Combat or integration?

[quaresma00]

I readed your '05_What's the difference between Cellranger aggr, Seurat merge and integration?.md', i'm not sure whether i understand.
If samples come from a same experiment ,we should use batch correction method like ComBat. And, if the samples come from multiple experiments, we should use data integration method. Is it right? If there are 10 samples in a GEO series, should i use the batch correction method or integration method?