Conversion Copywriting: Testing Words That Move the Number

Copy is one of the highest-return things you can test and one of the most under-tested. Teams will spend weeks on a visual redesign and never seriously test the words, even though the words are what carry the argument for why someone should act. When the design team at Chegg was stretched, I wrote the copy myself, and testing that copy was often what moved the number more than any layout change. This is how to write and test conversion copy that actually shifts the metric, rather than copy that merely sounds nice.

Copy is the argument, design is the stage

Start with the right mental model. Design arranges attention; copy makes the case. A beautiful page with a weak argument converts worse than a plain page with a compelling one, because the visitor ultimately acts on what the words persuade them to do. This is why copy deserves the same experimental rigor as anything else on the page, and why it is a mistake to treat it as the thing you fill in after the design is done.

Conversion copy has one job: to move a specific person one step closer to acting, by addressing what is actually stopping them. That framing, addressing the real blocker, ties copy directly to hypothesis-driven testing, because good copy is a change that acts on a belief about why the user hesitates, which is exactly the structure I lay out in hypothesis-driven experimentation.

Write to the hesitation, not the feature

The most common conversion-copy mistake is writing about the product instead of about the reader’s hesitation. “Our platform supports twelve integrations” is a feature. “Connect the tools you already use in two minutes” speaks to a worry the reader actually has. The second converts better because it addresses a blocker, not a spec.

To write to the hesitation, you first have to know it, which means the same diagnostic work that feeds any good experiment: reading support tickets, sales objections, exit surveys, and session recordings to find the doubts in the reader’s own words. The strongest conversion copy often reuses the customer’s exact language, because it proves you understand their problem. Copy grounded in real hesitation is a hypothesis about the user rendered as words, and it is testable in exactly the same way a design change is.

The principles that reliably move the number

A handful of copy principles hold up across almost every page:

• Lead with the benefit, then support with the feature. People buy outcomes. Name the outcome first, then explain how the feature delivers it.
• Be specific. “See your first result in two minutes” beats “fast and easy.” Specificity is credible; vagueness is ignored.
• Reduce perceived risk. Guarantees, social proof, and clear terms lower the cost of saying yes, which is often the real blocker.
• One idea per element. A headline that tries to say three things says nothing. Give each piece of copy a single job.
• Write the way your reader talks. Jargon and inflated language create distance. Plain, concrete words build trust.
• Make the next action obvious. The call to action should say exactly what happens next, in the reader’s terms, not “submit.”

None of these is a trick. They are all in service of clarity and reduced hesitation, which is what conversion copy is for. And every one of them is testable, which is the point: you do not have to argue about which headline is better, you can find out.

Test copy the same way you test anything

Copy is a perfect thing to A/B test, because changes are cheap to produce and the effects can be large. But test it with the same discipline as any experiment, not as a random rotation of clever lines.

Frame each copy test as a hypothesis: “We believe visitors hesitate because they do not understand what happens after they sign up. If the CTA and subhead spell out the next step, then form completions will rise, because we have removed the uncertainty.” Now the test teaches you something about your users, win or lose, rather than just crowning a phrase. Isolate the copy change so you can attribute the result, and read it with the same statistical care as any test, per statistical significance for product managers.

Because copy changes are so cheap to build, they are ideal for keeping an experimentation program moving, and they are exactly the kind of variant a growth practitioner can ship without waiting for the design queue, which is a throughput advantage I discuss in landing page optimization at scale.

What to test first

Not all copy carries equal weight. Prioritize the copy that the most people see and that sits closest to the decision.

• Headlines. The first thing read and often the only thing read. A stronger headline can move a page more than any other single change.
• Calls to action. The words on the button and around it, at the moment of decision, punch far above their size.
• Subheads and the opening line. They decide whether the reader keeps going or leaves.
• Risk-reducers near the action. Guarantees, trust signals, and reassurances placed exactly where hesitation peaks.

Start where attention and intent are highest, which is usually the headline and the primary CTA, and work down from there. This is the same concentrate-where-it-pays logic that governs experiment prioritization: test the words the most people read at the most decisive moment first.

Beware the clever line that does not convert

A warning that experienced writers need most: clever and converting are not the same thing. A witty headline that wins awards can lose to a plain one that simply tells the reader what they get, because the reader is not there to admire your writing, they are there to decide. Some of the highest-converting copy is almost boringly clear.

This is exactly why you test rather than trust taste. The line you are proudest of may lose to a blunt statement of the benefit, and the only way to know is to let the number decide. Testing keeps copywriting honest, because it repeatedly humbles the instinct toward cleverness in favor of clarity. Earn the right to your opinion about a line with a test, the same principle that governs the whole program.

