is the second data was something new or similar.
if it is new check whether the data is clean for the model to process.
if the dataset is similar I think the model will learn little from it.
I think that the cause can be in data