First, consider the following situation: you have just completed a fairly promising CRO test. Carrying out the task of increasing the number of visitors with their further conversion into requests on the pages of the site, a test was carried out to update the design on one of the pages by means of classical A / B testing. Thus, half of the visitors displayed the old design, while the other half had access to the page after the redesign.
This experiment was conducted during the month and its result was the predicted increase in the conversion rate to 10%. The manual got acquainted with the data obtained in the testing process. The result was the introduction of updated design on all pages of the site.
However, after the upgrade you found a sharp decrease in the number of requests. You made a discount on a quite logical seasonal reduction, but after a couple of months the situation has not improved. Along with this, regular monthly income began to fall. The thing is that the experiment was given too little time to compile a statistical report. Every day the page received about 50 views, respectively, it was necessary to wait for at least 150 thousand views, which would indicate an increase in the coefficient of trust to 95%. Thus, the test should have been at least 8 years old. Incorrect timing calculations resulted in serious financial losses in your company.
Risk related task
The incorrect definition of sample sizes is one of the most common problems in a CRO that marketers have to face. A marketing strategy becomes quite erroneous if it is based on too small a sample. By optimizing the conversion, you must strive to maximize true values. Conducting deliberately incorrect testing it is impossible to obtain statistically correct data necessary in solving the set tasks.
The purpose of this article is to study the most common mistakes made by marketers in the optimization process. When analyzing and studying new marketing techniques, try to take this information into account in order to avoid unpleasant mistakes.
6 serious conversion rationalization errors
- Convince yourself that conversion optimization is about A / B testing
It is absolutely illogical to say that CRO and A / B testing are absolutely identical things. In fact, A / B testing is one of the many types of conversion optimization needed to make the visitor want to perform the necessary actions. You are mistaken if you think that in order to conduct a CRO, A / B testing should be implemented wherever it is allowed. There are many examples in which this technique is devoid of any meaning. An example would be situations in which there is too small a sample that does not allow collecting data. If you are dealing with a page that receives about a hundred views a month, then achieving a serious statistical significance will take a couple of months.
If you still run the risk of A / B testing on a page with low traffic and after one and a half months you complete the test, then the data obtained can certainly be considered meaningless. A / B testing is a serious and high-quality technique, but in no case can it be limited to it. It is important to constantly discover new testing methods and experiment to determine the best one. If you are faced with the task of quickly and qualitatively increasing the conversion, make major changes instead of studying a specific variable. It is possible to use the site Weather.com. Marketers used to gradually edit some parts of one of their main sections, which led to an impressive result – the conversion increased by 225%.
- Do not define conversion goals when working with a specific project.
When studying news about the success of Weather.com, which increased conversion by 225%, do you have questions related to what we mean when we talk about conversions? If yes, then you have great prospects. With the help of conversion rate, you can define different things: leads, subscriptions, visitors, purchases – you need to build on the tasks assigned to the page. If you say that the number of conversions has increased, without specifying at the same time, what is the meaning of the “conversion”, then your statement loses any meaning. Returning to Weather.com: 225% growth meant an increase in the number of trial subscriptions. Knowing such details, the news from the site has become much clearer.
However, after refining the information on metrics, we cannot accurately understand the picture in general. It is important to take into account the time of the tests. In different periods of time, these figures may vary significantly.
Accordingly, even if a few days after the start of testing, you will receive 98% of statistical significance, but in no case should you stop. After a few days, the data may change dramatically. The same applies to the longer term. Information obtained on the basis of tests conducted in December will be strongly derived from similar data obtained in March. Seasonality is an important factor affecting conversion rates.
What else do you need to pay attention to? We recommend to consider the type of devices used by visitors. Using the desktop can be much faster to register on the site, if it includes filling a large form. In the case of mobile phones, the conversion will be much lower. This also includes channels. Try to get maximum information, ignoring the average. If some channels have a relatively high conversion rate, then they need to pay more attention. Remember also that conversion is not the most important indicator when doing business. It is necessary that they bring income. By lowering the prices of your products you will be able to significantly increase conversion, but this will have a negative impact on profits. Since this metric does not always accurately indicate business performance, it must be considered in conjunction with other data.
