I recently ran across a solicitation from an ISP looking to hire a consultant to help them quantify the price sensitivity of ISP products. For those of you not familiar with economics, understanding price sensitivity would help to accurately predict the number of customers that would abandon a product if prices are increased. Understanding price sensitivity would also allow an ISP to create a demand curve which could predict the number of customers a business should attract at a given price.

It turns out that it’s incredibly difficult and mostly impossible to define the demand curve for most products in the world, including broadband. This is not to say that price sensitivity isn’t something that everybody in the industry would love to know, and I’ve been asked this question for my whole career. Early in my career I worked on rate cases, where a telephone company sought permission from a regulatory body to raise rates. Telcos didn’t raise rates very often back in the 70s and there was always a lot of gnashing of teeth by management who worried that a few-dollar rate increase would lose customers. It turns out at a time when telephone service was a monopoly that almost nobody dropped service, but telco management didn’t trust this to be true.

Perhaps the easiest way to understand a demand curve is to look at one of the few examples where it might be measurable. Consider Uber in a big market like New York City. The company raises rates any time there are more requests for drivers than there are drivers. This happens at busy times of the day such as during the morning commute and during for events like a snowstorm. People in a major city have alternatives to Uber. If Uber is too expensive people can hail a cab, walk, or try another ride service like Lyft. By now, Uber has gathered enough data in a market like New York City that they probably can closely approximate the demand curve for their service in the city. However, even here, the demand curve will be specific to New York and not Uber anywhere else.

However, broadband (and most consumer products) don’t work that way. There are a few other economic terms that are relevant to the topic. One is substitution. Customers are a lot likelier to be price-sensitive if they have a product alternative they view as equal. There are portions of the cities with Google Fiber where the telco also built fiber and where the cable company upgraded to gigabit speeds. In these rare little pockets a customer might find the broadband alternatives to be pure substitutes for each other. In most of the country, broadband products are not considered as substitutes. For years we’ve seen millions of customers annually ditch DSL for cable company broadband and a lot of those households are likely to never consider moving back to DSL.

One of the key characteristics of an economic substitute is the ease of customers to change between similar products. It’s easy to pick a choice other than Uber. It’s not easy to change ISPs and households agonize over making a change. Changing to a new ISP might mean sitting at home for an afternoon waiting for an installation technician. It also means suffering through the call with the existing ISP as they try to keep you as a customer. It likely means driving electronics back to the canceled ISP.

One of the components needed to estimate a demand curve is getting a lot of data points. This is what is so awesome about Uber for an economist – they can get piles of data to understand how customers react to various pricing offers. Broadband doesn’t generate a lot of data points – ISPs don’t change prices very often. Even when they do change prices, ISPs don’t get the same kind of instant feedback that Uber gets over a price change. If an ISP raises prices, they are not going to be able to correlate this with a customer who gets around to dropping them six months later because of the increase.

To further complicate things, my firm does surveys and we’ve found that the majority of homes have no idea about what they pay for broadband. This is not surprising since in most markets over 70% of households have bundles that disguise what they are paying for each product. Customers also tend to remember the advertised prices, which they may or may not have ever received, instead of the actual prices they are paying. Customers are also regularly duped by deceptive billing practices where ISPs hide parts of the cost of a product in hidden ‘fees’.

I can’t blame a new ISP for asking the price sensitivity question, and it’s not an unreasonable question to ask. But I would venture to say that even Comcast and AT&T have little grasp over the demand curve for broadband. If they start raising rates every year as Wall Street is expecting, then over time the giant ISPs might start to get an tiny inkling of the demand curve – but even then, broadband is such a complicated product that predicting price sensitivity will always be a guess.