First - how does response time analysis of your service look like? Can you share it and also include a pure path?
Basically, I’d recommend putting a custom service (or services) in the application code to get a higher granularity unless it is a very simple service. If your service code calls some other methods, try to define custom services for those methods (if it makes sense of course).
Did your devs explain what do they mean by “applicaiton method is being tied up with SQL”? Recently I’ve encountered a similar case when identified issue was a large part of code handling database requests and was also defined as synchronized (this can be seen as lock wait time in the response time analysis). This caused massive slowdowns of requests (average response time is ~200ms, slowest 5% ~2s, some individual requests up to 1 minute and 99% of the time is lock wait).
Unless the response time analysis shows a large portion of time is being spent in DB or network, then it’s likely a code issue.