Accelerating ASO with ChatGPT: how we create keywords and texts

Until recently, the process of writing texts for primary ASO was entirely manual for us. This involved reading GDDs or playing a build, brainstorming keywords, researching keywords used by other games, selecting the most effective ones, and determining which ones bring in traffic. Then, an engaging game description incorporating these keywords would be written for the store.

An ASO specialist typically spends two to three hours per locale and platform on this task. However, thanks to neural networks, we’ve reduced this time to just 30 minutes. Neural networks can quickly generate texts for the appropriate store category with a high Confidence level of 0.8-0.9, which indicates the likelihood that the app will belong to a specific category on Google Play (anything above 0.8 is considered a good result). To be featured in the list of similar games, it’s crucial for a project to fall within the same category as its competitors.

We use ChatGPT to assist in brainstorming ideas, swiftly selecting and organizing keywords, and writing and editing texts. Meanwhile, Notion AI refines and polishes the final options. With up to 30 projects going through these tools each month, we’ve already saved tens of hours compared to the manual writing process.

In this article, ASO Manager Zlata Kazantseva and ASO Lead Rustam Saidrakhmonov will demonstrate the pipeline using Survive Squad as an example.

It’s worth noting that this task was performed using GPT-3.5, so the fourth version should perform just as well, if not better.