As a heavy user of DataNitro in the finance world, I can attest to how much it has improved my life. It allows me to interact with my data providers' excel plugins in a much more expedient manner. I could use their direct data feeds, but testing things on those is a PITA.

Largely, though, I agree that DN isn't really necessary/useful but for a small niche... but when you need it, it's EXTREMELY necessary.

In a prior life I used a lot of VBA to process some scenario modeling Excel spreadsheets. Python would have been much better. Python is also becoming more and more common in Financial Services.

I haven't used DataNitro, but it would have helped me in the past, and it fills a very real need. I think the niche may be bigger than you give it credit for.

I think it's comparable to XMLSpy, in how powerful it is, how niche it is, how incredibly useful it is to some people, and how expensive it is. https://datanitro.com/pricing.html

The pricing actually isn't that high if the target is Finance. If anything it may be too low. These are folks who pay ~$20K per user per year for Bloomberg.

I meant in absolute terms. There are people who are happy to pay for XMLSpy, too. Ditto for AutoCAD.

I'm a financial analyst that uses VBA quite frequently (and sometimes unnecessarily, admittedly, I just like to program). I first learned to program in Python and I find VBA not as enjoyable to work with. However, I don't think I could justify going out and getting this product. Very small niche indeed, I'm not sure who would really need it.

Two jobs ago, I built reports and dashboards, mostly in Excel. It was a nightmare. After three years of that, I did some demonstration of Python's capabilities and ended up working on a web-based enterprise reporting system.

These days, I work almost 100% in Python. While I'm not doing analysis much, iPython is awesome for "exploratory programming" - I can quickly build a function to do what I want, thoroughly test it, then import the and its associated tests back into my "real" environment.

The ease of integration external datasources alone makes Python extremely attractive for Excel.

I can easily imagine using iPython to control a small cluster of Amazon instances, using them to process some offensively large dataset (e.g., US Census data), generating a usably small resultset, and dropping it into Excel at the end to make pretty charts that don't look (visually) out of place with the rest of the executive's presentation.

You'd end up with Excel being the GUI of an arbitrarily powerful network of computing resources.

I do like FRP though, it'd be nice to be able to automatically update cells in an iPython notebook when dependencies change their data. Also, a notebook is quite a linear structure, a grid (or tree, possibly just an acyclic graph) might be a better representation.

Good point. Excel then functions as the kind-of dashboard to lots of more powerful code.

Spreadsheet is more convenient way of visualizing a lot of numeric inputs and outputs at the same time.