How to improve your website’s sales (without much traffic)

Today you'll discover three simple ways to improve your website's conversions without much traffic (or any at all).

But first, we need to review:

The six factors which affect how long to run your split test

As we learned in our Split Testing 101 guide, there are six factors which affect how long a test should run:

Traffic: More visitors make it faster to get reliable results.
Conversions: The more people convert, the quicker you'll get meaningful results.
Current conversion rate: Higher-converting websites see faster outcomes, needing less test time.
Number of variations: Testing more options takes longer as you gather enough data for each one.
Confidence levels: Higher confidence requires bigger samples and more testing to avoid wrong conclusions. Finding a balance is important.
Difference in conversion rates: If the expected change is big, you'll reach reliable results faster.

To get a feel for how these factors affect your testing time, use an A/B test duration calculator (from VWO) to estimate how long your test will need to run.

If your test will need months to complete, try these three approaches:

1\. Change the goal

You can reduce your testing time by changing your goal.

For example, let's say you get 30 sales per month. At that rate, it would take months to reach statistical significance.

So, to move faster, you could change your goal to focus on:

leads
clicks to the cart/checkout

This will let you run tests much faster (think weeks instead of months). And by testing faster, you'll learn faster.

Then, as your traffic grows, you can start focusing on sales as your main goal.

Of course, if your test will still take to long to run, you can always:

2\. Lower the confidence level

You can cut your testing time in half if you decrease your level of statistical significance.

For example, most split testing software declares a winner at 95% confidence. But if you accept 90%---which is perfectly fine!---you only need half the traffic.

To put this another way: a 90% confidence means that, nine times out of ten, your result is legit.

I'll take those odds.

But what if you have very little traffic? Or none at all?

Then you should:

3\. Run five user tests

If you're just getting started, I'd suggest running five user tests on your funnel to get feedback and iterate from there.

Why only five? Because studies show that five user tests uncover 85% of usability issues, so there's no point doing more than five user tests at a time.

Plus, you can do this before you even launch. It's a great way to iterate your MVP.

By using these three approaches---changing the goal, reducing the confidence level, and user testing---you'll iterate (and grow) much faster.

Happy testing. 🙂

Subscribe to Funnel Candy's newsletter for expert tips to grow your business (based on 1,000+ experiments on startups and sites like Amazon, Dropbox, and WIRED).