thing better—it’s about automating an entire workflow that incumbents serve with multiple products or clunky integrations. Large platforms are in a feature-bundling arms race. Look at the low-code space. Appian saw its stock jump after announcing strong results driven by its new AI-powered solutions in late 2025. Their AI-inclusive tier revenue more than doubled quarter-over-quarter. They didn’t launch a new product; they wove AI directly into their existing platform, instantly upgrading their value proposition and boxing out any startup trying to sell a standalone “AI for process automation” tool.

The New Playbook

Your goal is no longer to build a better tool. It’s to build a better, more automated workflow that creates high switching costs.

1. Map the entire workflow, not the task. Don’t build an AI to summarise meeting notes. Build the system that schedules the meeting, records it, generates the summary, creates action items in the user’s project management tool, and drafts the follow-up email. The value isn’t in one step; it’s in the seamless automation of all of them.

2. Go vertical, not horizontal. A horizontal tool like “AI for copywriting” competes with everyone. A vertical tool like “AI for generating Phase I Environmental Site Assessment reports” for commercial real estate has a built-in moat. It requires specific domain knowledge, proprietary data structures, and a deep understanding of a workflow that big tech will never bother to learn.

3. Build for deep integration. Your product should become the central nervous system for a specific business process. Use tools like Retool or Airtable not just as a database, but as a platform to build custom, AI-powered internal applications that are deeply embedded in your customer’s operations. The harder you are to rip out, the more defensible your business is.

4. Own the messy ‘last mile’. Fully automated AI is a commodity. A workflow that is 95% automated but requires expert human-in-the-loop validation for the final 5% is a powerful moat. You aren’t just selling software; you’re selling a guaranteed outcome. This is something incumbents with a purely self-serve model struggle to replicate.

Companies To Watch

• Harvey — San Francisco, USA — AI platform that automates full legal workflows from contract review to case research and litigation strategy. Perfect example of the article’s thesis - avoided the point solution trap by automating entire legal workflows instead of building narrow AI tools that incumbents like Westlaw could easily bundle. Shows how to build a defensible wedge that makes existing workflows obsolete.

• Sierra — San Francisco, USA — Conversational AI platform that powers full customer service workflows, from query resolution to retention plays. Exemplifies the new playbook by automating complete customer service workflows instead of building narrow chatbot features that platforms like Zendesk bundle for free. Demonstrates how to own the entire workflow rather than a single touchpoint.

• Adept — San Francisco, USA — AI agents that automate complex enterprise workflows like sales outreach and data entry via natural language actions. Shows the evolution from point solutions to workflow automation - instead of building narrow AI tools that get commoditised, they’re creating AI agents that handle entire business processes. Perfectly illustrates the shift from better tools to obsolete workflows.

• Ema — Mountain View, USA — Universal AI employee platform that handles multi-app workflows across HR, finance, and IT. Demonstrates the article’s core thesis by avoiding the commoditisation trap of single-purpose AI tools. Instead of building narrow automation features, they’re replacing entire cross-functional workflows that incumbents can’t easily replicate with simple bundling.

• UiPath — New York, USA — End-to-end robotic process automation platform that orchestrates AI agents to replace entire back-office workflows from invoice processing to compliance auditing. Pre-AI example of the thesis in action - avoided the point solution trap by automating complete workflows before incumbents could bundle individual features. Shows the enduring value of workflow-level thinking that’s now even more critical in the AI era.

• Zapier — Sunnyvale, USA — No-code automation platform that chains AI-powered apps into custom workflows, obsoleting manual integrations across sales, marketing, and ops. Illustrates how workflow-level thinking beats point solutions - while individual integrations get bundled by platforms like Salesforce, Zapier’s cross-tool workflow automation remains defensible. Shows the power of owning the entire process rather than individual connections.

• Glean — Palo Alto, USA — Enterprise search and knowledge AI that replaces fragmented internal search workflows with instant, contextual answers. Avoids the commoditisation of individual search features by owning the entire knowledge discovery workflow. While point search tools get bundled into platforms like Slack, Glean’s comprehensive approach to enterprise knowledge creates higher switching costs.

• Casetext — San Francisco, USA — AI legal research platform (CoCounsel) that automates full case prep workflows. Success story that validates the thesis - avoided building narrow legal research features and instead automated complete case preparation workflows. The $650M acquisition by Thomson Reuters shows the value of workflow-level thinking over point solutions.

