The vendors and utilities have core simulation codes that do a very good job of predicting cycle length, reactivity and local power distribution, enrichment depletion, in conjunction with the operation of the plant. There's now about 40+ years of experience.
In PWRs, the plants design the boric acid content in the water, and that is monitored daily. The core power distribution and core exit temperatures are monitored so the operators are aware of the core power level and radial power distribution. It's fairly accurate to the last EFPD (effective full power day).
Some plants design for coastdown (gradual power reduction) and feedwater temperature reduction, and they can run out to essentially no soluble boron in the primary coolant, and the reactivity essentially goes to zero or slightly negative, and then the plant is shutdown with control rods.
However, many times, some surplus reactivity remains, i.e. the fuel is not full depleted because the unit may have had a mid-cycle outage, a trip, or a period of reduced power operation, and the cycle simply runs out to the scheduled end of cycle refueling outage.
The core simulation codes have been benchmarked against experimental data, including gamma scanning of fuel rods, as well as hot cell PIE (post-irradiation examination) in which various fission products and transuranic isotopes are measured.