How about this, instead labelling something as overrated, start with a more positive and differentiating point of view:

- when to use unit-tests ?

- when are other tests more useful ?

- which bugs can unit-tests find ?

- in what situations is changing code to make it testable some-times beneficial ?

- when is it a complete waste of time to make code unit-testable ?

Things to consider when answering these questions above:

- what is the customer impact if there is bug ?

- what is the developer impact if there is bug ?

- what is the developer impact writing tests ?

- how quickly can we fix it and release a new version ?

- how quickly can we fix the hotfix in case we messed up ?

- how can we find out if there is a bug ?

[...]

I think it is great that he starts thinking about the value of unit-tests and testing in general in his particular context.

But i think it helps keeping an open mind knowing that some approaches work better than others depending on the context.