How to Optimize Your Site Even If it Doesn’t Have a Lot of Traffic
The conventional wisdom is that your site needs a healthy amount of traffic if you want to optimize it for conversions. That’s because, statistically speaking, you can’t be confident that your split test results are accurate without a large sample size.
That certainly makes sense at face value. After all, if a presidential polling organization released a poll showing Hillary Clinton tied with Donald Trump at 50% each but only polled two people, you probably wouldn’t take that poll seriously.
Nor should you. Sample size matters.
Unfortunately, it’s that sample size that creates a challenge if your site doesn’t have a lot of traffic.
Let’s say you’d like to split test your call to action (CTA) text but your site is currently only seeing about 50 visitors per day. Your conversion rate is currently 2% but you’d like to increase it to 3%.
So, you use an online tool like the one provided by Optimizely to determine the sample size necessary to give you an accurate assessment. The tool informs you that you need 2,900 visitors per variation or 5,800 visitors in total.
That test would last 116 days (5,800/50) or almost four months!
That’s not realistic. Plus, you’ll also jeopardize the results of your test if you let it run for too long.
So is all hope lost if you’d like to optimize a site with minimal traffic?
No.
Here are a few ways to optimize your site for conversions even if you don’t have a lot of visitors.
1. Reduce Your Split Testing Confidence Level
The general rule of thumb for split testing is that you should go for a confidence level of 95%.
That’s all fine and dandy when you’ve got a boatload of traffic. However, if you’re struggling to bring in visitors, it might be worthwhile to lower your confidence level a bit.
Ask yourself this: can you live with an 90% confidence level? How about something even lower?
If not, then skip on to the next point.
On the other hand, if you’re willing to accept a 10% or 15% chance that your test results are wrong instead of just a 5% chance, you might be able to move the needle a bit by reducing the confidence rate.
Truth be told, though, you usually won’t move it much unless you’re testing something that you believe could have a significant impact on your conversions (that is, you’re expecting a 100% lift or better).
Still, every little bit helps.
2. Test Big Changes
I hinted at this in the last point. If you want to test with limited traffic, consider testing big changes.
Why? Because big changes can produce big results. You can see lifts that are much higher than you would notice if you just made small changes.
The reason that benefits you if you’re working with limited traffic is because big lifts mean you can use a smaller sample size.
Let’s reconsider the example from above, except instead of trying to increase your lift from 2% to 3%, you’re trying to double it from 2% to 4%.
Now, instead of 2,900 visitors per variation, you only need 710 visitors per variation.
And that’s at a 95% confidence level. If you want to lower the confidence level to 90%, you only need 670 visitors.
If your site is only receiving 50 visitors per day, that means you can complete your test in two weeks. That’s much more realistic.
3. Test Micro-Conversions
It might be the case that your money page only receives a limited amount of traffic but there are pages on your site that are higher up in the sales funnel that receive much more traffic. If so, test conversions on those pages.
For example, you might have a lead collection page. That’s a page that visitors will see long before they get to the shopping cart.
Although your lead collection page doesn’t contribute directly to your sales, it certainly contributes indirectly to your sales.
Although your lead collection page doesn’t contribute directly to your sales, it certainly contributes indirectly to your sales. That’s why you want to test conversions on that page as well.
Plus, you’re likely to see much more traffic on pages that are higher up in the sales funnel. That’s why it looks like a sales funnel, after all.
If you’re running a site that only receives 50 visitors per day on the money page but 200 visitors per day on the lead collection page, you’re in a pretty good position to optimize conversions on that page.
Recall that you need a sample size of 5,800 visitors. If the lead collection page receives 200 visitors per day, then you can wrap up your test in just 29 days.
Once again, that’s at a 95% confidence level. Lower the confidence level and you can reduce that time even more.
4. Buy Traffic
You might not be in a position where you can buy traffic all the time, but you can buy it just long enough to run your split test. If so, consider allocating some resources to that effort.
Check out the AdWords keyword tool. See if you can find a search-term relevant to your niche that receives a steady number of searches every month. Then, run an optimized AdWords ad that will drive traffic to your site during the testing period.
Also, if you’re targeting people based on demographics and interests, consider running Facebook ads that will send traffic to your site. You’ll likely find that Facebook ads are cheaper than AdWords ads.
As a bonus, you’ll build some brand-name recognition when you buy traffic while you’re running your test. So it pays off in a couple of ways.
