In this project we had good work done by
Posted: Wed Dec 18, 2024 4:24 am
Before working with me, the customer used stock images and creatives drawn by the designer in the ads, but not all of them gave results. By the way, I also tested the designer's images. We could launch 200-300 similar creatives for testing per month, but they worked poorly.
In general, we launched each hypothesis for A/B testing. In one ad set, we tested several ads with different visuals at once. The best of them were collected into a new ad set and tested again. And so on until we derived the very formula that creates conversion ads :
1. We use unusual pictures and sometimes videos as targeted industry database creatives . We use the latter to improve the quality of the ad - this is an internal hidden parameter in Yandex, which is not available in Google Ads.
Pictures created by a neural network and retro-style visuals. The latter gave parents a sense of déjà vu and cut through the information noise quite well. Examples: interfaces of old programs, mockups of push-button phones, screenshots of WhatsApp or TG correspondence.
Images with text. Tests have shown that such creatives perform better when ads are shown on mobile devices. The user first catches the visual with his eyes, then reads the text on the picture, if there is one, and then the text of the ad. The text on the picture increases the likelihood of reading the ad and subsequent actions that interest us. Therefore, I recommend putting the strongest offer on the picture. For example, we duplicated the main offer: “Applications are open for a free programming course for children.”
2. Use all the space to the maximum to highlight the ad in the feed:
In the space under quick links, list the benefits. They will strengthen the key offer and attract the attention of users. For example: "free course", "for children aged 9-17", "only 20 places".
In the promotional campaign, we duplicate the offer if there are no other benefits. The purple color of the text will highlight the ad against the background of other offers and increase the size of the ad in the search results, which will help to break through banner blindness. Example: "Python course for children from 9 years old for free!"
The mockup idea worked best: the outdated computers stood out from other creatives and broke through banner blindness. The AI-generated images worked because they were unique. But the stock images worked the worst precisely because everyone uses them. Users don’t notice them.
In general, we launched each hypothesis for A/B testing. In one ad set, we tested several ads with different visuals at once. The best of them were collected into a new ad set and tested again. And so on until we derived the very formula that creates conversion ads :
1. We use unusual pictures and sometimes videos as targeted industry database creatives . We use the latter to improve the quality of the ad - this is an internal hidden parameter in Yandex, which is not available in Google Ads.
Pictures created by a neural network and retro-style visuals. The latter gave parents a sense of déjà vu and cut through the information noise quite well. Examples: interfaces of old programs, mockups of push-button phones, screenshots of WhatsApp or TG correspondence.
Images with text. Tests have shown that such creatives perform better when ads are shown on mobile devices. The user first catches the visual with his eyes, then reads the text on the picture, if there is one, and then the text of the ad. The text on the picture increases the likelihood of reading the ad and subsequent actions that interest us. Therefore, I recommend putting the strongest offer on the picture. For example, we duplicated the main offer: “Applications are open for a free programming course for children.”
2. Use all the space to the maximum to highlight the ad in the feed:
In the space under quick links, list the benefits. They will strengthen the key offer and attract the attention of users. For example: "free course", "for children aged 9-17", "only 20 places".
In the promotional campaign, we duplicate the offer if there are no other benefits. The purple color of the text will highlight the ad against the background of other offers and increase the size of the ad in the search results, which will help to break through banner blindness. Example: "Python course for children from 9 years old for free!"
The mockup idea worked best: the outdated computers stood out from other creatives and broke through banner blindness. The AI-generated images worked because they were unique. But the stock images worked the worst precisely because everyone uses them. Users don’t notice them.