7 Tips For More Website Earnings - 7. Testing
 |
It's about testing and the best way to do it to increase your income.
I have already addressed the topic a few times, but I find it important to do this again in detail here since you can never really get the most out of your sources of income without testing.
Parts of the article series:
Why Test?
 |
| 7 Tips For More Website Earnings - 7. Testing |
The first question that often arises is why testing should be done in the first place.
After all, on the web, including on this blog, there are always articles with best practices and tips that you can easily implement and that's good.
However, these tips and guides are just a good introduction. As a rule, they shorten the path to the goal, but they will never be enough on their own to find the optimal solution for your situation.
Every website is unique. The combination of many factors, such as layout, content, style, images, target group, industry, topic, etc. ensures that no website can be compared 1: 1 with another.
And that means that a procedure that has produced very good results on website A can produce more or less different results on website B.
You should therefore use tips, instructions, and best practices as a starting point, but then definitely carry out your tests.
Test Opportunities
What test options are there?
There are many in and of themselves, but not every change is a meaningful test. Just because you have read a good tip on a website and then simply build it into your own and then decide after 4 weeks to see what it has achieved, that is not a valid test.
The term “test” not only includes changing elements but also accurately measuring the effects. The meaningfulness of the test results is crucial, otherwise, you could read the coffee grounds straight away.
That is why you should always try something different but in a measurable and meaningful framework.
Split Test
The simplest method is the so-called split test (split-run test). This is also known as the A / B test.
The term A / B test already gives an idea of what it is about. You compare 2 variants and at the end, you can see which of them works better.
It is important to note the following points:
- A Change
To be able to judge what worked better, you can only make one change in the two variants. So e.g. just change the heading and otherwise leave everything the same.
If you change several things, in the end, you cannot judge which of them had positive effects and which maybe even negative. - Contemporaneous
You have to run the two variants of a split test in parallel. Only then are the results comparable. If, on the other hand, variant A is displayed 1,000 times and then variant B 1,000 times, important framework conditions may have changed.
For example, A's 1,000 fade-ins could have happened in the morning, while B's 1,000 fade-in happened in the afternoon. And there are sometimes different people on the Internet.
It is crucial to always display the variants alternately. - Significance
A statement about the results can only be made if a statistical relevance is achieved.
On the one hand, this can be ensured by simply making enough overlays and, for example, measuring enough conversions. It is difficult to say exactly what is “sufficient” in individual cases.
That is why there are tools such as on visualwebsiteoptimizer.com, where you enter the displays and the number of conversions, and then the tool calculates whether the test results are clear or whether the test should continue to run.
On the one hand, split tests can be carried out by hand. I have described the measurement of clicks (in this case the conversions) in the article “Measure clicks - meaning and implementation”. The evaluation then takes place manually by comparing the displays of the two variants and the number of conversions with the tool linked above. However, this is quite time-consuming.
Alternatively, there are various providers of test tools. One of the few that is free comes back from Google, of course, and is called Google Website Optimizer.
With it, you can create A / B tests step by step, and then you only have to integrate the output code into your pages.
Google then collects the test data and outputs it in an evaluation.
The following is a short introduction to the Website Optimizer in English.
I will present this tool in a separate series of articles. It's not perfect, and it certainly lacks many of the features that the paid solutions offer. Still, in my opinion, the Google Website Optimizer is quite useful for many applications.
Advantages/Disadvantages Of The Split Test
Split tests generally have a few advantages and disadvantages:
Advantages
- Split tests are quite easy to implement. The Google Website Optimizer is easy to use and only has to change one element, which saves time and money.
- Split tests work with relatively few visitors. Of course, you have to achieve significance and show a certain amount of traffic for this, but you can still use a split test with relatively low visitor numbers.
- The results are clear. Since there are only 2 variants, the results of a split test are always clear if they are significant.
Disadvantage
- A / B split tests often take a long time, especially if your traffic is not that high. You keep testing the winner of the previous test against a new variant. That drags on.
- The interaction of individual elements can hardly be tested. It is often said that the result is more than the sum of the individual parts. With a split test, you can hardly measure whether the combination of different elements brings in more than the use of the best individual elements. Example: The heading A and the text C are not individually the best in the test, but together they give the best results. This can only be found out by chance with A / B tests.
- As already said, the Google Website Optimizer is very beginner-friendly, but this comes at the expense of many professional features, which would make the test results even more meaningful.
Multivariate Test (Multivariate Tests)
The disadvantages of the A / B test are partially eliminated by the so-called multivariate test.
You don't just test 2 variants, you create some combinations.
For example, you could create 3 different images and 2 different headings and then test them. That would be 6 different variants in combination:
- Heading A + picture A
- Heading A + picture B
- Heading A + picture C
- Heading B + picture A
- Heading B + picture B
- Heading B + picture C
This method makes it possible to find out the best combinations.
These tests can also be easily implemented with the Google Website Optimizer.
However, due to the many possible combinations, you should have enough traffic to achieve statistically significant results for the individual variants.
Only The Facts Count
In the end, A / B tests and also multivariate tests have a right to exist, especially when emotions and opinions are involved.
Just because I think variant A is better doesn't have to apply to my target group.
So it is better to test than to engage in discussions. The concrete results are then decisive.
Of course, you should also test when you are not discussing with someone. You can increase your income significantly if you test again and again and find out, for example, the most profitable affiliate banner, the best place for AdSense ads, etc.
Post a Comment