Yes - I agree, the point is that most of the time an Excel "app" is solving a very specific problem using data coming from some other (internally managed, usually legacy) system.
Like: I run a report, transform the result in .csv with some ultraedit macro, slurp it up in Excel and off I go.
(off I go can often being something like: upload the result to the same or a different legacy system, again in some custom format).
Maybe I can pester my own IT dept. to add an extra option to the report so that it spouts off a .csv directly. That's all, the rest still lives in Excel.
Then I am free to play with data as much as I like (take also in account that in some cases you want these in a Excel just to be able to manipulate them better, while the old app works record by record, you can make changes across a thousand records using the Excel interface... and keep also in mind that this is maybe something you need to do once or twice a year, when you renew your catalog prices or whatever).
Now, what is the cost of:
- going out on the market looking for a SAAS that can get my data exactly in the format I use internally, applies the required transformations and send the result back in a format I can use
- assess that the SAAS vendor is indeed trustable for my desired level of security, SOX compliance, etc.
- add one more vendor to my portfolio of vendors/licenses/purchasing orders
So the reason not to go to SAAS is actually a combination of security/external resource dependency/cost/bureacracy.
Where "cost" is mostly inertia+sunk cost fallacy. I am sure that a SAAS may cost less than the hours spent manually doing all the stuff I mentioned above (including mantaining the Excel spreadsheet) but the latter is a "hidden cost" because it something that happens infrequently and is part of the normal chores of whoever is using the Excel sheet itself.
While introducing a new SAAS app will be an IT cost (to identify/approve it, add a recurring subscription etc.)