The Cherry Picking Dilemma

iProvoI ran across an article written by somebody in Provo, Utah who claims that the penetration rate for paying data customers on the Google network has fallen to around 20% from a previous penetration rate of 30% when the network was operated by the city. I have no idea if the 20% penetration rate cited is accurate, but it is not surprising since Google also offers 5 Mbps for free in the City as part of the deal for buying the network from the city. I’m sure that a lot of households and students are taking the free option.

But the article did prompt me to think about cherry picking – the phenomenon where telecom carriers tend to mostly pursue customers who spend the most money. This topic is of particular interest when talking about Provo because the network that Google now operates was once an open access network. And I think the pre- and post-Google situations are worth comparing.

Back when the city operated the network they operated it on an open access basis, as required by Utah law. This means that the city was prohibited from being an ISP, but they could sell access to other last-mile service providers on the fiber network. Provo sold lit fiber loops on the network for roughly $30 per month. ISP using the network were then free to sell any services that a customer wanted to buy.

An open access network leads to a form of cherry picking in that no ISP is going to buy a $30 fiber loop and then offer a standalone inexpensive data product. There just is not enough profit in such a situation to sell a standalone $40 or $45 data products. Instead an ISP in an open access network will either price standalone data high, or else bundle it with lots of other stuff. You can contrast this to Qwest who would have competed against iProvo by selling low price DSL. I am sure that Qwest had some data products in the $30 per month range. They would have been much slower than the iProvo fiber but would have been attractive to the budget-minded customer.

And then consider Google who is definitely a cherry picker. They sell a gigabit of data for $70 per month. There are very few markets where a significant percentage of households are going to find that affordable, regardless of how attractive they might find the speed. I don’t know what Google’s target penetration rate is, but they can’t be shooting for the same overall penetration rate as the cable company can shoot for. The cable companies have a full range of products from slow to fast, from cheap to expensive.

I work with hundreds of ISPs and the one thing that I have consistently seen in every market across the country is that when customers have a choice between a low and a high priced data product that the vast majority of them will take the lowest priced data product that will give them a speed they can live with. Cable companies don’t expect more than a few percentage of households to buy their fastest and most expensive data product.

And so, even if the author of the article is right, I’m not sure that this is a negative thing for Google. If the city was selling broadband to 30% of households in an open access environment, then one has to imagine this represented a broad range of products at different prices and speeds. There would have been no really cheap products due to the $30 monthly loop rate, but there still was probably a range of packages between $50 and $200 with various combinations of data, video, and voice.

If Google has been able to get 20% of the people in Provo to pony up $70 per month for broadband they might be very happy with the results. They bought the network for $1, but obviously had to make some capital investments to get the network capable of gigabit everywhere. I see nothing automatically distressing about a 20% penetration rate of a very high margin product.

There are a lot of other new ISPs hitting various markets around the country today. A few of them are also peddling gigabit as their only product like Google. But most competitive ISPs still sell a mix of products. Whenever I talk to these companies I always caution them that given a choice that very few people are going to buy the gigabit if there is an affordable 100 Mbps alternative. I’ve seen a number of business plans that predict a high penetration rate of the fastest data product, but I’ve just seen human nature rear its head in almost every market I’ve ever worked in. Given a choice people will save money when they can. And all of the marketing in the world won’t get the to spend more than they are comfortable with.

3 thoughts on “The Cherry Picking Dilemma

  1. “I work with hundreds of ISPs and the one thing that I have consistently seen in every market across the country is that when customers have a choice between a low and a high priced data product that the vast majority of them will take the lowest priced data product that will give them a speed they can live with.”

    That illustrates that people are purchasing telecommunications service as a utility as they did for decades in the pre-Internet period with voice phone service. Since it’s a recurring monthly cost, they are rational economic actors in seeking to keep that dollar amount as low as possible. It also illustrates the bankruptcy of the business model that’s based on packaging and selling bandwidth tiers. Consumers ultimately desire reliable, good value telecom service, not bandwidth.

  2. I’m somewhat sure that I’m the author to whom you refer. Note that my numbers on penetration have since been backed up by research reported on in Multichannel, so I stand by the 20% as being accurate.

    I still maintain that Google is seeing an overall sharp decline in subscribers and revenues. According to the numbers I can find, about 1/4 of their fiber subs are buying a TV package (they do not sell TV unbundled or do phone service). Compare this to iProvo numbers showing that over 40% of data subscribers also took a video package. Not only did Google suffer a decrease in paying fiber customers, they saw a sharp decrease in both the number of video subs at the rate at which people subscribed to video. They also gave up any and all revenue from phone services that they do not sell (which about 2/3 of subs took). Privately, Google Fiber employees will admit that they are well off the mark from where they expected subscriber numbers to be.

    So here’s the question: why is Google losing a rather substantial amount of business despite their significant brand power and marketing? It’s the question nobody seems to want to ask.

    The point is well-taken that a 100Mbps product can cannibalize a 1Gbps product. The cost differences between delivering the two are negligible. If offering a 100Mbps product means you both increase your data customer share by 50% and add on a substantial number of video subscribers, it seems like it would more than offset the losses of offering a cheaper product. Google is far too committed to the gigabit brand to see the money they are leaving on the table.

    • Your questions are exactly why I call this a dilemma. Provo is something a little different for Google than their usual market. In other places they only build to neighborhoods where they get enough customers to agree to buy service. By just building to such neighborhoods they greatly hold down costs by not building anywhere, and each neighborhood meets their financial goal. But in Provo they inherited a network that covered the whole City. I suspect that if you broke the City down into neighborhoods there are a few that would meet Google’s expectations, but you will see many others that do not. And this is due to the high-priced product.

      This is something that cities have to recognize when they deal with Google or the other overbuilders that are using the same model – the city is not going to get a ubiquitous solution to their broadband needs but instead are going to get a product that satisfies the top 20% of the market who can afford it. If that’s what a City really wants then obviously that is okay, but most cities I talk to want broadband for everybody and they don’t get that from the gigabit builders.

      The 20% number you cite is no surprise because my guess is that in most cities if you took that business plan to the whole city you’d see a similar penetration rate. There might be some cities with a higher percentage of affluent households where you might do better, but probably not many. Your Provo statistics clearly show he impact of cherry picking while it would be disguised in other places where something less than all of the city gets fiber.

      Provo is also a little different than other markets in that people were able to elect free, but very slow broadband as part of the deal for Google to take over the network. If I recall it’s 5 Mbps which in today’s world is a miserable speed, but free is damned attractive and I picture tons of homes in the city suffering with that until some day when they will finally buy something else. You can just picture kids yelling that the Netflix doesn’t work but Dad yelling back that it’s free!

      You also ask why they don’t sell other products like voice. I’ve asked the same question of a number of ISPs, but it’s the new trend. I think it’s part of a desire to keep the product line simple and also a belief that voice is an obsolete product. But still something a little less than 50% of homes have a voice line and the product has a very high margin.

      I must say, though that if you were the author of the Provo article that I’m impressed you found my blog. The universe of folks who care about broadband is connected pretty well, I guess.

Leave a Reply