AI Automation

AI-Generated Website Copy: Where It Helps Conversion and Where It Kills It

Every service-business owner now has, sitting in a browser tab, a tool that will write their entire website in the time it takes to make coffee. The temptation is obvious and the savings are real, and the result is that AI is being poured into service-site copy at scale, as of April 2025, 74.2% of new web pages contained AI-generated content (1). The reflexive debate, "should I use AI to write my site or not," is already over; almost everyone is, in some measure. The debate that actually matters, and that almost no one is having with any precision, is where AI helps your conversion and where it quietly destroys it. Because it does both, and the line between them is sharp. The same tool that can sharpen your structure and accelerate your drafts can also strip your site of the exact qualities that make a stranger trust you enough to hire you. Using AI everywhere is as costly a mistake as using it nowhere. This piece draws the line, the specific places AI-generated copy lifts conversion and the specific places it kills it, so you can use it as a scalpel rather than a firehose. The data is cited; the framework is mine.

Joshua Agonya Pi'Rwot

By Joshua Agonya Pi'Rwot

Founder, Business Growth Accelerator

Executive summary

AI can write your website in minutes. The question is not whether to use it, it's knowing exactly which words it should never write.

Section 1

The principle: AI is excellent at the median and dangerous at the distinctive

To know where AI helps and where it hurts, understand what it fundamentally is: a tool that generates from the average of everything that exists. This makes it superb at producing competent, conventional, median output, and structurally incapable of producing the distinctive, specific, and personal. That single fact predicts the entire map. Anywhere your conversion depends on competent execution of a known pattern, AI helps. Anywhere your conversion depends on being specifically, credibly you, AI hurts, because the averaging machine erases exactly the specificity that builds trust. The market data confirms the stakes. Despite 74.2% of new pages containing AI content, 86% of the top-ranking pages in Google remain human-written (2), and the consistent finding is that the winning formula blends AI's efficiency with human originality and judgment (2). The web is drowning in AI-generated median, which means the median no longer differentiates, and the parts of your site that must differentiate are precisely the parts AI must not write. AI is a brilliant intern and a terrible founder. Let it draft, structure, and accelerate. Never let it decide what you believe, claim what you've done, or speak in your voice about why you're different. The intern executes; the founder is the reason anyone hires you.

Section 2

Where AI-generated copy HELPS conversion

These are the jobs where competent execution of a pattern is the goal, and AI excels. Structuring and outlining. AI is genuinely excellent at organizing a page, proposing a logical section order, turning a messy brain-dump into a clean structure, identifying gaps. The architecture of a converting page follows known patterns, and AI knows them. First drafts to react to. A blank page is the enemy of momentum. AI-generated first drafts give you something to sharpen, and editing is faster than originating. The draft is raw material, not the deliverable. GEO and SEO formatting. This is a strong fit: AI is excellent at the answer-first structuring, question-shaped headings, and clean hierarchy that get content cited by AI engines, the very formatting that lifts AI-search visibility. Machine-friendly structure is something a machine produces well. Variants for testing. Generating ten headline or CTA variations to test is exactly the kind of high-volume, pattern-based work AI accelerates, and it feeds the continuous-testing discipline that conversion-centered design requires. Tightening and editing. AI is a capable line editor, cutting flab, fixing flow, flagging jargon. Used to tighten human writing, it improves clarity without erasing voice. In all of these, AI is operating on the median-execution layer, where competence is the goal and personality is not yet at stake.

Section 3

Where AI-generated copy KILLS conversion

These are the jobs where the distinctive, specific, and personal is the conversion mechanism, and where AI's averaging is poison. Your proof and claims. This is the most dangerous place of all. AI will happily generate plausible-sounding results, testimonials, and case-study details, and fabricated proof is not just ineffective, it is a legal and reputational catastrophe, and it is detectable by the buyer who checks. Proof must be real, specific, and yours. Never let AI write it; only let it format the real proof you provide. Your point of view. What you believe about your field, the positions you take, who you will and won't work with, these are the human judgments that differentiate you, and AI hedges them into mush because the average has no opinions. AI-written positioning reads as the generic competence everyone now has, which is to say it doesn't position at all. Your origin and "why." The about page, the founding story, the reason you do this work, these are the human-trust surfaces, and an AI-generated version is a tell. In a web where buyers are increasingly alert to AI sameness, a machine-written "our story" actively erodes the authenticity it's supposed to build. Your voice on the trust-critical pages. The hero, the "why us," the personal note, wherever a buyer is deciding whether to trust the specific humans behind the work, the copy must sound like a specific human. AI's brand-safe blandness is the opposite of the texture that earns trust.