The Warnings

The Thin Wrapper Trap. If your entire product is a slightly nicer UI on top of a third-party LLM model, you don’t have a business. You have a feature that’s waiting to be copied. I’ve watched founders raise money on this premise, only to see their churn skyrocket the moment their customers’ existing tools get the same AI capability for free.

Mistaking a Feature for a Workflow. Founders fall in love with a clever AI-powered feature. They build a company around it. But customers don’t buy features; they buy solutions to whole problems. If your “wedge” only solves one small part of their painful, multi-step process, you’ll be Sherlocked by the platform that owns the rest of that workflow.

The Bottom Line

Stop selling a smarter hammer. Start selling the automated workshop, with you as the foreman who guarantees the job gets done.

Part of Startup Principles for an AI World — 30 principles for building in the new era. New issue every week.

Sources:

• Appian Q4 2025 Earnings Call Transcript

• Paul Graham, ‘The Best Way to Start a Startup’

Part of Startup Principles for an AI World — 30 principles for building in the new era. New issue every week.

Arthur is the AI native startup operating system I’m building in public — not hype, but a system that turns input into structured execution and tracks founder progress. If you want to follow or use it, it’s open for early access.
→ Access Arthur

Enjoy and hope to see you soon.
❤️ & ✌🏼

Derek

Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.

If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.

Please let me know what you think of it, love a feedback loop 🙏🏼

• ❤️ Love it, keep them coming!

• 👍 Great, always a good read.

• 😊 Enjoyable, but let’s see some variety.

• 👎 Okay, seen better.

• 🛑 Get a different job.
  
  Subscribe below and follow me on LinkedIn or Twitter to never miss an update. 

Why This Mattered Before

You were told to focus on growth first. Get users, build something they love, and figure out how to make money later. The mantra was “make something people want.” Revenue was a secondary concern to engagement.

I used to hear this in every single office hours session. Founders would say “we’ll figure out monetisation once we have traction.” And for years, that actually worked. For traditional software, the marginal cost of serving another user was near zero. Once the code was written, adding the ten-thousandth user cost little more than the tenth. You could afford to give it away because it cost you almost nothing.

The Graveyard

The public post-mortems for AI startups that burned to death subsidising free users haven’t been written yet. The market is moving too fast. But I can tell you what I’ve seen.

In the cohorts I’ve worked with since 2024, the pattern is painfully consistent. A founder builds a clever wrapper around a powerful model. They launch a free beta and get a rush of sign-ups. They see user numbers climb and think it’s traction. But their OpenAI bill climbs faster. I’ve watched teams torch a meaningful chunk of pre-seed in a single quarter on inference, then lose the vast majority of ‘users’ the moment they introduced pricing. They’re left with no money, no paying customers, and a product no one actually valued enough to pay for.

What AI Actually Changed

Here’s the thing. AI turned whole product categories into near-zero marginal-cost businesses overnight. Shipping, support, onboarding, even chunks of delivery are now automated. So charging £20/month doesn’t prove you’ve got a business—it might just prove you’ve got cheap compute and a Stripe account.

Every time a user interacts with your product, it triggers an API call to OpenAI, Anthropic, or your own self-hosted model. That is a real, metered, non-zero cost. It shows up as a line item on your monthly bill. Your Cost of Goods Sold (COGS) is no longer just server hosting; it’s a direct function of user engagement. An active free user is not an asset, it’s a liability.

Giving away an AI product for free is not like giving away software. It’s like giving away coffee. The first cup isn’t free to you, and the thousandth cup certainly isn’t. You’re paying your supplier for every single one. If you don’t charge the customer, you’re just running a very expensive charity

.

The New Playbook

1. Charge from Day One. This isn’t just about covering costs. It’s the fastest way to validate that you’ve built something of actual value. Free users will tell you they love it; paying users prove it. Price on the value you create. Just last week on Hacker News, I saw a tool that negotiated a hospital bill down by thousands; people will happily pay a percentage of that saving.

2. Know your Unit Economics. You must know your exact cost-per-user and cost-per-query from the start. Instrument everything. If it costs you £0.50 in tokens to serve an average user session, and you’re charging £10 a month, you know your ceiling for engagement. If you don’t track this, you’re flying blind into a cash-flow crisis.

