Cogent (AS174) sells IP transit and they sell it cheap. Everyone knows this. It is how they position themselves. It’s their competitive advantage. It’s why they think they will take over the world. They plan to undercut the prices and margins of all of the bigger carriers, combine that with strategic (cut-rate) acquisitions, and wait for everyone else to go broke. So far they’re doing a pretty good job of executing on that plan.
However, I recently learned that “Cheap” isn’t actually Cogent’s secret market advantage at all. I’m not denying price matters to people who buy from Cogent. Heck, that’s why Renesys decided to buy from Cogent for a development installation for Babbledog. (There will be more about Babbledog here shortly, as I’m sure it’s something that many of Renesys’s network-centric customers will have some questions about.)
This is a tale of a network-clueful small company trying to get connectivity at a well-connected building in Boston at a reasonable price. This is a tale of sorrow and woe, misery and despair. I’d like to say it has a happy ending, but on review, I believe that many of you will conclude, at least for the IP transit industry, that the ending is not happy at all.
Babbledog is colocated at 1 Summer St in Boston. We chose it because it is relatively close to where many of us live and work, and has a long list of providers colocated in the building from whom we intended to buy. That’s probably where our trouble started.
I should say parenthetically that the Markley Group, the landlords of the 1 Summer facility, were easy to deal with on the whole, the space and power are good and affordable and the location is fine with us. Our problem here was never the facility; it was the providers.
Initially, we ruled out the very smallest of providers, since many of them (however good) were simply single-homed to other providers. Then we ruled out providers who were no longer in the building, either because of mergers and acquisitions (Wiltel bought by Level 3 and still being integrated, Verio/NTTs business in the building bought by Cogent, Broadwing/Focal bought by Level 3 and still being integrated, SBC no longer selling as SBC but as ATT; the list goes on and on). Once we sorted out who was left from a more up-to-date list, we began getting quotes. We asked for quote on a 10 mb/s commit on an ethernet handoff to the Markley Meet Me Room (MMR) on the 4th floor. I know that this is a low commit, but it was a development project and we thought we would start small. The majority response to this request for pricing was silence. Most salespeople from most of the organizations we contacted simply ignored us.
When we pushed a little bit harder, using some of our industry contacts to shake the lazy salespeople loose from the collective stupor into which they seemed to have sunk, we began receiving amusing responses. The names are changed to protect the guilty, but here are some anecdotes
- Provider A1: “We’re not in that building.” Renesys: “Markley Group says you are. I just saw your cabinet/fibre.” Provider A: It’s not on my list. We’re not in that building.
- Providers L and G: “We’re in that building, but we can’t sell to you. Maybe later. Try back in the fall.” Renesys: “Ummmm. What? This is a 10 megabit circuit. This shouldn’t cause capacity or provisioning issues for anyone under any scenario.” Providers: “Yeah, we’re just reorganizing some stuff right now and don’t want to take any new customers.”
- Providers R, X, C: “I’ll get back to you tomorrow with a quote.” Renesys: “Hey, it’s been three days. Can we get a quote, pls.” Providers: “Oh, dude, sorry. later this afternoon.” Renesys: “Ummmmm. Now it’s been five more days. Do you have any prices that you can share with a potential customer?” Providers: “Real Soon Now.(tm)”
- Provider Q and X2: “Here’s a quote.” Renesys: “Dude, that looks like a 1998 price.” Providers: “Yep. pretty much.”
The most amusing (and frustrating, at the time) response was from Provider A2. They actually got us a price (only took a week) and it was reasonable, so we accepted the terms and signed a contract. Three days later, the salesperson came back to us and said that he had mispriced the cross connect because he didn’t realize that we were in the Markley suite on the 4th floor. This is understandable because when we asked for the quote and ordered the circuit we only specified that four or five times. We obviously should have been clearer. And more repetitive. And clearer. And said it more often. And clearer. You know.
The upshot of this miscommunication was that Provider A2 decided they really needed to charge us several thousand dollars more install fees in order to cover the cross connect from their facility on another floor that they hadn’t noticed in the original order. We declined politely and pointed out to them that everyone had told us not to try to buy from them because their cross connect fees were so high (and they were hard to deal with, according to others). In the end, they did the right thing: they relented on the extra money and installed the circuit. I hope they get extra business at the facility out of it. I’m personally recommending them to other tenants, I know.
Post-turn-up provider A2 has been fabulous. Great NOC. Good support during BGP turn-up (including helping us identify a rookie mistake in our BGP setup). All good. But the sales, provision, install portion was horrible.
Contrast all of these experiences with Cogent. We called Cogent for a price, they gave us two or three options on the same day. We picked one of those, they sent a contract and a form. We signed the contract and filled out the form, they installed the circuit and even gave us the fiber-to-copper transciever so that we could take a copper hand-off (lame, I know, but this is a 10mb/s circuit). There were a few bobbles in setting up BGP because of Cogent’s somewhat notoriously complicated two-session BGP configuration, but ultimately we figured that out as well.
The ugly fact here: Cogent’s price is only about 15% cheaper than Provider A2’s price above. In other words, even though Cogent is still the price leader, A2 is completely competitive. Price would never be a differentiator between the two at these levels. And yet, the purchasing experience with Cogent was nearly frictionless. The purchasing experience with Provider A2 was akin to removing impacted wisdom teeth. The purchasing experience from every other provider contacted was more akin to curing cancer: it wasn’t going to be possible for us to do under any reasonably likely set of circumstances.
The moral of the story: Cogent’s secret weapon isn’t their price. It is their deviously clever business plan of knowing what they sell and where they sell it, answering questions about those services, selling them, provisioning them and billing for them. It seems like a crazy plan, but it might just work. As long as Cogent keeps executing this wily and sneaky plan, they seem likely to continue to experience market acceptance. I don’t think that they are in any danger of any of their competitors discovering this cunning plan any time soon. Most of them couldn’t find their own networks with a map, a flashlight and a customer to guide them, based on our recent experience.
Financial analysts rarely look at companies’ performance at this level of detail. So they continue to report on larger trends, like numbers of lit buildings and price per megabit per second. It’s very difficult for them to find the time to look more closely at actual execution issues. I suggest analysts covering the communications industry pick 5-10 multi-tenant datacenter facilities and just try to buy something from one of the other tenants. My guess is that they won’t have any more luck than we did. Unless they talk to Cogent.