In this particular case you're also looking "closer" at an interaction than you can model with the particle intuition. For simple electron/photon scattering you can pretend that things are particles and use that intuition to guide you but when you start looking really close at things you can't really do that anymore and have to work with fields and waves in a way that belies normal physical intuition. Part of the way that manifests is "virtual particles" popping in and out of existence (really field interactions that aren't really particle-like but with the virtual particle trick can be modelled almost as if they were). These interactions do affect the real world and the way they affect it depends on the fields involved. If there's some unknown 4th generation of lepton for instance a virtual one of those could interact with the experiment, but since you didn't know about it you didn't include it in your calculation.
Also depending on how you're calculating things you aren't just saying "1+2=3", you're saying "1±theory error + 2±theory error = 3±theory error" so you end up with a range of calculations. That might manifest as "well if there's a graviton then I need to calculate it this way but if there isn't then I need to calculate it this way, and we don't have experimental evidence to guide me one way or the other".
Lastly, QFT is quantum mechanics + special relativity and I think it goes without saying that both of these are just detailed, difficult fields with difficult maths involved. So you're taking an already difficult field and since you're trying to be accurate out to a lot of decimal places while interacting with messy real-world experiments you're forgoing all of the nice physics-class "massless particle in a spherical box" approximations that normally let you skip over all of that trouble.
See https://indico.fnal.gov/event/7309/contributions/101335/atta... for some lecture notes on the base problem.
0: https://en.wikipedia.org/wiki/Path_integral_formulation 1: https://en.wikipedia.org/wiki/Feynman_diagram 2: https://en.wikipedia.org/wiki/Effective_field_theory 3: https://en.wikipedia.org/wiki/Renormalization