Sequential fitting works well for when one set of phases will be fit to a series of individual histograms. Not all phases need to be present in all histograms, so phase transitions will be fine. However, here is a case that GSAS-II cannot handle at present: Suppose that you have a series of datasets collected as a function of temperature, and you have multiple datasets at some or all temperatures, and you want to do combined fitting at temperatures where you have multiple datasets. At present, combined fits require a separate refinement. The collection of multiple datasets under a single set of experimental conditions is is a routine case for TOF diffraction (other than with Powgen). Combined x-ray/neutron parametric studies are likely to become more common. Likewise, for parametric resonant scattering experiments. Allowing sequential fits to use groups of histograms is something that I am slowly working on.
Another possible case for a new type of sequential fitting would be for where one wants to compare how different models fit a dataset (or a combined fit against multiple datasets.) This is not something that I am currently planning to tackle, but do want to get to eventually. One possibility for implementing this would be that one creates a series of project (.gpx) files and fits all of them. This needs a program that can be used to provide an overview to a series of projects. That has been started called G2compare.py but is currently not yet in a usable state. (This was sort of modeled after my VAX program CMPR, which was later made into a Windows/Linux/Mac tool, but that too is also now an antique.) The other way that this might be implemented would be to handle this as a different mode of sequential fitting. Stay tuned to see how and when this is worked out.