Effect size is strongly affected by severity - people who aren't very ill just don't have as much to gain compared to people who are gravely ill. Widening diagnostic criteria and more liberal prescribing will inevitably lead to a reduction in the observed effect size.

Antidepressants were bona-fide miracle drugs when we first started using them on desperately ill inpatients who experienced every moment as exquisite torture. We saw the most miserable lives completely transformed in a matter of weeks. They have become merely "sorta-kinda useful sometimes" now that we're mainly prescribing them to broadly functional people who are feeling a bit sub-par.

SSRIs are a pretty poor fit for the latter cohort, because SSRIs cause significant emotional blunting in the majority of patients, to the extent that some people hypothesise that emotional blunting is the fundamental beneficial effect. Feeling quite numb is an incredible improvement if you are constantly unbearably miserable. If you have a more normal range of emotional experience than relentless misery, it is likely a sideways move at best; if your core complaint is that you feel numb and apathetic, they're probably actively harmful.

SSRIs are very widely used because of their extraordinary safety, but they're often thoughtlessly prescribed by overworked primary care doctors. There are a wide range of antidepressants (and drugs that have antidepressant effects despite not being marketed as such) that are likely a better option for a large proportion of patients.