I think that a lot of the poor perceptions of "mapping companies" come from a combination of smoke-and-mirrors and the placebo effect.
I'm all for tinkering with engine management, particularly to take benefit of things like higher RON fuels into account. The M54 is mapped, slightly conservatively, to run on 95RON fuel - the norm for most European markets. While you're probably not going to see significant headline gains from mapping to run on 97RON+, it may help to smooth out some flat spots and increase the area under the torque curve a smidge at the bottom end.
In fact, even if not to run on higher-octane fuel, some small driveability gains are eminently possible with most NA engines. However, the process is expensive relative to FI engines, where the mapping process is often to simply increase boost and fuelling to the extent of the existing safe ECU learning parameters for headline power increases.
For an NA engine, there's much more fine-tuning required, needing almost an individual map for each customer (based on observations from the standard map/performance). This is why there's so much scepticism about mappers who promise NA gains (be that power or driveability) for modest fees, as the likelihood is that they are just chucking a generic map on and letting you drive away. Imagine a 3D graph showing all possible permutations for throttle position, rpm and air/fuel ratio - that is a simplification of we're working with here. I personally wouldn't trust anyone to just chuck a generic map on an NA car and hope for the best.
The crux of the issue is that we as humans are built to post-rationalise and self-reassure (a bloke called Dan Ariely is a well-published author in the area). As such, we can drive away from one of these cut-price 'mapping' sessions, and we'll be convincing ourselves we can feel a difference, and that any observed differences are a clear improvement on how it was. It's just how the human brain works.
If tuners were more transparent about the process, doing things like before/after dyno printouts to demonstrate improvements, it would help significantly with the trust issues inherent in the industry. However, most 'tuners' who chuck a generic map on probably don't have access to a decent dyno, and most customers paying a knockdown price don't want to spend £100 on two dyno runs. Instead, they would prefer to use a 'bum dyno' and convince themselves that it was worth the expense.