5. Use Qualitative Analysis
You might not have enough traffic for quantitative analysis, but that doesn’t mean that you can’t do any qualitative analysis.
What’s the difference? Quantitative analysis, as the name implies, is based on raw, hard numbers. It’s fairly objective.
Qualitative analysis, on the other hand, is more subjective. It’s based on evaluating your site from a user experience perspective.
Fortunately, there are plenty of ways that you can perform qualitative analysis on your website, even with limited traffic.
For starters: did you know that you can record user sessions? Thanks to tools like HotJar, you can watch what visitors do once they get to your site.
And you might see trends that will be obvious even if you don’t have the requisite sample size for quantitative analysis.
Visitor Recordings Can Remove a Lot of Guess Work
Or, has the HotJar website puts it, you can “learn what’s holding your customers back.”
Of course, it’s going to be time-consuming to watch recordings of what your visitors are doing. You might have a little bit of extra time, though, if you don’t have a lot of traffic yet.
If you don’t have that kind of time, HotJar also offers heat-maps to show you exactly where visitors are placing their cursors on your website. That should give you some insight into what’s working and what’s not.
In addition to tools like HotJar, there are a number of other qualitative analysis methods, including:
- User surveys
- Usability studies
- Sales questions
- Customer service questions
Any combination of those options would make an excellent substitute for quantitative analysis if your site isn’t yet attracting a lot of visitors.
6. Drop the Multivariate Testing
You might have a number of things that you’d like to test all at once. For example, you might want to test a Headline A with Image A, Headline B with Image A, Headline A with Image B, and Headline B with Image B.
Sorry, but when you’re dealing with low traffic, multivariate testing is a bad idea.
Why? Because you need an even larger sample size to determine which option works best. Since sample size is your Achilles’ heel right now, it’s best to stick to testing only two options at once.
Break your multivariate test down into groups of two.
You can, however, break a multivariate test down into groups of two. Then, run your tests like a NCAA bracket, pairing off two options at a time and selecting the winner of each option to move on in the contest.
That’s a great way to handle multivariate testing when you’re working with limited traffic.
7. Try Sequential Testing
Instead of running your control and variation tests concurrently, try running them sequentially.
For example, run the control for a week. Then, run the variation for the next week.
Afterward, compare analytics for the two tests. If you noticed that one option performed significantly better than the other, then consider using that option going forward.
There are a couple of caveats here, though.
First of all, the test might not have statistical significance because the site has low traffic. So you’re going to have to make a judgment call.
Second, be sure that the two tests are run during similar business cycles. For example, you wouldn’t want to run the second test during the Fourth of July holiday when there was no holiday during the first test.
8. Crowdsource Your Feedback
If you’re looking to get feedback about the usability of your site, there are plenty of tools online where you can do that.
Basically, you provide people with an image of your site or its URL and they comment on it. The feedback can be valuable, but keep in mind that it’s a self-selecting poll (that is, people are deciding to participate on their own).
Here are a few options for you if you’re looking for quick feedback on your site:
9. Read CRO Case Studies
If you can’t be practical, be academic.
You can learn from the tests of others when you’re not confident in your own tests. There are plenty of CRO case studies that will give you a wealth of information about how to optimize your own site.
Keep in mind, though, that it’s best to learn from case studies in your own industry. If you’re running an e-commerce site that sells men’s fashion products, a case study about optimizing a landing page that sells diet pills won’t necessarily apply to your audience.
Also, some case studies are garbage. As of this writing, the first link in the list above highlights four case studies that produced inaccurate results (at least, in the opinion of the author).
The point is this: if you’re going to go academic, go all in. Don’t just read one case study and consider it gospel. Read several of them, draw comparisons, and form your own conclusions.
10. Do the Other Optimization
Perhaps I’ve buried the lead here. If your site isn’t getting enough traffic for a test, then maybe that’s the problem.
It might be time to do some search engine optimization (SEO) and put the conversion rate optimization campaign on the backburner.
A good SEO campaign will bring more visitors to your site. Then, you’ll be in a better position to run effective tests.
If you’re not comfortable handling SEO yourself, get with a reputable SEO firm. Sure, it will cost you some bucks, but that’s part of the investment required to be successful in digital marketing.
Wrapping It Up
Just because your site has low traffic, that doesn’t mean you should cast aside all hope of optimizing it for conversions. Even with limited traffic, there are several ways that you can conduct split tests and improve the user experience. Of course, you should also look at ways to increase your traffic.
