This is a confusing, vague question but I'll give it a shot
First thing to consider is whether you want success chance to vary with or inversely too your factor for consideration(level, experience, defense, whatever)
your fomula can be a fraction with things that you want to improve your chances in the numerator and things you want to decrease your chances in the denomenator(this can just be 1 if nothing decreases your chances)
Suppose we wnat your chance to recruit someone to go down with your level(so it gets more difficult) but to go up with your strength. Our basic formula so far might look something like chance = strength/level
This would be a linear function. Depending on how you want things to scale you may want a different type of function. If you have a theoretical maximum you could for instance use a logarithmic function to taper off towards some value
chance = ln(strength/level)
With this you get diminishing returns.
You also may not want throw in multiplicative factors or additive constants to tweak the data if it doesn't feel right with how you want to progress.
chance = m * ln(strength/level) + k
Finally since this is a percent chance you may want to try and normalize it and convert into a percent. For this we divide by your best possible values for chance and multiply by 100
chance = 100 * (m * ln(strength/level) + k)/(m * ln(MAX_STRENGTH/MIN_LEVEL) + k)
I hope this helps.