Why is the cost of AI data collection so high?

[Lunaproxy]

I don't know why, but I feel that the cost of my data collection every time is extremely high. It may be largely related to the fact that I don't have clear requirements. Also, I haven't made good use of data sharing and open-source data. Can anyone give me better suggestions regarding the data collection process? Currently, I'm collecting data related to e-commerce. Or if you have better methods to save costs and improve efficiency, your guidance is most welcome, and I would be extremely grateful.