- Do not perceive statistics
Starting training in the field of CRO, many owners of institute diplomas in statistics or marketing are confident that their knowledge will be enough to conduct tests. Many experiments in this area do not help in achieving goals. CRO is based on statistics. Accordingly, the lack of your knowledge in this area will be a big obstacle when testing, which will negatively affect the effectiveness of marketing strategies. What can happen if you complete the test without typing the cherished 98%? Is there really little 90 for statistical significance?
Of course, there is little and it is connected with the following: Imagine that this is not about statistical significance, but about setting a bet. Do you dare to put on the result, the reliability of which is 90%? Most likely no. It is better to constantly improve your knowledge in the field of statistics. It will have to spend a lot of time, but this will increase your professionalism and, consequently, efficiency.
- Do not conduct tests on pages that already show good results.
If you have been observing the positive work of a campaign or a single page for a long time, then this is not a reason to stop conducting tests and experiments. It is in them that there is a huge potential for reaching even greater heights. For clarity, we study a few examples from the company HubSpot.
“Historical Optimization” is a large project implemented under the leadership of Pamela Vaughan. The main idea was to constantly modernize and duplicate outdated entries in order to increase traffic and get leads. In this case, the work was carried out not with all the records, attention was paid only to those who had previously brought good results. After reviewing the reports on the experiment, Pamela stated some interesting facts:
- during the month, about 70% of the views of the entries related to the updated posts;
- 92% of the volume of leads for the month gave the same updated records.
Why did this happen? The posts described above were well ranked in Google before the experiment, bringing a lot of traffic.
Among the tasks set there are two main ones:
- whether it is possible to get additional leads from records with initially good traffic and insufficient conversion;
- Is it possible to get traffic from records with good conversion and not high traffic?
Thus, it was possible to increase the number of leads by processing old posts.
Marketing Acquisition employee Nick Barosso began to audit the HubSpot site, paying special attention to the leads. It is noteworthy that over 50% of entries with high traffic led visitors to inefficient offers. In order to improve the conversion rates on these records, it was decided to conduct a test in which the main CTA-link was changed to the one that sent to the record with a good conversion. After 7 days, the edited record brought 100% more leads than usual. Accordingly, it is not always worthwhile to focus your attention on projects that require solid work. Often, good results can be achieved from pages that initially have high traffic or a sufficient number of leads. Optimization ultimately leads to an obvious improvement in the situation.
- Ignore research, relying only on tactics
Always in matters that relate to the CRO you need to remember the main thing – the processes. Do not waste time on working through individual optimization tactics; stick to a systematic approach. Always the first stage of work should be research and in-depth analysis. This will help to find the main problems and work out a strategy for solving them. Let’s go back to the HubSpot example. Only after careful analysis, Nick Barosso was able to make sure that posts with high traffic are related to offers with low conversion. How to find information for analysis? Marketer Paddy Mugan recommends that you look in more detail at how sales go, examine the funnel, determine the points at which the most failures occur and try to figure out the reason for this.
The main steps on which optimization is based:
- Building hypotheses.
- Identify the optimal solution.
- Experiment to verify the correctness of the decision.
- Study of the results of the experiment.
- The following experiment.
Do not forget to fix the data obtained during the tests, it will help if you need to re-test. You should always remember the last step and repeat your experiment. Be sure to note for yourself the recommendations associated with the possible improvement of their testing methodology.
- Stop further work after failing testing
We advise you to listen to one piece of advice: “You can’t call a test bad until it doesn’t hurt anything. As a result, you will achieve a positive result or discover something new for yourself. ” Do not stop after testing in the hope that the desired result will meet the requirements. Pip Ladja once said the following: “A CRO professional is ready to change his mind at any time.” It is also not necessary to ignore the bad results of testing, the data obtained may be useful in the future.
We hope that this material will become a motivator for improving your work results. Consider the errors described above, this can save both personal time and the resources of the company for which you are performing tasks.