Copy is a conversion surface everywhere, not just landing pages

The discipline extends well beyond landing pages. The subject line of an onboarding email, the microcopy in a checkout flow, the empty-state message in a product, the error message that either reassures or frustrates, all of these are conversion copy, and all of them can be tested. Teams that treat copy as a first-class, testable conversion surface across the whole experience, not just on marketing pages, find lift in places others never look.

The mindset is to see words as levers everywhere a user is deciding something. Wherever there is a decision, hesitation, or confusion, there is copy doing a job, and that copy can be improved with the same hypothesis-and-test discipline. That is how copy stops being an afterthought and becomes one of the most reliable sources of conversion lift you have.

Common conversion-copy mistakes

• Writing about the product, not the reader’s hesitation.
• Vagueness where specifics would be credible.
• Cramming multiple ideas into one headline or CTA.
• Trusting clever over clear, and never testing which wins.
• Treating copy as decoration to finish after the design, rather than the argument the design serves.
• Only testing copy on landing pages, ignoring emails, product, and microcopy.

Avoid these, write to the real hesitation, and test your words with the same rigor you test everything else, and copy becomes what it should be: one of the highest-return levers in your whole conversion practice.

A worked rewrite, before and after

Principles are easier to trust when you see them applied. Take a real pattern of headline: “The All-in-One Platform for Modern Teams.” It is smooth, it is confident, and it says almost nothing, because it names no outcome, addresses no hesitation, and could sit atop a hundred different products. A reader skimming learns nothing about whether this solves their problem, so they leave.

Now rewrite it to the hesitation. Suppose the diagnostic work, the support tickets and exit surveys, told you the real doubt is “will this actually save me time, or is it another tool to manage?” The rewrite becomes something like “Cut your weekly reporting from three hours to twenty minutes.” It leads with a specific outcome, it speaks to the exact worry the reader carries, and it is concrete enough to be credible. It is also, notably, less clever than the original, and that is the point: it trades polish for clarity and relevance, which is the trade that converts. You would still test the two against each other rather than assume, but the rewrite is grounded in a belief about the reader, so even if it loses you learn whether the time-saving angle was the right hesitation to address.

Copy and design are tested together, not in sequence

A trap experienced teams fall into is testing copy and design as if they were independent, first settling the layout, then swapping words inside the frozen frame. In reality they interact, because a headline that works in a spacious hero fails when crammed into a narrow column, and a benefit that needs a supporting image to land will underperform as text alone. The words and the space they live in make the argument together.

The practical consequence is that your strongest copy test often changes a small amount of design with it, giving the words room to work, and your strongest design test carries copy chosen to suit the new layout. This does not violate the discipline of isolating variables; it means the variable you are testing is sometimes “this message in this treatment” rather than the words in a vacuum. Be honest about which you are testing. When the hypothesis is about the message, hold the layout as steady as the message allows. When it is about presentation, hold the words steady. What you must not do is pretend the two never touch, because that is how a technically clean copy test produces a result that evaporates the moment the design around it changes.

Voice is a conversion asset, not just a brand one

The last thing worth saying is that a consistent, human voice compounds across an entire experience in a way individual winning lines do not. When every headline, button, error message, and email sounds like it comes from the same trustworthy person who understands the reader’s problem, the cumulative effect is a relationship, and a reader who trusts the voice hesitates less at every subsequent decision. Voice is usually filed under brand, but its effect is measurably conversion.

This is why the copy work should not be a series of disconnected tests that each optimize one line in isolation. The winning lines should teach you something about the voice, what register, what level of directness, what kind of specificity your particular readers respond to, and that learning should propagate into how everything is written, the same way a winning component propagates across a page system. Test the lines, but harvest the learning about voice, because a coherent voice earned through testing is a conversion asset that keeps paying long after any single headline has been replaced.

The short version

• Copy is the argument; design is the stage. It deserves equal rigor.
• Write to the reader’s real hesitation, in their words, not to the feature list.
• Follow the principles that reduce hesitation: benefit-first, specific, risk-reducing, one idea per element, plain, clear next action.
• Test copy as hypotheses, isolated and read with statistical care.
• Test the highest-attention, highest-intent copy first: headlines and CTAs.
• Distrust clever; let the number decide. And test copy everywhere, not just on landing pages.

The words carry the sale. Test them like it.

---

I am Deepanshu Grover, a Growth Product Manager in Paris. I wrote and tested the copy behind a 34% conversion lift across 200+ pages at Chegg. If your words are untested, connect on LinkedIn or get in touch.