3. Use Strategic Freemium, Not a Free-for-all. A free tier can still work, but it must be a marketing tool, not the core product. Offer a low-cost version that uses a cheaper, less powerful model (e.g. Haiku instead of Opus). Or, offer a severely rate-limited or feature-gated experience. The goal of the free tier is to demonstrate value so clearly that users upgrade. It’s a product tour, not an open bar.

4. Align Billing with Costs. Usage-based billing is your friend. By charging per-credit, per-report, or per-token-bundle, you ensure your revenue scales directly with your single biggest cost centre. This protects your margins and forces high-volume users to become your most valuable customers, not your biggest expense.

Subscribe now

The Warnings

The Pivot Trap. I’ve watched founders realise too late that their free users and their paying customers want different things. The person playing with your AI toy at 2am is not the same person who’ll expense a £200/month SaaS tool. If you build for free users first, you’re building the wrong product. The Timing Trap. You think you’ll introduce pricing ‘once we have more features.’ But your free users are training you to build features that don’t drive revenue. Every month you delay is another month you’re learning the wrong lessons.

The Bottom Line

Charge early. Cover your costs. Prove your value. Don’t fund hobbies with venture capital.

Part of Startup Principles for an AI World — 30 principles for building in the new era. New issue every week.

Arthur is the AI native startup operating system I’m building in public — not hype, but a system that turns input into structured execution and tracks founder progress. If you want to follow or use it, it’s open for early access.
→ Access Arthur

Enjoy and hope to see you soon.
❤️ & ✌🏼

Derek

Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.

If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.

Please let me know what you think of it, love a feedback loop 🙏🏼

• ❤️ Love it, keep them coming!

• 👍 Great, always a good read.

• 😊 Enjoyable, but let’s see some variety.

• 👎 Okay, seen better.

• 🛑 Get a different job.
  
  Subscribe below and follow me on LinkedIn or Twitter to never miss an update. 

Why This Mattered Before

The logic was sound: identify an emerging channel, build deep expertise in its mechanics, and create a product that makes that channel work better than anyone else could manage in-house. The channel was the product. Master it, and you’d own the customers who depended on it.

Discord proved this brilliantly. They didn’t just build another chat app — they built the communication infrastructure for gaming communities. Low-latency voice, robust server management, community moderation tools that understood how gamers actually organised themselves. By 2021, they’d raised nearly $1 billion and hit a $15.2 billion valuation. The product was inseparable from the channel it served.

Intercom did the same for customer messaging. They weren’t selling software — they were selling the channel itself. Chat widgets, in-app messages, email campaigns, all unified into one platform that became the primary way SaaS companies talked to users. By 2024, they were generating $343 million in revenue with a $1.3 billion valuation. The channel was the entire value proposition.

Both companies succeeded because the technical complexity and channel-specific knowledge created genuine barriers. You couldn’t just bolt on Discord’s voice quality or Intercom’s message orchestration as an afternoon project.

The Graveyard

In the cohorts I’ve worked through since 2022, I’ve watched this pattern destroy dozens of early-stage companies. The most common casualty: startups building AI-powered customer support tools as standalone products. They’d raise a seed round on the promise of “intelligent chatbots that understand context” or “automated ticket routing that learns from your team.” Solid products. Real customer validation. Then Meta, Zendesk, and Salesforce started bundling comparable AI features directly into their core platforms — often at no additional cost.

The breaking point wasn’t that the AI got better. It was that the incumbents already owned the channel. They had the customer data, the existing integrations, the sales relationships. Adding an AI layer was trivial compared to the distribution advantage they already held. The startups were selling a feature. The incumbents were giving it away to protect their platform position.

I watched one founder in a 2023 cohort spend eighteen months building a sophisticated AI email outreach tool. Beautiful product. Real engagement metrics. Dead within six months of HubSpot shipping their AI email writer. The channel — email outreach — became a commodity the moment the platform players decided to compete.

Subscribe now

What AI Actually Changed

The specific mechanism: AI collapsed the technical barriers that made channel expertise defensible. What used to require deep domain knowledge and months of custom development now takes an afternoon with the right API calls.

