I think exactly the same as you. I also mentioned this in a previous form title. Newly uploaded models are ranked last in long tail and direct searches, that is, among the models that are directly linked to them. But that's not the only problem.
Let's analyze the status of your “Moai Ultimate Low-poly 3D model” model in the search results.
First of all, we need to point out that the problem started after June 2023 (I'm not exactly sure, but it could be 1-2 months earlier or later). So, if the model was loaded before this time, we can say that everything is actually working fine.
As for the "Moai Ultimate Low-poly 3D model" model, you uploaded this model months after the problem started, that is, during the period when the search algorithm was broken.
Problem 1: Irrelevant products are ranked higher.
First of all, before moving on to the problem of uploaded models being in the last place compared to other competitors, we should address the problem of irrelevant models/products being more prominent in the search results. Solving this problem without solving the other one will not make sense. Let's examine a few examples. If you pay attention to the "moai" search results we made targeting the "Moai Ultimate Low-poly 3D model" model, you will see that the following models/products are ranked higher:
“Zbrush - 20 Rock Formations and Stone Sculptures v03 3D model”
“Easter island 3d model terrain 3D model”
"The Division Bell Tribute 3D print model”
Since AI has been integrated into CGT searches, the first advantage we expect AI to provide is that it makes inferences from the user's search and brings to the fore the most suitable models in the database. However, we see that the titles of the models (even though they should be the most important element in searches) have the same or lower priority than the descriptions and model tags. The 3 models that appear in the "moai" search above are not directly related and are ranked higher than the product we are actually targeting, just because the word "moai" is mentioned in the description. (Note: This word is a search term that already has quite a few models, whether relevant or irrelevant. When you do this search, you will see that only 136 items are listed. In other words, the problem we are talking about increases to incredible levels in more competitive searches.)
I wish this was the only problem for problem 1. Please also check out this status in other searches:
In searches containing 2 or more words, if there is a model that contains these search terms separately and is both older and more successful, it is shown higher. In other words, it is very important to make separate matches instead of direct matches with the search. We would expect AI to solve such problems, not cause them…
For example:
When you search for the word "train interior", you will see that the product "Gym Equipments Low-poly 3D model" is on the first page and at the top. So why? Because the words "training" and "interior" are included in the tags. So, in fact, when you search for "train interior", this is the product. It's pretty simple for AI to understand that the product is not about that. Do many different searches and you'll see similar situations pop up even more frequently in some searches.
Problem 2: You will come in last place no matter what.
If you uploaded your model after June-July 2023, It doesn't matter how good is the keywords search, analysis, high quality and abundant preview images, detailed and effective description texts, etc. You will be in last place among the models competing directly with you. There doesn't seem to be any escape from this situation. In addition, when you click directly on the 3D models section in CGT without any search, the list of models that we call the most successful models, which generally change never or very rarely, was almost the same 1-2 years ago (that is, when the search results were working as they should). So some people say these need to be changed, but I think there is no problem there and new updates are not the reason for this general listing. We need to understand this so we can better identify problems.
Please review all links posted by @jaguarbestproduction. While all the products uploaded before the date I mentioned were very good and had the required rankings, we must say that those uploaded after this date had no chance. Moai is a good example here, but a risky one. Why is it risky? Because the sample has too low competitive (136 models) to be accurate to show
Believe me, I want CGT to work correctly and as it should. I want to see that the conversations we have here are taken into consideration by the CGT team, because I think one of the biggest criteria for CGT to become a more robust asset market is "Search Results". Yes, there are too many assets now and this is true. It can be said that it is very difficult or even impossible to analyze whether preview images fully represent the product in question by a human or by an algorithm that does not use AI, but AI can now scan and analyze both texts and images very effectively. This is truly a great advantage, but unfortunately we have not yet been able to take advantage of it in CGT. I hope this problem will be solved as soon as possible.