A few days ago I wrote an article to realize and Google run UAC like lying flat cast Facebook, specifically made an experiment.Test: Vote for Facebook ads like UAC
Experimental content: single account only run A + AC, other advertising series are not on, every 2-3 days to the advertising campaign inside a supplement advertising material, the purpose is to continue to iteratively increase new material, close the old decay material to achieve long-term campaign stable delivery.
Experimental results: did not succeed, and finally found that the old A + AC inside the new advertising, but also grab the old good material, but the old material cost will continue to rise.Continued to test 2 rounds of new material, and finally found that the material on the post can not even get exposure, and finally had to create other A + AC ads, re-add these new materials to the new campaign, and so on.
Guess the reason for the failure:
1, The logical difference between UAC and Facebook A + AC is still relatively large, UAC is to take the material inside the overall combination of new ads, but A + AC itself is a single ad run by itself, Without a combination, there is a clear tendency in the current ad system toward “material as location,” which causes a single ad to clearly learn a more vertical demographic (at the adset level), and new material uploads may not necessarily match the demographic learned by the old ad (at the current adset).
2, My product itself belongs to a broad crowd of products, pan-entertainment products in line with the crowd in the adset learning will be scattered more serious, different adset learning out of the crowd will be different.
3,There is no continuity in my material, this time the material obviously uses a large difference in the attributes of the material test, the learned crowd will not match the previous material run out of the crowd.
4, Facebook’s own advertising system logic, the attribute label of the crowd is very strong, resulting in the algorithm tends to find the front transformation of the crowd similar to the crowd exposure, improve the exposure conversion rate.
Several factors added up to make this round of testing ultimately ineffective.
Follow-up thoughts:
1, For this product, can only follow the logic of the wash campaign, regularly new, if there is a similar vertical material, you can continue to try by adding.
2, Advertising material copy, back to the old road line, run a period of time material decay to directly copy the corresponding ads directly, In the case of the same material attributes, copying can make a new wave (covering the previous correct crowd but may not have downloaded it at that time), and it has been shared before, copying the material can also solve the problem of poor reviews (many products are actually louder).
What products may work: Vertical products, I think it is still possible to try. The materials are relatively continuous, and they all cover the people who conform to the product. This kind of material can continue to broaden the possibility of adding to the original population, and the conversion rate of the crude oil people will not lose much.