Outbound sales automation used to be hard. Tools like Outreach and SalesLoft built entire companies on sequencing, personalisation, and analytics. Then ChatGPT made it trivial to generate hundreds of personalised cold emails in minutes. Clay, Apollo, and a dozen others bundled AI writing directly into their platforms. The channel — cold email — stopped being a product and became table stakes.

Content generation followed the same path. Companies built businesses around “content marketing automation” or “SEO content at scale.” Jasper raised $125 million on this premise. Then every CMS, every social media tool, every marketing platform added AI writing. WordPress plugins, LinkedIn’s built-in post composer, HubSpot’s content assistant — all free or near-free. The channel commoditised in eighteen months.

Referral mechanics and growth loops used to require serious engineering. Tools like Viral Loops and GrowSurf sold the infrastructure to run referral programmes. Now? Most CRM and marketing automation platforms ship with AI-powered referral tracking and optimisation built in. The channel became a feature, not a product.

The pattern repeats: AI made the technical execution cheap enough that platform players could bundle it without thinking twice. The channel stopped being defensible the moment it became a commodity capability.

The New Playbook

1. Build distribution into the product experience, not around it. Your product should create its own distribution as a natural byproduct of use. Loom did this — every video shared was a mini demo of the product itself. The distribution was embedded in the core value, not bolted on as a separate channel strategy. If your “channel” can be replicated by someone else’s feature update, it’s not a product.

2. Own proprietary data that makes the channel work better for your specific use case. If you’re building on top of a channel, the defensibility comes from data incumbents can’t easily replicate. Scale AI didn’t just build labelling tools — they built proprietary training data that made their tools uniquely effective for specific AI use cases. The channel (data labelling) was commoditised, but the data moat held.

3. Make the AI create network effects, not just automate tasks. Tools like Notion AI or Figma AI get more valuable as more people use them because they learn from collective usage patterns within teams. The AI isn’t just a feature — it’s the mechanism that creates lock-in. If your AI just automates what one person does faster, you’re building a feature. If it gets smarter as your customer’s team grows, you’re building a moat.

4. Ship faster than incumbents can bundle. The only way to win the feature race is to not be in it. Move so fast that by the time a platform player copies your current feature set, you’re already three versions ahead. Linear does this relentlessly — they ship meaningful updates every week, staying ahead of Jira and Asana’s AI features by sheer velocity. Speed becomes the moat when technical complexity isn’t.

The Warnings

The “AI wrapper” trap. I’ve watched too many founders build what they think is a product but is actually just a thin UI on top of OpenAI’s API. They raise a seed round, get early traction, then watch their entire value proposition evaporate when OpenAI ships a native interface or when their customers realise they can just use ChatGPT directly. The cost: six to twelve months of runway burned before the pivot. If your product is just “ChatGPT but for [vertical],” you’re not building a company — you’re renting time until the channel owner notices.

The “bundling is coming” delusion. Founders convince themselves they have eighteen months before the big platforms bundle their feature. They’re wrong. In the cohorts I’ve tracked, the median time from “this could be a feature” to “this is now a free feature” dropped from thirty-six months pre-2020 to fourteen months post-ChatGPT. The acceleration is real. If your entire moat is “we do this one channel thing really well,” you don’t have years — you have quarters. Plan accordingly or die surprised.

The Bottom Line

The channel isn’t the product anymore — the product is what makes you impossible to unbundle even after the channel gets commoditised.

Companies To Watch

These companies built their core product around owning a distribution channel. Each one faces the same bundling pressure described above — watch how they respond.

Aspire — Singapore — Influencer discovery and campaign management platform. Built their entire product around owning the influencer distribution channel. TikTok Shop and Meta are now building native creator marketplaces. The channel they own is becoming a platform feature.

PartnerStack — Toronto, Canada — Partner and referral programme infrastructure for SaaS. Their moat is being the dedicated layer for partner distribution. As HubSpot and Salesforce build native partner tracking, the standalone case gets harder to make.

Yotpo — Tel Aviv, Israel — UGC, SMS and email acquisition platform for e-commerce. Built a $1.4B valuation on owning those channels for Shopify merchants. Shopify’s own marketing tools are eating directly into that territory.

Impact.com — Santa Barbara, USA — Affiliate and partnership automation platform. Owns the affiliate channel for enterprise brands. AI-powered attribution and fraud detection are now standard features in major ad platforms.

