Conversion Rate Optimization With Low Traffic: What Actually Works

Most conversion advice assumes you can run A/B tests. On a site with a few thousand visitors a month, you can't, at least not in a way that gives you an answer this year. That doesn't mean you can't improve your conversion rate. It means you work differently: fewer tests, bigger changes, and more listening. This is part of our guide to increasing your conversion rate without more traffic.

Why an A/B test can't decide on a small site

An A/B test splits your visitors between two versions of a page and checks whether one sells more. The catch is that small differences look exactly like luck unless you have a lot of visitors.

How many depends on three things: your current conversion rate, how big a change you want to detect, and how sure you want to be. Using the standard formula for comparing two rates, at the usual settings (95% confidence, 80% power), here is what it takes:

• To tell whether a page converting at 2% really moved to 2.5%, you need about 13,800 visitors on each version, so about 27,600 in total.
• To detect a bigger jump, from 2% to 3%, you need about 3,800 on each version, about 7,700 in total.

Now put that against real traffic. If your sales page gets 1,000 visitors a month, the first test takes over two years. Even the second one takes more than seven months, and in seven months your offer, your prices and your traffic will all have changed.

What happens in practice is worse than waiting. People stop the test after two weeks, see that version B is "ahead", and ship it. With that little data, B being ahead is often just chance. You end up changing your page based on noise, and sometimes making it worse.

On a small site, an A/B test usually finishes with noise that looks like a result.

Fix the obvious problems first, without testing

You don't need a test to know that a broken checkout button is costing you sales. Many conversion problems are not questions of taste; they're plain mistakes, and fixing them is safe.

Go through the page with this list:

• The five-second test. Show the first screen to someone who doesn't know your business for five seconds. Can they say what you sell? If not, rewrite the headline until they can.
• The phone. Open the page on a phone, on mobile data. Is the button visible without hunting? Does it load in a few seconds? Many small sites are built on a desktop and never checked.
• The price. Is it on the page, or does the visitor find out at checkout? A surprise price is a reason to leave.
• The next step. Is there one clear button per offer, with what happens next written next to it?
• The proof. Is there at least one specific result from a real customer, with a name and a before and after?
• The path. Make a real purchase or booking yourself, start to finish. Broken forms and dead links are more common than anyone admits.

We go through the full list of reasons visitors leave in why your sales page isn't converting. Fixing the ones that apply needs no statistics at all.

Fix what's clearly broken before you test what's arguably better.

Make bigger changes, not smaller ones

The sample sizes above have a useful lesson hidden in them: a bigger change needs far fewer visitors to show up. Detecting a jump from 2% to 3% takes about a quarter of the visitors needed for 2% to 2.5%.

So on a small site, testing a new button colour is a waste of time. Even if it helps, the effect is too small to see. Change something that could plausibly move the number a lot:

• a completely different headline that says what the offer is, instead of a slogan;
• a different first screen that leads with the result for the buyer instead of the founder's story;
• a different offer structure, like adding a cheaper starter option;
• a different path, such as asking visitors a few questions first and sending each one to the option that fits.

Big changes are also easier to learn from. If the number moves a lot, you'll see it even without a formal test.

With little traffic, only big changes are visible, so make the changes that could matter most.

Compare before and after, the careful way

When you can't split traffic, you can still compare periods: the page before the change against the page after. It's less precise than an A/B test, because other things change at the same time. You can reduce that with a few rules.

1. Use full weeks. Compare four full weeks before with four full weeks after. Buying patterns on weekends differ from weekdays.
2. Change one thing at a time. If you rewrite the headline and the price in the same week, you won't know which one moved the number.
3. Keep a change log. Write down the date of every change to the page, your offer, your prices and your ads.
4. Watch for outside events. A holiday, a sale, a newsletter that went out, a post that went viral. Any of these can swamp the effect of your change. If one happened in either period, say so in your notes or pick different weeks.
5. Split by source. Compare visitors from the same source in both periods, such as email with email, ads with ads. A shift in where traffic comes from can look like a change in the page.

A before-and-after comparison won't prove anything to a statistician. It will tell you whether a big change helped, hurt, or did nothing visible, and that's enough to decide what to keep.

A careful before-and-after comparison is the A/B test of the small site.

Ask real people

When the numbers are too thin, people fill the gap. You don't need many of them.

• Watch five people use the page. Ask friends of friends who match your buyer, not your own friends. Give them a task ("find out if this is for you and what it costs") and watch where they hesitate. Five sessions often surface the main problems.
• Ask your buyers. A one-line email after purchase: "What almost stopped you from buying?" The answers are your list of doubts the page should answer.
• Ask people who didn't buy. If you have their email, from a booked call that didn't close or an abandoned signup, ask what was missing. Fewer will reply, but their answers are worth more.
• Read your inbox. Every question someone emails you before buying is a question the page didn't answer.
• Use session recordings carefully. Tools that record visits show where people scroll and stop. They show what happened, not why, so pair them with the questions above.

Five honest conversations will tell you more than five months of an underpowered test.

Measure closer to the decision

One more way to get readable numbers from little traffic: measure a step that happens more often than the sale.

If 1,000 visitors produce 20 sales a month, you're working with 20 events, and 20 is too few to see much. If the same 1,000 visitors produce 150 clicks on "Buy" or "Book", you have 150 events, and changes show up sooner. Clicks to the offer are not sales, so check that the two move together over time. They're still an earlier and more frequent signal.

This is the main number Snacked reports: clicks to each of your offers, out of everyone who opened the funnel, on every plan. Add a short snippet to your thank-you page and it counts the actual sales too. It doesn't run A/B tests for you, so the before-and-after rules above still apply.

When sales are too few to read, watch the click to the offer, and check it against sales over time.

Recap

• On a small site, A/B tests need more visitors than you have, and stopping them early means acting on noise.
• Fix the obvious problems without testing, then make big changes and compare full weeks before and after, one change at a time.
• Fill the gaps with people: watch five visitors, ask buyers and non-buyers, and track clicks to the offer as an earlier signal.

There's more on what to change first in our guide to increasing your conversion rate without more traffic. For a quick list of what's wrong on your page today, try the free sales page grader.