Fernside Studio
Background
Archive
Journal Entry

Using AI to Write Website Copy: What Works and What Doesn't

Documented
Capacity
10 MIN READ
Domain
AI & Automation

In 2026, 90% of content marketers use AI writing tools. For website copy specifically—the kind that sits above the fold and drives conversions—AI has become a standard part of the process. Not the whole process. Part of it.

At Fernside Studio, we use AI tools in our copywriting workflow, particularly for Studio Sites and blog content. But we’ve learnt what AI genuinely helps with and where it creates more work than it saves. Here’s what we’ve found after writing copy for dozens of SMB sites with AI in the mix.

What AI Tools Are Actually Good At

1. Beating Blank Page Syndrome

The hardest part of writing website copy is starting. You know what your service does, but translating that into clear, compelling copy feels paralysing. AI tools excel at generating first drafts that give you something to react to.

According to Siege Media’s 2025 research, writers using AI tools spend 30% less time on initial drafts. That tracks with our experience. Instead of staring at a blank Google Doc, you’re editing a rough structure within minutes.

How we use it: We feed AI tools a brief outline—target customer, core problem, desired outcome—and ask for three different hero section variations. We never use what it produces verbatim, but it surfaces angles we hadn’t considered.

2. Exploring Different Messaging Angles

AI can generate multiple versions of the same message quickly. This helps you see which angle resonates before committing to a direction.

For a recent accountancy firm site, we asked Claude to rewrite the same value proposition five ways: emphasising speed, cost savings, compliance expertise, peace of mind, and founder access. The client immediately gravitated toward the “peace of mind” angle—something we might not have prioritised without testing variations.

Practical tip: Generate 3–5 versions of your core message. Use them as conversation starters with your team or trusted customers to identify which angle lands.

3. Rephrasing Clunky Sentences

You know what you mean, but the sentence feels awkward. AI tools excel at offering clearer alternatives without changing the underlying meaning.

This works especially well for technical services where you need to explain complex offerings simply. Paste your jargon-heavy draft into ChatGPT or Claude with the prompt “Rewrite this for a busy founder who doesn’t work in this industry” and you’ll get something more accessible.

4. Structuring Long-Form Content

For service pages or guides where you need to cover multiple sections, AI can help organise your thinking. Feed it a brain dump of points and ask for a logical structure.

We use this regularly for our blog content service. The AI suggests heading hierarchies and content flows that we refine based on SEO priorities and reader intent.

Where AI Falls Short (And Creates More Work)

1. Brand Voice Consistency

AI-generated copy tends toward corporate blandness. Even with detailed prompting, tools like ChatGPT revert to filler phrases—“unlock potential,” “seamless experience,” “delve into.” According to Medium’s analysis of ChatGPT copywriting, this “AI fluff syndrome” persists despite custom instructions.

Claude performs better here—its Styles feature helps maintain tone more consistently—but it still lacks the personality that makes copy memorable. Your brand voice is built on specific word choices, sentence rhythms, and deliberate constraints. AI doesn’t inherently understand these nuances.

How we handle it: Every AI-generated draft goes through heavy editing to match Fernside’s calm, precise voice. We remove filler, tighten sentences, and replace generic phrases with specific claims. This editing phase often takes as long as writing from scratch would have—but we’re working from a solid structural foundation.

2. Knowing Your Specific Customers

AI can’t interview your customers. It doesn’t know the exact language they use when describing their problems, the specific objections they raise, or the proof points that matter most to them.

When writing copy for a high-stress service, understanding emotional context is critical. AI tends to intellectualise problems rather than reflecting how customers actually feel about them.

Practical tip: Before using AI, write down 3–5 direct quotes from customer conversations—the exact words they used to describe their problem. Feed these quotes into your AI prompt as context. The output won’t be perfect, but it’ll be closer to reality.

3. Emotional Resonance

AI struggles with humour, nuance, and emotional weight. It can describe emotions, but it rarely evokes them. For website copy—especially CTAs and value propositions—this matters.

Compare these two CTAs for a web studio:

  • AI version: “Get started with our comprehensive website design services today”
  • Human-edited version: “Book a Launch Sprint—live in five days”

The second is specific, time-bound, and actionable. It creates urgency without hype. AI rarely produces this level of precision without heavy prompting and editing.

4. Conciseness (AI Loves to Waffle)

AI-generated copy tends to be 30–40% longer than necessary. Tools like ChatGPT prioritise sounding helpful and thorough over being concise. This directly conflicts with good web copy, which respects reader attention.

Research from 2026 on AI content quality shows that 94% of AI content accuracy issues could be prevented through systematic human oversight—including trimming unnecessary elaboration.

How we fix it: After generating AI copy, we cut 30–50% of the word count. Every sentence earns its place. If a paragraph doesn’t advance understanding or drive action, it’s gone.