Talon.One — Berlin, Germany — Promotion and loyalty engine. Built around owning the promotions distribution layer. CDPs like Segment are bundling similar capabilities as native features.

Referral Rock — Washington DC, USA — Referral programme infrastructure. Their pre-AI moat was referral loop mechanics as a standalone product. Growth platforms now ship this as table stakes.

Part of Startup Principles for an AI World — 30 principles for building in the new era. New issue every week.

Arthur is the AI native startup operating system I’m building in public — not hype, but a system that turns input into structured execution and tracks founder progress. If you want to follow or use it, it’s open for early access.
→ Access Arthur

Enjoy and hope to see you soon.
❤️ & ✌🏼

Derek

Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.

If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.

Please let me know what you think of it, love a feedback loop 🙏🏼

• ❤️ Love it, keep them coming!

• 👍 Great, always a good read.

• 😊 Enjoyable, but let’s see some variety.

• 👎 Okay, seen better.

• 🛑 Get a different job.
  
  Subscribe below and follow me on LinkedIn or Twitter to never miss an update. 

Why this exists (and when it matters)

Search changed quietly — but not universally.

Your website is no longer read only by people. It is read, summarised, compressed, and repeated by machines — before a human ever sees you. When those machines misunderstand you, your market misunderstands you. No amount of polish fixes that.

That said, GEO is not equally important for every startup.

If you sell enterprise software through relationships, GEO won’t replace sales. If your growth is dominated by app‑store algorithms, GEO won’t outrank Apple. But if discovery, explanation, comparison, or recommendation matter to your business — GEO is no longer optional.

GEO is not hype. It is alignment under compression.

This guide shows:

• where GEO actually sits (and where it doesn’t),

• what it practically changes for founders,

• and how to implement it without agencies, theatre, or over‑engineering.

Is GEO worth your time? (Decision tree)

Use this page before you read anything else.

Step 1 — How do people find you today?

• Mostly through inbound search, content, or recommendations → go to Step 2

• Mostly through outbound sales or relationships → GEO is secondary, not critical

• Mostly through app stores or closed platforms → GEO has limited impact

Step 2 — Do people ask machines questions like these about your category?

• “What tool should I use for ___?”

• “What’s the best way to ___?”

• “Explain ___ and give examples.”

• Yes → go to Step 3

• No → GEO is useful but not urgent
  
  

Step 3 — Can a machine accurately describe you in one sentence today?

• Yes → GEO will amplify clarity

• No → GEO will expose misalignment (this is where the leverage is)

Step 4 — Do you care if AI systems cite or recommend you?

• Yes → GEO is worth your time now

• No → stop reading and revisit later

Rule of thumb: If AI‑mediated discovery or explanation touches revenue, trust, or pipeline — GEO matters.

1) What GEO is

Generative Engine Optimisation (GEO) is the practice of making your business legible to large language models.

Not rankable. Legible.

Important distinction: GEO is not creative or persuasive copywriting. It is literal clarity under compression — writing so machines can extract, summarise, and repeat you without interpretation.

Where SEO tried to win positions, GEO tries to win summaries.

Machines now answer questions like:

• “Who should I use for X?”

• “What’s a good tool for Y?”

• “Explain Z and cite examples.”

They don’t browse like humans. They extract → compress → cite.

If your site is vague, contradictory, or bloated with marketing language, machines cannot safely quote you. When they can’t quote you, they don’t recommend you.

GEO’s job: ensure that what you actually do is what machines repeat about you.

GEO is not

• A replacement for SEO

• A traffic guarantee

• A magic file or plugin

• Creative copywriting

GEO is

• Message integrity

• Machine trust

• Citation readiness

• Literal clarity under compression

If SEO asked “can you be found?”, GEO asks “can you be understood in one sentence?”

2) Where GEO sits

Think in layers:

Reality (what you actually do)

↓

Website (what you say you do)

↓

Machine understanding (what LLMs extract)

↓

LLM answers (what gets repeated)

↓

Human decisions (trust, traffic, sales)

Most founders obsess over the website layer and ignore the machine layer. That gap is now costly.

GEO sits between your website and machine understanding. Its job is to remove ambiguity before compression happens.

