You might think that the big ISPs deliver the same broadband products everywhere. But I’ve been seeing evidence that broadband speeds are definitely a local issue. One of the products that we’ve been using to help clients assess a new market is to get a lot of people in the potential market to take speed test. We’ve mostly been using the Ookla speed test, but probably any speed test is sufficient as long as everybody in a market takes the same test.

The results of these speed tests surprised me a bit because they showed a wide variance in the products of the major ISPs. For example, I’ve seen markets where Comcast is delivering a little more download speed than they are advertising. But I also saw tests results from a Comcast market where the speeds were about 20% less than advertised. I’ve seen the same thing with AT&T where there are markets that get only half of the advertised speeds and other markets where they were mostly delivering what they are promising. I’m not sure if there is any better demonstration that speeds are a local issue than by seeing that the big ISPs don’t deliver the same speeds in every market.

There is a long list of reasons that can account for the differences in speeds. A big one is the age and quality of the network cables. Older telco copper and older coaxial cables can cause a lot of problems with quality. The size of customer nodes is always an issue. If everything else is equal, a cable company node serving 100 customers is going to have better broadband speeds than one serving 200 customers.

The other big issue that affects customer performance is what I call network choke points. A chokepoint is any place in a broadband network that restricts the flow of data to and from customers. There can be a choke point directly within a neighborhood if the nodes are too large. There can be a chokepoint between a node and the core network if the electronics for the connection are undersized. There can be a chokepoint on local network rings if they don’t provide enough bandwidth. There can be electronics chokepoints at a headend if a router or other major piece of electronics is overwhelmed. And finally, there can be an overall chokepoint in a network if the data pipe going to the Internet is too small.

Chokepoints don’t have to always be a problem. Many chokepoints only appear during the busiest hours of usage on the network, but don’t impede data speeds when data traffic volumes are smaller. And this means that chokepoints are often hyper-local. They might affect one neighborhood but not the one next door, and only at some times of the day. I’m guessing that the slowest results I saw in the big ISP speed tests were during the peak evening hours.

These chokepoints obviously don’t only affect the large ISPs and plenty of smaller ISP networks have chokepoints. I’ve seen numerous network chokepoints appear in recent years due to the explosive growth of the use of broadband. A network that may have been functioning perfectly a few years ago will develop chokepoints as the amount of total bandwidth on networks overwhelm some portion of a network.

ISPs often are challenged to keep up with the upgrades needed to avoid chokepoints, because generally the only ways to relieve chokepoints is to replace cables or to upgrade electronics, which can be expensive. Smaller ISPs often don’t have the immediate capital available to fix chokepoints as they appear. The big ISPs tend to ignore chokepoints as they appear and to make large fork-lift upgrades periodically instead of making the constant small upgrades needed to keep the network working perfectly.

I always advice my clients to keep a running list of all of their chokepoints. With good network engineering and monitoring practices a company can see chokepoints coming long before they materialize and hopefully can plan to make the needed upgrades before they degrade the customer experience.