1. Look beyond conversions

First, you can choose to look at a different KPI (key performance indicator) instead of conversions. For example, instead of focusing on the number of product requests, pay attention to the number of click-throughs to a desired page. You can also choose to focus on an earlier moment in the sales funnel. For example, test not the customers who actually buy a product or sign up, but customers who put a product in the shopping cart. If everything is measured properly on your site, you can easily see in your data what behavior converters exhibit prior to their purchase.

2. Good research for indispensable information

Next, with small numbers, it is extra important to do good research. You want to understand your (potential) customers as well as possible. What behavior do your customers exhibit? Can you define segments in this? Why do they visit your website? And what products are they interested in? Of course you can partly read this from your own web data, but there are also plenty of other sources to supplement this information.

Using online surveys through tools such as Hotjar and Usabilla it is possible to ask questions of your customers online (for example, after check-out) to get answers to the above questions. But face-to-face interviews can also provide valuable insights. Or visit customer service or sales departments. In addition, there are plenty of other methods (clickflows, eye tracking, etc.) that give you information about your visitor's behavior.

After collecting all the data, you get to work analyzing general user behavior and dive into the (scientific) knowledge available about your industry. What does the market look like? Are there any new developments going on? Who are your competitors and what do you see happening there? Based on all this information, knowledge of behavioral psychology and the insight you have into the behavior of your visitors, you create targeted hypotheses.

3. Small numbers, big changes

Whenever possible, of course, it's always nice to validate your hypotheses. When there are not enough conversions for an A/B test, you can also choose to put changes live immediately. With all the information you have, you can often already partially reason out whether there is a chance of success.

If you want to know what and how big the effect of your adjustments is, you can opt for a 0-1 measurement (or a before-and-after analysis). You analyze the period before and after the new elements go live and then compare them to the year before. Always make sure that the changes are big enough to have a clear impact, though, and take into account any marketing campaigns or other major changes in the market. These can cause large discrepancies in your data.

Suppose you put your changes live on the first Monday of the year. After about 3 weeks, you compare the results of the test period with the weeks before, the control period. This will look like this:

In the test period, you can see that the conversion rate went up by 15.6%. An indication that the changes have been positive. But this uplift could also be caused by the period of the year. Therefore, to rule out these seasonal influences, you also look at the same period of the previous year. In the example, you can see that in 2016 the conversion rate was actually lower in the test period than in the control period. This is another indication that the changes had a positive effect on the conversion rate.

Conversion Optimization is for everyone

No matter how many or few conversions you have, structure is indispensable in a good optimization program. Make sure the right elements on your site are measured and you have enough information about your visitors' behavior. Really try to understand how and why your visitors behave in a certain way. Only with this knowledge is it possible to build targeted hypotheses and thus measure the impact of your changes and actually turn your results into insights.