One word: observability. Normally programmers don't care or consider observability, so they write services that just fire off HTTP requests and use the result. No effort is made to log these calls, or the result. IT teams can help here, by capturing HTTP traffic and exposing it through tools like Splunk. But if you can get your programmers to use a message queue, then you can much more easily inspect API calls in flight, making it easier to find the source of an error that could exist somewhere across multiple services.