3) How machines read websites

The most common GEO failure (run this first): Run a robot summary on your homepage, about page, and one product/feature page. If you get three different one-sentence answers, machines don’t know who you are yet. Fix consistency before you optimise anything else.

Machines do not:

• admire your design

• infer intent

• give you the benefit of the doubt

They do:

• look for explicit statements

• prioritise consistency

• penalise contradictions

• prefer quotable structure

Practically, that means:

• short, literal sentences win

• specifics beat adjectives

• proof beats promise

• one truth beats five angles

Use case: Two SaaS sites offer “AI‑powered insights”. One states:

“We analyse customer conversations to reduce churn by identifying at‑risk accounts.”

The other says:

“We deliver next‑generation intelligence for modern teams.”

Only one is quotable. Only one is citable.

4) Machine‑readable setup (the boring bits that matter)

This is not the work — it’s the plumbing. But bad plumbing breaks everything.

Baseline (must‑have)

• robots.txt — allow legitimate crawlers; block junk

• sitemap.xml — declare what exists

• Schema.org (JSON‑LD) — explicitly label who you are, what you sell, FAQs, reviews

This is the proven foundation.

GEO‑native layer (recommended)

• Canonical AI page (/ai or /about/ai) — a single, quotable truth source

• llms.txt (emerging) — guidance for LLM crawlers (optional hedge, not core dependency)

Rule: if you only do one thing, create the canonical AI page.

What goes on it:

• one‑sentence description

• ICP

• core problem

• outcome

• proof

• primary CTA

No marketing. No storytelling. Just truth.

Canonical AI Page Template (copy/paste)

Create a simple page at /ai or /about/ai. This page exists for machines first, humans second.

Page title:

What [Company Name] Does (for AI Systems)

1) One-sentence description

We [do X] for [specific ICP] so they can [measurable outcome].

2) Who it’s for (ICP)

Primary users are [role] at [company type / stage].

3) Core problem

They struggle with [clear, specific problem].

4) Outcome

Our product helps them [specific result], typically [time / cost / performance change].

5) Proof

Evidence includes [metric / named customer / case outcome / benchmark / demo link].

6) What we are not

We are not [adjacent category / common misclassification].

7) Primary CTA

Next step: [Book a demo / Start trial / View docs]

Example (fictional)

One-sentence description We analyse B2B sales calls for SaaS teams so they can reduce churn and shorten deal cycles.

ICP Revenue and customer success teams at B2B SaaS companies (10–500 employees).

Core problem Teams don’t know which conversations signal churn or stalled deals.

Outcome Customers identify at-risk accounts 30–60 days earlier and improve close rates.

Proof Used by 120+ SaaS teams. Average churn reduction: 18%.

What we are not Not a generic call recorder or CRM.

CTA Book a demo.

Rule: If this page and your homepage tell different stories, machines will trust this page more.

5) The 5 GEO rules

Rule 1: One‑sentence truth

If a machine can’t summarise you in one sentence, neither can a human.

Rule 2: Language consistency

Use the same words everywhere. Synonyms confuse machines.

Rule 3: Proof over promise

Metrics, customers, results. Not vibes.

Rule 4: Hierarchy matters

Hero → explanation → proof → action. Every time.

Rule 5: Be quotable

Short sentences. Clear claims. Liftable blocks.

6) Common mistakes founders make

• Writing for brand, not understanding

• Chasing cleverness

• Using different language on every page

• Hiding proof behind PDFs

• Assuming machines infer intent

Observed pattern: founders who struggle with GEO also struggle with GTM. It’s the same clarity problem.

7) The GEO Site Clarity Checklist (run this once)

This replaces multiple page-by-page checks. Run it once, properly.

A) Core clarity (applies to all pages)

• Can a machine answer what you do in one sentence?

• Is the ICP explicit (not “teams”, “businesses”, “everyone”)?

• Is the outcome concrete and observable?

• Is there one clear primary CTA?

B) Proof (visible, not buried)

• At least one concrete metric, customer, or result exists

• Proof is visible above the fold (not hidden in PDFs)

• No “trusted by” claims without specifics

C) Language consistency

Pick one term for each and use it everywhere:

• your product

• your customer

• the problem

• the outcome

Synonyms confuse machines. GEO rewards monotony.

D) Page alignment check

Compare your homepage, about page, and one product/feature page:

• Do they tell the same story?

