Web Copywriting
How to Improve Website Copy When You Cannot A/B Test
Nearly all conversion advice assumes traffic that most sites do not have. At two hundred visits a month, a split test will never reach significance — but the copy can still get better.
Almost all conversion advice is written for sites with traffic to spare. Run the test, wait for significance, ship the winner. For a site getting two hundred visits a month, that advice is not merely impractical — following it produces confident decisions based on noise, which is worse than not testing at all.
Why the Numbers Do Not Work
Detecting the size of improvement copy realistically produces — a few percentage points — needs thousands of visitors in each variant. At a few hundred visits a month, the test would need to run for years before the result meant anything.
What happens instead is that someone stops the test at three weeks because one variant is ahead, and ships a change that was random variation. The site then carries a "proven" improvement that improved nothing, and the next decision is built on top of it.
Stopping a test the moment it looks like a winner is the single most common error in low-traffic testing. Results wander early and settle late; an early lead is the expected behaviour of noise, not evidence.
What Works at Low Volume
The methods below trade statistical certainty for something a small site actually has: access to individual people, and the ability to make large changes quickly.
- Watch five people use the page. Five sessions will surface problems no split test would have thought to test. This is the highest-return activity available to a low-traffic site and it costs nothing.
- Ask the people who did convert. "What nearly stopped you?" produces the objection list your page should be answering. Ten answers beats a test you cannot run.
- Read what people type. Search queries, chat transcripts and the first line of enquiry emails all record confusion the page created.
- Make big changes, not small ones. Button colour is untestable at this volume and irrelevant at any volume. Restructuring a page around a different argument is large enough to notice without statistics.
- Use paid traffic to test, briefly. If you need a decision quickly, a small ad budget can put a few hundred targeted visitors through two versions faster than organic will.
Measure Further Up the Funnel
Enquiries are too rare on a small site to move visibly. Metrics closer to the top happen more often and shift sooner: how many people reach the form at all, how far they scroll, how many start filling it in and abandon it.
These are not the outcome you care about, and they are directionally useful. A change that doubles the number of people reaching your form is doing something real, even before it shows up as enquiries.
Form abandonment is worth instrumenting before anything else. People who began filling in a form were interested and then stopped — that is the most concentrated source of copy problems on any site, and it needs no traffic volume to interpret.
Before-and-After, Honestly Framed
Where you cannot test, you can still compare a period before a change with a period after — provided you are honest that it is not a controlled experiment. Record what changed and when, keep the window long enough to cover a normal cycle, and note anything else that moved: a campaign, a season, a competitor.
This is weaker evidence than a test and considerably stronger than opinion, which is the real alternative on a small site.
Fix the Obvious Things First
Most low-traffic sites have unanswered objections, no price signal, a form asking for more than it needs, and a call to action that does not say what happens next. None of that requires testing to identify, and fixing it will outweigh anything a split test would have found.
The diagnostic questions are the same ones behind a service page that gets traffic but no enquiries, and the material comes from your customers' own words. Our website copywriting service starts there rather than with a testing tool.
Questions
Questions about this article
Still unsure about something? Talk to a specialist — no script, no sales pitch.
More than most small sites have. Detecting a realistic improvement — a few percentage points on conversion — generally needs thousands of visitors per variant, which at a few hundred visits a month means a test running for years. The honest answer is usually that you cannot test, not that you should test longer.
Only up to a point. A test running for months is contaminated by seasonality, changing traffic sources and any other change you make to the site meanwhile. At some duration you are no longer measuring the variant, you are measuring the year.
Big structural changes, judged over a long period against a clear baseline — not button colours. If a change is large enough to matter at low volume, you will usually see it in enquiries rather than in a significance calculation.