Obviously what has been said so far is spot on, every manufacturer varies but they tend to assume at the least that you need a base layer with a clo value of 1 (heavy base layer) and a camping matt or non compressible insulation underneath; at least from the testing methods i have read up on. I will add something that may be of use.
The rest of this is a take it or leave it thing. It has solid basis for applying your own system, but it might give you a headache. Personally i love this kind of thing but i am sitting a physics heavy nat science degree and love theory craft, charts and numbers every bit as much as i love being outdoors. It certainly is going to require that common sense and foresight be applied, so evaluate safety for yourself (just a little disclaimer there, yknow, just in case).
We know every body has different tolerances and we know different variables, in different situations will have differing effects and the results by nature will be subjective. It sounds like there is no way of comparing bags, too subjective, but there actually are reliable ways of
comparison from tangible data. Most insulation values have an insulation rating that is much more universal forthe sake of comparison, for example: fil power, r-value, clo rating are all comparable units with meaning between various products.
From these values you can account for your own variables for a personalised comfort rating instead of say snugpaks variables, you might not start off with a scale for temperatures but you sure can end up with one that is accurate for yourself (or at the very least a rough guide if you cba crunching numbers).
A very crude way of doing it is to look at loft, and this is a very crude example and method. Say, for example, 100mm of loft in 900 fil down. Most bags with the same rating of down and loft are going to be about the same ignoring draughts etc. So long as you give yourself a bit of a margin of error you can estimate an increase or decrease in loft for a static material insulation rating to gauge an appropriate purchase. Say you are warm naked at 20C, you know 100mm of loft gets you to freezing with no wind, you
probably need around half that for 10C in no wind. At the least you know another bag of the same fil rating and loft is going tobe about the same for you with no wind. Just be sure you are comparing the same fil power down for a given loft.
Just keep in mind that there is a myth going around that loft is an absolute indicator for all materials, say 100mm of any synthetic is the same as 100mm of any down. This is not true, ray jardine wrote it in a book a while back and it is dead wrong.
There are more accurate ways to approach this and ways of creating your own system too. From r-value or clo (two proper units of measure unlike comfort ratings) you can find what works for you in a given situation for a given r-value or clo rating. With a bit of trial and error you have a point of reference for temperature value to use between manufacturers. I have kept track of my own logs and have my own sliding scale of what works for me at given temperatures for a given CLO value. Obviously you are always going to have to estimate some variables, for instance heavy wind and no shelter, but imo you are better off removing the variables or over preparing (there is just no realistic way of precisely accounting for things like that). So long as you always wear the same thing inside your bag, keep topatterns like a hot meal [or not] before bed things like your own heat convection/radiation rates are essentially covered automatically.
This can get as complicated and as precise as you want it to be for your use. Originally i used a pre calculated static formula as a guide for clo to warmth,
one that accounts for no real world variables; eventually i was able to come up with my own formulas, by tweaking the already existing static formula to my own findings for different temperature bands. For example, i found that as i got below freezing i needed more clo value per degree celcius to keep warm, and below -5 again i needed even more clo per degree to stay warm. I then looked at the difference, noticed a bit of a pattern and predicted estimates for equal intervals there after (ifi reach those temps and find otherwise i will re-evaluate the pattern).
If you want the standard issue or my modified formulas
as a guide, or to modify yourself pm me or ask. I think, from memory, the standard one is some kind of industry standard for interpreting CLO into an average warmth rating for household use. I got the formula off a government environmental health website that
was well considered and made clear it did not account for variables and differing body shapes (a friend from the town hall pointed me towards it to toy with claiming they use it for reference in UK government); it also appears in many places elsewhere online, usually in an imperial measure format for the USA.
It can be as simple as you like: notes for a rough estimate and future reference; or you can take it to a new level and get to a point where you can actually somewhat accurately predict what will happen.
Additional note if that lot sparked interest:
Just keep in mind to understand what the unit you are using for comparison means. For example CLO value is often given as clo/gm^-2 (CLO per gram per meter squared); so to interpret it you need to multiply the clo/gm^-2 rating of the insulation by the gm^2 rating of the insulation (all layers and absent of shell weight ofc). Just to clarify here m^-2 indicates per meter square in proper formula use, not a minus figure. It is unlikely you will find a clo value opposed to clo/gm^-2 value stated by the manufacturer of insulation as they have no way to know what the finished item dimensions are. In simple format the conversion formula is: CLO=((CLO/gm^-2)*(gm^-2))
R-value is probably more user friendly, i don't know for sure i have not looked into it but it seems to be in common use by people who are not expected to have mathematical fluency or a calculator. I am told r-value converts into clo so you can compare between the two and draw a comparison for a vast range of materials although the formula i was given was incorrect and i have not looked fr the proper version. I use clo because the insulation i use, climashield, is rated in clo on the data sheet and i am familiar with it so there was no point in converting to r-value for me.
Down fil rating probably loosely converts into r-value and clo too, but i don't use it and well, i am not going to spend hours online looking for something that might not even be needed
Hope some of this is of use, if anyone see's caveats in this anywhere please do say so that i can revise my own methods.