Section 4

The Draft/Publish Line, in one view

The line is not "AI good" or "AI bad." It runs straight through your website, separating the median-execution layer (structure, drafts, formatting, variants, let AI fly) from the distinctive-trust layer (proof, point of view, story, voice, keep it human). The businesses that win use AI to do more of the first kind of work faster, which frees them to invest more care in the second kind, the only kind that actually differentiates them in a web full of AI-generated sameness.

Section 5

Execute This With AI

Here is a workflow to find where your site sits on the wrong side of the line, with any capable AI model. Step 1, Inputs. Paste your homepage and about-page copy. Note which sections, if any, were AI-written. Step 2, Run the line-audit prompt: You are a conversion copy editor. AI excels at MEDIAN execution (structure, drafts, formatting, variants) and fails at the DISTINCTIVE (proof, point of view, origin story, trust-critical voice). Audit my copy against this line. My copy: """ [PASTE] """ Do five things: 1. Flag every passage that reads as generic/AI-median and belongs to the distinctive-trust layer (proof, positioning, story, voice), these need a human. 2. Flag every passage where AI could safely HELP (structure, formatting, tightening). 3. Identify any place that states a result/testimonial/claim, confirm it must be real and mark unverified ones [VERIFY: must be your real fact]. 4. Rewrite ONE trust-critical passage to sound specifically human (ask me the questions you need; do not invent my experience or results). 5. Tell me where using AI more would free me to invest more in the human layer. Never fabricate proof, opinions, or my story. Step 3, Voice check. "Read my hero and 'why us' aloud in your head. Do they sound like a specific person with real opinions, or like competent AI copy anyone could have generated? Show me the difference with a rewrite that uses only what I've told you." Tools and expected output. Any frontier chat model. Expect a layer-by-layer flagging of your copy, verification markers on every claim, and a human rewrite of a trust-critical passage. The QA discipline, the whole point: the model can identify where your copy is too AI-like and can structure and tighten freely, but it must never generate the proof, the positioning, or the story that belongs to the human layer. The instant you let AI fabricate those, you've rejoined the undifferentiated 74% and forfeited the trust that converts. AI drafts the median; you own the distinctive. The question was never whether to use AI to write your website, you already are, and you should, for the right jobs. The question is whether you know where the line runs: let AI structure, draft, format, and tighten, and it makes you faster and sharper; let it write your proof, your positioning, your story, and your voice, and it strips out the exact specificity that makes a stranger trust you. Use it as a scalpel, keep the distinctive layer human, and you get the speed of AI without the sameness that is quietly making most of the web forgettable.

Section 6

Keep reading

Keep reading in the AI-Native Design & Agentic UX cluster and across the library: [How AI Is Changing What a "Good" Service Website Looks Like](/blog/how-ai-is-changing-what-a-good-service-website-looks-like), [What "Most Developers Now Use AI to Build" Means for Your Website](/blog/what-most-developers-now-use-ai-to-build-means-for-your-website), [Will AI Agents Replace Landing Pages? What the Evidence Suggests](/blog/will-ai-agents-replace-landing-pages-what-the-evidence-suggests). Also relevant: [The Service-Business Website Priority Stack: What to Fix First When Everything Needs Work](/blog/the-service-business-website-priority-stack-what-to-fix-first-when-everything-needs-work), [What Is AI Automation? A Plain-English Guide for Founders](/blog/what-is-ai-automation-a-plain-english-guide-for-founders).

Joshua Agonya Pi'Rwot

Written by

Joshua Agonya Pi'Rwot

Founder, Business Growth Accelerator · Country Director, AVODA Group Uganda · EMBA

Joshua helps service-business operators turn scattered marketing into a clear path from first attention to booked call. He is Founder of Business Growth Accelerator and Country Director of AVODA Group Uganda.