We evaluate the result of launching a hypothesis

[sakibkhan22197]

13% conversion rate in response - 1.5 times higher than a regular pop-up;
216 requests for demo access - 3 times more than a regular pop-up;
401 applications for a free course - 4 times more than a regular pop-up.
The AB test is not always appropriate, especially if:

the options are too similar, the difference between them is minimal,
the target audience is too narrow for reliable results,
The time for testing is limited and not enough to collect reliable data.
We've covered how to conduct A/B testing in detail in this article.

It is also important to remember two types of testing errors. The first is to list of afghanistan cell phone number reject a valid hypothesis, believing that it does not work, although in fact it does. The second is more dangerous: accepting an incorrect hypothesis and wasting resources on implementing an ineffective feature. To avoid these errors, it is important to carefully plan and analyze the results of AB testing.

Want to get more leads, but don't have time to study the service?
Contact the Carrot quest growth team. They will find where you are losing leads on the site and set up mechanics that will increase conversion by 1.5–5 times.

Sign up for a consultation
As with the preliminary estimate, you can use calculators to calculate the final estimate. When there is not enough data to use them, you can evaluate the results of the A/B test manually using the Z-test. Here's how it works:

1. Formulate the null (H0) and alternative (H1) hypotheses. The null hypothesis is a statement that the data will not differ between the test and control groups. The alternative hypothesis states that there is a difference between them. After recalculating the data, it will be immediately clear whether the null hypothesis can be rejected. For example, an online school tests two mailings: one with a promo code, and the other with a lead magnet. The null hypothesis states that "the conversion rate to application will not differ between these mailings," and the alternative hypothesis states "a promo code for a discount increases the conversion rate of the mailing."

2. Collect data. For example, if you are testing two newsletters with different offers, you need to measure the conversion to purchase for each of them.

3. Conduct a Z-test. This will help determine whether there is a statistically significant difference in conversions between the two groups. If the difference between the options is less than the established significance level (usually 0.05), then the null hypothesis is rejected in favor of the alternative.