A simple argument for investing in your pricing

I having full-time dedicated pricing experts on payroll makes sense as a rare thing.

(1) When pricing is hidden, this means in practice you have every sales rep trying to maximize pricing for the individual person. Meaning there are in fact millions of "pricing people" in companies around the world.

(2) When pricing is NOT hidden, pricing is rarely updated, which is a good thing. Ads, comparisons, press releases, etc all may reference pricing. Changing pricing on your existing customers really can cause churn (which for SAAS outweighs the typical margin gain).

(3) The author doesn't mention discounting, which is in fact playing with pricing, and something companies do a lot of.

(4) For transparent pricing, finding the right price doesn't really change that much. So if you only have a few products, investing in a full time person seems like maybe they would be really busy for the first 1-12 months then...do nothing? You really probably need new products coming out regularly to justify someone dedicated to the role.

I don't know of published applications of demand estimation in software in particular. I know that Microsoft and Amazon do this to estimate demand for cloud services, but I don't think any of that is publicly available.

In principle, _any_ time that Amazon wants to make a decision about offering an in-house version of some product that is already sold in their stores this is the kind of thing they will do first (again, none of these estimates are public to my knowledge).

There are prominent applications in:

- automobiles (Berry, Levinsohn and Pakes (1995) - the originators of the technique),

- alcohol (Miravete, Seim and Thurk, (2018) - relevant especially to the case of a high-dimensional product space),

- the minivan (Petrin (2003) - relevant to studying a new product),

- breakfast cereal (Nevo (2003) - this is a surprisingly innovative and competitive market category!)

- radio stations (Sweeting (2013) - probably the current state of the art econometrically)

- studying vertically organized markets with unobserved prices (Villas-Boas (2007))

- (There are other applications beyond those listed here - demand estimation is a foundational issue for answering many, many, many economic questions.)

Depending on the specific features of the software demand estimation problem you are thinking about, you may find any of those references helpful.

Two very recent surveys have been published by four of the top people in this area:

1. Gandhi and Nevo: https://www.nber.org/papers/w29257

2. Berry and Haile: http://www.econ.yale.edu/~pah29/Foundations.pdf

Plus there is a now-standard Python implementation of the estimator:

https://pyblp.readthedocs.io/en/stable/