There is a particular kind of website that breaks my heart. The business is good. The product genuinely helps people. The founder can explain it beautifully over coffee. And then the homepage opens with “We are passionate about delivering innovative solutions” and every visitor who might have bought quietly closes the tab.
This article is the framework I use to stop that happening. It has nine steps, it works for a one-page service site or a forty-page ecommerce build, and none of it requires you to be a natural writer. What it requires is that you stop guessing.
Why most website copy quietly fails
Failed website copy rarely fails loudly. Nobody emails to say your value proposition is unclear. They just leave, and your analytics record a perfectly ordinary bounce rate that tells you nothing about the sentence that caused it.
In nine years of auditing sites, I see the same four failure points again and again:
- The copy describes the business instead of the reader. “We’ve been trading since 2011” is a fact about you. “You won’t be the first person we’ve talked through this” is the same fact, aimed at the person reading.
- Every page tries to do every job. When a page has five calls to action of equal weight, it has none.
- Objections are ignored rather than answered. The doubt in your visitor’s head does not disappear because your page declines to mention it. It just goes unaddressed, and unaddressed doubt wins.
- Proof is generic. “Trusted by hundreds of clients” proves nothing. “Forty-one clinics in the Waikato use this every Monday morning” is checkable, and checkable is persuasive.
Each of the nine steps below exists to eliminate one of those failures before it reaches the page.
Open a single document called Message Source. Everything you gather in steps one and two goes in there, verbatim. You will write the whole site out of that document, and you will be astonished how much of the finished copy is quotation rather than invention.
Step 1: Interview five real customers
Not a survey. A conversation, twenty minutes, recorded with permission. Five is the number where patterns start repeating and new information stops arriving; below three you are collecting anecdotes, above eight you are procrastinating.
Ask these six questions, in this order, and resist the urge to help them answer:
- What was going on in your business the week you started looking for something like this? This gets you the trigger event, which is where the buying journey actually begins.
- What did you type into Google? Their literal search phrasing is often better than any keyword tool output.
- What else did you consider, including doing nothing? “Doing nothing” is your main competitor in almost every market and almost nobody writes copy against it.
- What nearly stopped you? The most valuable question on the list. Whatever they say is an objection your copy must handle by name.
- What did you expect to happen, and what actually happened? The gap between the two is your honest promise.
- How would you describe us to a colleague in one sentence? Write down their exact words. This is frequently your headline.
Transcribe the recordings. Do not paraphrase. A customer who says “I was drowning in spreadsheets and I just wanted someone to take it off me” has handed you a headline; a summary that reads “customer wanted efficiency improvements” has thrown it away.
What to do if you have no customers yet
Pre-launch, substitute two things. First, mine reviews of the closest adjacent product — not to copy their claims, but to collect the language of the same buyer. Second, interview five people who fit your ideal customer profile about the problem rather than your solution. You will get less certainty and more direction, which is enough to start.
Step 2: Mine the reviews nobody reads
Review mining is the highest ratio of insight to effort in all of copywriting, and almost nobody does it properly. The technique: gather 100–200 reviews of products in your category from Amazon, Trustpilot, G2, Google, App Store, Reddit threads, wherever your market talks. Then read only the three-star reviews first.
Five-star reviews tell you what delight looks like. One-star reviews are usually about delivery or a bad day. Three-star reviews are where people carefully explain what they wanted, what they got, and the precise gap between the two. That gap is your positioning.
Sort every useful line into five buckets:
| Bucket | What goes in it | Where it gets used |
|---|---|---|
| Desired outcome | What they were trying to achieve | Headlines, page intros |
| Pain | The cost of the status quo | Problem sections, ad copy |
| Objection | What made them hesitate | FAQ, guarantees, mid-page reassurance |
| Trigger | What made them act now | CTA framing, email subject lines |
| Vocabulary | Words they use that you don’t | Everywhere |
Two hours of this will change what you write more than two weeks of brainstorming.
Step 3: Write the message hierarchy first
A message hierarchy is one page long and it settles the arguments before they happen. Mine has four parts:
- The one big idea — a single sentence that, if the reader remembered nothing else, would be enough. It should be specific enough that a competitor could not paste it onto their own site.
- Three pillars — the three reasons the big idea is true. Not five, not seven. Three is what people retain.
- Proof for each pillar — a number, a named client, a demonstration, a guarantee. If a pillar has no proof, it is a wish, and it comes out.
- Ranked objections — the top five, in order of how often they came up in interviews, each with the honest answer beside it.
Get sign-off on this document from whoever has veto power over the copy. It is a one-page argument about strategy, and people find it much easier to disagree productively about strategy than about a paragraph they simply do not like.
If you cannot fill in the proof column, you do not have a copy problem. You have a business claim you cannot yet support — and finding that out on a Tuesday in a Google Doc is far cheaper than finding it out in the market.
Step 4: Choose one job per page
Write the job at the top of each page’s draft, in this format: “A [who] who arrives here [in what state] should leave [having done what].”