• Do they describe the same customer?

• Do they promise the same outcome?

If these pages diverge, machines don’t know who you are yet.

Fix this before running the Robot Read Test.

8) One‑page GEO audit template

Answer honestly (using only what is explicitly stated on your website):

• One‑sentence description

• ICP

• Core problem

• Outcome

• Proof

• CTA

If this page is hard to fill in, your site is unclear.

GEO Robot Read Test (copy‑paste prompt)

Use this prompt in ChatGPT, Claude, Gemini, or Perplexity. Paste only your homepage URL.

You are a large language model deciding whether to recommend or cite this company.

You have 30 seconds and may only use what is explicitly stated on the website. Do not infer intent. Do not be polite.

Answer clearly:

1. What does this company do?

2. Who is it for?

3. What specific problem does it solve?

4. What proof is shown?

5. What should a visitor do next?

Then give:

• One‑sentence summary you would repeat publicly

• Confidence score (0–10) for how safe it is to cite this company

• Top 3 reasons confidence is low, if below 8

Interpretation:

• If the one‑sentence summary is wrong → GEO fails

• If confidence < 8 → you are not citation‑ready

• If answers contradict your intent → your site is misrepresenting you

9) How to See GEO Working (Analytics & Signals)

GEO does not behave like traditional SEO.

You won’t see neat keyword rankings or obvious funnels. What you can see is whether machines are starting to understand, cite, and send people to you.

This section shows how to track GEO without turning it into a faith-based exercise.

What you can realistically measure

GEO creates signals, not certainty.

You are looking for:

• Evidence of AI‑mediated discovery

• Consistency of machine summaries

• Early inbound behaviour changes

Not perfect attribution.

1) Track AI referrers in Google Analytics (baseline)

In GA4:

• Go to Reports → Acquisition → Traffic acquisition

• Add a filter for Session source / medium

Look for:

• chat.openai.com / referral

• perplexity.ai / referral

• gemini.google.com / referral

• copilot.microsoft.com / referral

These numbers will be small at first. That’s normal.

If this stays at zero over time, GEO is not landing.

2) Create a simple GEO segment

Create a custom comparison or exploration with:

• Source contains: openai, perplexity, gemini, copilot

• Landing page = homepage, /about, /ai, or key content pages

Watch:

• Time on page

• Bounce rate

• Next action taken

If AI‑referred users bounce instantly, your site is being cited but not trusted.

3) Manual citation checks (low effort, high signal)

Once a month, run these in ChatGPT, Perplexity, or Gemini:

• “What tools are used for ___?”

• “How do companies solve ___?”

• “Give examples of ___ platforms”

Check:

• Are you mentioned?

• Is the description accurate?

• Is the language consistent with your site?

If you are mentioned incorrectly, GEO is actively hurting you.

4) The consistency test (most founders miss this)

Run the Robot Read Test on:

• Homepage

• About page

• One product or feature page

If each produces a different one‑sentence summary, machines don’t know who you are yet.

Fix consistency before chasing traffic.

5) What not to obsess over (yet)

• Exact attribution from LLMs

• Immediate revenue impact

• Daily dashboards

• Vanity mentions

GEO compounds quietly. Early clarity matters more than early volume.

Rule of thumb

If, over 30–60 days, you see:

• AI referrers appear (even small)

• Machine summaries converge

• Fewer “what do you actually do?” conversations

→ GEO is working.

If not, your site is still ambiguous.

Final word

GEO is not optimisation. It’s alignment under compression.

The founders who win won’t shout louder. They’ll be easier to understand — by machines first, humans second.

Fix clarity. Everything else follows.

For the ❤️ of startups

✌🏼 & 💙
Derek

Insights by Fusion42

We cut through the noise of the startup and investor ecosystem by delivering actionable knowledge, data-driven insights, and connections to make the startup world transparent and accessible. No fluff, no BS—just the insights you need to get **it done.

By DeReK WaTSoN

Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.

If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.

Please let me know what you think of it, love a feedback loop 🙏🏼

• ❤️ Love it, keep them coming!

• 👍 Great, always a good read.

• 😊 Enjoyable, but let’s see some variety.

• 👎 Okay, seen better.

• 🛑 Get a different job.
  
  Subscribe below and follow me on LinkedIn or Twitter to never miss an update. 

Read more