Practical Tips for Using AI in Your Website Copywriting

1. Feed It Your Best Existing Copy

AI tools perform better when you give them examples of your actual brand voice. Before asking for new copy, paste 2–3 samples of your strongest existing content and say, “Write in a similar tone.”

This works especially well if you’re extending an existing site or adding new service pages. The AI can pattern-match against your established voice more effectively than working from generic prompts.

2. Use AI for Structure, Not Finished Prose

Think of AI as a structural assistant, not a ghostwriter. Ask it to:

  • Suggest heading hierarchies for service pages
  • Outline the key sections for a what to include on your website audit
  • Organise scattered notes into a logical flow

Then write the actual copy yourself, using the AI-generated structure as scaffolding.

3. Edit Ruthlessly

Set a rule: Every AI-generated sentence must be edited before publication. This forces you to engage critically with the output rather than copying it wholesale.

Companies implementing systematic AI oversight achieve 67% better content performance and 45% fewer brand consistency issues compared to those using AI without human guidance.

At Fernside, AI-generated drafts go through at least two editing passes—one for clarity and concision, another for brand voice alignment.

4. Ask for Multiple Versions, Then Synthesise

Don’t settle for the first output. Ask for 3–5 variations of the same section, then synthesise the strongest elements into your own version.

This approach combines AI’s speed with your editorial judgment. You’re not choosing an AI-written option—you’re using multiple AI attempts to surface ideas you then shape into something better.

5. Use AI to Overcome Specific Blockers, Not Replace Thinking

AI works best when you know what you need but can’t quite articulate it. It’s less effective when you haven’t thought through your core message at all.

Before opening ChatGPT or Claude, answer these questions:

  • Who is this copy for, specifically?
  • What do they need to believe to take action?
  • What’s the single most important thing this section should communicate?

Then use AI to explore how to say it, not what to say.

How Fernside Studio Uses AI in Our Copywriting Process

We use AI tools—primarily Claude—as part of our website copywriting workflow, but always with human oversight. Here’s how it fits in:

  1. Client interview: We discuss their service, customers, and key differentiators. No AI involved—this is about understanding context the tools can’t access.

  2. Messaging framework: We map out core value propositions, proof points, and objections manually. This becomes the brief we give to AI tools.

  3. AI-assisted drafting: We use AI to generate multiple versions of key sections—hero copy, service descriptions, story-driven frameworks. This speeds up the initial writing phase.

  4. Heavy editing: Every AI-generated section is rewritten to match Fernside’s voice, remove filler, and sharpen claims. We typically keep 40–60% of the AI output and rewrite the rest.

  5. Client review: We present the edited copy for feedback. If revisions are needed, we use AI again to explore alternative phrasings—but the final decision is always human.

This hybrid approach gives us the speed benefits of AI without sacrificing the precision and personality that makes copy convert. We’re faster than writing everything from scratch, but the output still feels unmistakably Fernside.

The Limits of AI for Conversion Rate Optimisation

Here’s what AI tools genuinely struggle with:

  • Knowing which proof points matter most to your specific audience
  • Creating urgency without sounding pushy or generic
  • Writing CTAs that feel natural within your brand voice
  • Balancing detail with brevity on service pages
  • Adapting copy based on real user feedback and conversion rate data

These tasks require judgment, customer knowledge, and iterative testing. AI can assist, but it can’t replace strategic thinking.

If you’re writing copy for a high-stakes page—pricing, what belongs on a business website, or service descriptions—use AI to speed up drafting, but invest serious editing time before publishing.

Should You Use AI for Your Website Copy?

Yes, with caveats.

AI tools are genuinely helpful for:

  • Overcoming writer’s block
  • Generating multiple messaging angles quickly
  • Rephrasing technical jargon into plain language
  • Structuring long-form content logically

They’re less helpful (and sometimes counterproductive) for:

  • Capturing your specific brand voice
  • Writing emotionally resonant copy
  • Creating concise, punchy messaging
  • Understanding your customers’ actual language and concerns

The best approach: Use AI as a drafting assistant, not a copywriter. Feed it context, generate options, then edit ruthlessly. The final copy should feel like you wrote it—because, ultimately, you did.

Next Steps: Get Your Website Copy Right

If you’re building a new site or refining existing copy, every week with generic, AI-sounding copy on your site is a week prospects are forming the wrong impression of your business. Your competitors are already investing in copy that sounds human, specific, and conversion-ready.

Fernside Studio can help. We use AI tools in our process, but every word is human-edited to match your brand and convert your ideal customers.

Option 1: Launch Sprint — £750 fixed for a custom one-page site, including strategy call and copy refinement. Live in five days.

Option 2: Studio Site — From £2,400 for a multi-page marketing site with bespoke copywriting and onboarding workshop.

Option 3: AI Consultancy — If you’re exploring when to build custom AI tools vs. using off-the-shelf solutions, we can help you audit your workflow and identify high-value automation opportunities.

We only take on a few projects each month. Check availability and we’ll confirm your earliest build slot within 24 hours.

Sources

Say hello

Quick intro