Will the selected ad work? Is it worth increasing the budget for an advertising campaign or is it better to redistribute it? A/B testing will provide accurate answers to these questions if you do it correctly.
What are A/B tests in AdProfex
This is a tool which helps you to constantly experiment with ads, comparing the effectiveness of each of them. The goal is to determine the creative for an advertising campaign that will take the bulk of traffic and will not compete with other ads.
For example, through A/B testing, you can compare creatives with the same text but different images and find out which one will get the higher CTR.
- The tool allows you to test different variants of teasers based on the already existing one.
- At the end of the split, the creative with the highest CTR remains. It is determined by the greater number of clicks with an equal number of impressions.
- With A/B testing, audiences do not cross and ads do not influence each other, so you will get objective results.
- The function is available for all GEO.
- You can compare 50 teasers within a campaign at once.
It is worth remembering that CTR is an ambiguous indicator (not flexible), it does not guarantee profitability. But if your advertising campaign is set up correctly, that is, it reaches the target audience of the offer, the offer itself is well presented on the landing page, then a high CTR shows that your creative is good and the way the offer is presented on it attracts the audience.
Of course, there are nuances. A/B testing will not be correct if you use misclick (promises of gifts that do not exist and other unfulfillable theses) or visual tricks, for example, in the form of a “play” icon. Yes, you will get a high CTR for such ads, but this will not bring conversions or approve.
What does the testing process look like?
- Log in to your platform account.
- Go to the “Ads” section and select the desired teaser.
- Find the “A/B tests” button next to it.
- Think of several alternative options for the selected ad: change the text and/or image.
- Set the number of clicks you want to receive during the test, we recommend starting from 1000. This is necessary for the ad to achieve a statistically significant result.
Please note that this may happen earlier, that is, it is not necessary to get 1000 clicks; the system may achieve the indicators it needs earlier. But this does not always happen, so a limit of 1000 clicks is needed so that the test still ends when it reaches statistical significance, and not the click limit.
If you tick the “do not complete the test automatically”, the ad will run until one of the options reaches the specified limit. If you do not tick, the system will choose a winner and end the test when it receives enough data, that is, possibly without reaching the limit you set. Once the specified value is reached, the test will stop. The speed of running an ad depends on the current reach of the ad.
- Launch and wait for the result; if necessary, testing can be interrupted. Then it will have the status: stopped. If the test went as planned, its status will be «Completed successfully.»
- The system itself will select the option with the highest CTR and leave it in operation. Other options will be stopped.
How much does performance improve?
A/B testing alone does not guarantee improved results. But it helps to identify ideas that are more attractive to the audience for less money.
Let’s see what results A/B testing can bring us in practice and what conclusions can be drawn from them.
So, we took two flows on one news.
We used 1000 clicks on it and received a modest CTR of 0.8%.
Here it is.
We prepared 3 options for testing.
Option 1
Replaced the picture, left the title.
Option 2
We left the picture, but changed the title a little.
Option 3
Replaced the image and title.
It was the last option that showed an excellent CTR — 3.2%.
What does a creo with a higher CTR give us?
- we can lower the cost of a click by the percentage by which our CTR has increased and buy traffic cheaper for the same volumes;
- we can leave the price the same and buy more traffic.
The choice depends on our tasks.
However, you need to remember that A/B testing does not guarantee an increase in profitability. New creo variants may have a negative impact on the advertising campaign. Therefore, you need to check the statistics after the winnning of the new ovariant in the A/B test. Pay attention to the creo variants with the “play” icon, they attract attention and encourage the user to click. But often users are not interested in the content, they do it simply because of the icon. Thus, you get false test results and low conversion due to misclick.
Why you can trust AdProfex test results
You can compare creos manually, or you can use an AdProfex tool. Will there be any difference in the quality of the results?
Pros and cons of manual and automated testing.
| Manual A/B testing | Automated A/B testing | |
| Time | Time-consuming (you need to monitor the progress of the test) | Saves time |
| Errors | Does not exclude spontaneous errors | Eliminates errors when determining results |
| Process | Double-checking results is a time-consuming process | Allows you to quickly double-check results |
| Reliability | Causes incorrect results due to traffic differences | Gives greater reliability, since the tests are carried out under the same conditions. |
When teasers are tested by the AdProfex algorithm, audiences do not cross and ads do not influence each other, so the results are objective. If the winner is determined by the higher number of clicks with an equal number of impressions.
4 golden rules for conducting an A/B test of a creative
- Don’t be afraid to experiment: use different titles and images, including unexpected ones.
- Look for inspiration in other showcases and ads, pay attention to how headlines are constructed and creos are selected. Sometimes something completely unexpected can work.
- It is not at all necessary to change the creo completely. Practice shows that replacing one word can seriously affect CTR, while a complete replacement may not give anything.
- Don’t do an A/B test just for the sake of an A/B test, do it when you have good ideas that you want to test.
So, A/B tests are a simple and informative tool that is already available in the AdProfex native network. To test it, you need to register, create a campaign and launch your first ad.