For example: “A clinic owner who arrives from a Google search for ‘patient booking software NZ’, sceptical about switching, should leave having booked a demo.” That single line kills a surprising number of bad decisions. It tells you the page cannot also be a recruitment pitch, cannot open with company history, and must handle switching cost early.
Then choose one primary call to action for the page and make every other action visibly secondary — smaller, quieter, lower down. Two equal buttons is a coin toss you asked the visitor to perform.
Step 5: Earn the second sentence
The job of your headline is not to sell. It is to earn the second sentence. The job of the second sentence is to earn the third. Copy is a chain of small permissions, and most sites break the chain in the first eight seconds by opening with something that is true of everyone.
A working hero section has five components:
- The headline — the outcome or the mechanism, in the customer’s language.
- The sub-headline — who it is for, and how it works, in one sentence.
- The primary CTA — one, unmissable.
- A risk reducer — “no card required”, “30-minute call, no pitch”, “cancel any time”.
- A shard of proof — a number, a logo row, one short quote.
Test it with the competitor swap: paste your headline onto your closest competitor’s homepage. If it still reads as true, it is not a headline, it is a category description. Rewrite until the swap fails.
Step 6: Handle objections in the order they arrive
Objections are not a FAQ problem. They are a sequencing problem. A doubt that arrives at the pricing section and is answered four screens later has already cost you the reader.
Take your ranked objection list from step three and place each answer where the doubt appears:
| Objection | Where it arrives | How to answer it |
|---|---|---|
| “This won’t work for a business like mine” | Immediately after the hero | Name the segment explicitly: “Built for clinics of 2–15 practitioners” |
| “Switching will be a nightmare” | As soon as you mention change | Show the migration timeline, with who does what |
| “It’s too expensive” | At the price | Reframe against the cost of the status quo, with a number |
| “How do I know you’re any good?” | Before the CTA | Named, specific proof — not a stock testimonial |
| “What if it goes wrong?” | At the CTA | Guarantee, exit terms, or a genuinely low-commitment first step |
Naming an objection out loud does not plant it. It was already there. What naming it does is prove you have had this conversation before, which is exactly what a nervous buyer is looking for.
Step 7: Make proof specific
There is a hierarchy of proof, and most sites use the weakest kind:
- Demonstration — showing the thing working. Strongest, and least used.
- Numbers with context — “cut average onboarding from 11 days to 4 across 23 clients”.
- Named testimonials with specifics — full name, role, company, and a concrete before/after.
- Third-party validation — awards, certifications, published reviews.
- Client logos — fine as reassurance, weak as evidence.
- Adjectives about yourself — “trusted”, “leading”, “passionate”. Worth nothing. Delete.
When you request a testimonial, do not ask “would you write us a review?” Ask three questions: What was the situation before? What changed? What would you say to someone considering it? Then edit lightly for length and send it back for approval. You will get usable specifics instead of “great service, highly recommend”.
Step 8: Write CTAs that finish the sentence
A good call to action completes the sentence “I want to…” from the reader’s point of view. “Submit” finishes nothing. “Get my free copy audit” finishes it nicely.
Four rules that consistently earn their keep:
- First person beats second. “Start my trial” usually outperforms “Start your trial”. Test it, but start there.
- Name the next screen. If the button leads to a booking calendar, say “See available times”. Unexpected next steps are where people abandon.
- Reduce the perceived commitment. A line of microcopy under the button — “Takes about two minutes. No card needed.” — is often worth more than the button text itself.
- Repeat the primary CTA after every major proof block, and nowhere else.
Step 9: Measure one number, then edit
Before launch, write down the single number this page exists to move, and its current value. Conversion rate to enquiry. Demo bookings per week. Newsletter signups per 1,000 sessions. One number.
Then leave it alone for four weeks. Traffic is noisy and week one tells you nothing. At day 30, look at three things: did the number move, where do people now stop scrolling, and what are they asking in the enquiries you do receive? That last one is the richest — if three people ask the same question, your page has a hole exactly the size of that question.
Below roughly 1,000 sessions a month to a page, A/B testing will not reach significance in any useful timeframe. At that traffic level, do not run split tests — run better research instead. Five customer interviews will beat an underpowered test every single time.
The 12-point pre-launch checklist
Run this before anything goes live. It takes about twenty minutes and it has saved every project I have ever run at least one embarrassment.
- Does the headline fail the competitor swap test?
- Can a stranger say what you sell after five seconds above the fold?
- Is there exactly one primary CTA per page?
- Does every claim have proof within one screen of it?
- Are your top three objections answered by name?
- Is the price, or the price range, findable in under ten seconds?
- Does each testimonial include a full name and a specific outcome?
- Is every button label a verb phrase the reader would use?
- Have you removed every instance of “passionate”, “innovative”, “solutions” and “seamless”?
- Does the page read cleanly on a phone with only the headings visible?
- Is there a next step for the reader who is interested but not ready?
- Have you written the title tag and meta description, rather than letting the CMS guess?
Nine steps and a checklist will not make you a copywriter. But they will get you a website that argues for your business the way you would if you were in the room — which, most days, is the whole job.