The reality of cheap web hosting

Since I started building sites using WordPress, I’ve tended to use cheap hosting – very cheap hosting. I’ve run high-profile government websites quite comfortably on shared hosting deals costing £50 a year, or less. Some had daily page views running into the thousands; at least one was for 10 Downing Street. It seemed in keeping with the low-cost ethic, and it didn’t let me down.
But over the last few months, I’ve come to understand a bit more about how cheap hosting actually works. The reality, I’ve realised, is that all web hosting is effectively free of charge. When you pay a fee for hosting, you’re really paying for support – or perhaps more accurately, the promise of support when you need it. An insurance policy, in other words.
Looking back, I can recite instances where a cheap hosting company has suspended accounts unilaterally and without warning, because traffic or other activity hit a notional limit. Or where a global setting was changed on a shared webserver, breaking key functionality on one of my sites. For the vast majority of their clients, these wouldn’t have been problematic: most websites won’t trouble their traffic limits, or use difficult functionality. But mine did.
Cheap hosting means zero tolerance. You aren’t paying them enough to employ someone to get in touch proactively before things go wrong; or to respond to your anguish afterwards. They will employ unilateral limits, and make unilateral changes, based solely on a cold analysis of what will suit the majority of clients. Based on automated tests and calculations, not human beings. I’m not blaming them; you can’t really expect them to do anything else.
But that isn’t good enough for serious publishing efforts. They do deserve better – advance warnings, responsive support in a crisis, proactive maintenance to stop bad things ever happening. And that comes at a price.
In the context of my crowdsourced business plan, one emerging idea is long-term site support. In a WordPress context, that means updating the underlying technology; updating WordPress itself; updating themes and plugins; and at each stage, testing to make sure everything still works as intended. So I’m talking to some people about the possibility of providing a WordPress-optimised, centrally managed hosting service, aimed at government and corporate usage. We feel WordPress has reached a certain level of maturity, and it’s probably time the hosting arrangements did so too.
If we do it, it’ll be the best, slickest, smoothest, friendliest, smartest, most tailored solution we can imagine. But it won’t be cheap.

12 thoughts on “The reality of cheap web hosting”

  1. You probably know this, but just in case it’s helpful…
    Domain mapping on a WPMU installation would allow you to run thousands of WordPress sites on one installation. It’s how wordpress.com works. They provide a domain mapping plugin, too: http://wordpress.org/extend/plugins/wordpress-mu-domain-mapping/
    If you’re going to become a WP hosting provider, you’ll find this a much more efficient and cost-effective set up than managing multiple single install sites.
    WordPress and WPMU will merge in the 3.0 release, but the functionality of WPMU will be retained (it has to be, right? It’s what Automattic make their living out of).

  2. Why will it not be cheap?
    Use of a dedicated server at rackspace can run you maybe £300 a month, that’s only 6 “£50 hosting” a year websites a month you need (so 72 sites total) to break even. Surely you can get that many corporate and governemt WP sites hosted over the course of a year? Heck, add £10 to the fee, £60 a year and you only need 5 a month.
    Quality doesn’t always have to be costly if you know how to market and support it correctly.

  3. Only just seen this, but I can warn you that you have to be really really careful in what you offer and at what cost.
    Providing terrific, high uptime and stable support that can tolerate DDoS, hack attempts and big traffic spikes isn’t easy. Many cheap hosts run big clusters which can serve a lot of pages quickly. They can also RAID everything up and have redundancy.
    Eight cores of server will fall over when you make the front page of the Huffington Post. It’s like Digg, only worse!

Comments are closed.