Contents:

Summary

Overview

Research details

Table hosting details

Grand Table of Hosts

Final conclusions

Further details

External links

Key Google searches:
Negative N1 = "hate host x" N2 = "host x sucks" N3 = "host x sux" N4 = "avoid host x" N5 = "problems with host x"	Positive P1 = "love host x" P2 = "host x rocks" P3 = "host x is great" P4 = "host x is * great" P5 = "recommend host x"

Summary
Since comparing all of the hosts many years ago, a lot has changed, so this update was well needed. In a nutshell, I have found that probably the 'best' web host for most people would be ICDsoft (shared hosting). And for virtual or dedicated hosting, I would recommend Liquidweb. I have spent months of grueling research to find that out, so for those who want to see how I came to these conclusions, read on...

Overview
This article should inform those looking to buy a quality hosting provider for their forthcoming site. How would one decide between the plethora of hosts though? If you wanted to avoid a degree of risk, you might have researched the various web hosting service review sites on the internet (and maybe come unstuck). More likely though, you would probably look on message boards, forums, and ask friends for their opinion. This site takes that concept to its logical conclusion. We've scavenged the whole web using Google, and searched for various positive and negative phrases such as "I love host x" or "host x sucks". This way, we can find a rough consensus towards a particular host if there's enough data available.

We first reviewed the hosts on this page back in 2005 (you can see the archive here). At the time, Dreamhost was ranked highest, and for the curious, it brought a commission of one affiliate sale per month on average. Then in late 2006, we reviewed most of the good ones again, and this time, the relatively rare host www.asmallorange.com came top (unlucky for me actually, as they had no affiliate scheme). This latest update has seen asmallorange.com knocked off its perch. For the full history of this page, see the boxout to the below/right.

History of this page: Initial page created: 14/10/2005 Update 1: 22/12/2005 (63 hosts). Added term: 'problems with hostx', and updated top 10's search terms + specs. Update 2: 08/08/2006 (62 hosts). Updated top 20 results, and included old statistics for comparison. Removed Directnic.com entry, thanks to the keyword confusion over whether people were praising Directnic due to their hosting, or their registrar services. Also altered ranking formula so that there's less penalization against less known hosts (always weigh the risk yourself though). Turned "i hate hostx" and "i love hostx" into just "hate hostx" and "love hostx" to obtain more Google results. Update 3: 02/01/2009 (57 hosts). Updated top 20 results, (still included old statistics for comparison). Removed high cost per gigabyte / mega-expensive hosts (Datapipe, Rackspace, 34sp, Maxasp), and intermedia.net (name confusion over other brands). Resulting top two hosts have affiliate schemes, so for full disclosure I have joined those. Update 4: 27/01/2009 (60 hosts). A mini update this time. Added 3 new hosts - hostpapa.com, fluidhosting.com, and the unusually good icdsoft.com which (yay for me!) has an affiliate scheme also. Update 5: 11/02/2009 (60 hosts). Sifted through top 6 results to properly filter out affiliate scheme links, and also to take account of duplicate names (e.g. "liquidweb" = "liquid web", or "asmallorange" = "a small orange"). Thus most of those scores have dropped slightly (about 2-3% on average), but the placements are exactly the same as Update 4, except asmallorange has moved from 5th to 3rd, and hostmysite has dropped from 6th to 7th. Update 6: 17/03/2019 (60 hosts). Updated links to ICDsoft and tidied up formatting.

Back to how it all works. As mentioned earlier, using Google, we look to see how many websites report that they 'hate' or 'love' host x. Naturally, the question will arise; how can one be sure the results containing glowing praise about a host are genuine and accurate? Of course, the conclusions can never be perfect due to lack of data, word/service ambiguity, bias, and plain advertising. But it turns out that the majority of comments are written on personal sites, forums and message boards. Under such circumstances, posts are more often genuine than not, and for the forums at least, moderators are likely to detect if a recommendation is spam or not.

Research details
As before, most of the placings have remained fairly consistent. You can always verify the results for yourself - in the main table further down the page, you'll see the exact number of searches made in Google for a particular search term. Naturally, the relatively small sample sizes for each search will cause the real result to veer off slightly, but we have factored risk into the overall score as well.

It wasn't easy. I had to filter out Google duplicates using custom made software, and look for spam and advertising among the results. For the top 5-10 hosts, each link was checked by hand to see if it was a genuine post from a real customer without any affiliate motive (I'm sure many would recommend their host regardless, but we choose to be on the safe side, especially as it would be unfair to hosts without an affiliate scheme in place). For full details on my approach towards filtering and phrase selection, see towards the end of the page.

Unfortunately, it wasn't possible to include every hosting site out there, simply due to the fact that there are not enough users to comment on them. Rarer hosts such as Hostican.com, Successfulhosting.com and Parcom.net seem as though they could be pretty good, but they simply aren't popular enough to be included.

If you're getting impatient, at this point you may just want to jump to the main table further below. Otherwise keep reading on.

Table hosting details
The gigantic table below sorts web hosts by how good or bad they supposedly are. (Try this older link if you want to roughly see the hosts sorted by popularity). You'll find that some high ranking hosts may not have the specs you require (like if you want more than 20 MB of disk space, or higher than 1 GB of bandwidth per month). Equally, many of the cheap or free hosts are lowly ranked due to the obvious limitations (forced adverts, lack of space/bandwidth and limited filetypes and features), but since they're free, they're worth considering (especially if you're a beginner to site creation and want to test the waters). If you're going that route, Bravenet, Doteasy, Micfo and Directnic (relatively unheard of names) seem to be the best here. Angelfire is the worst, with Tripod following closely behind :-) Four hosts we left out the table include Datapipe, Rackspace, 34sp, and Maxasp. These seem decent hosts, but the cost/gigabyte ratio for bandwidth and disk space is very high (however, you may get extensive support and other features).

The score At its most basic level, the score is the number of positive comments divided by the number of negative comments. But greater weighting is given to hosts with many members, thus increasing the 'reliability' of the given score. In other words, hosts with a normally good mark, but who don't have a large internet presence will be marked down (due to the margin of error factor risk, and the fact that they won't be as established, making them more of a risk again). And for the same reason, an apparently poor, but minor host's score will increase towards the 'center' of the list if their score was normally low. For the curious, the following formulae were used: reliability = (popularityOfHost/30) / (1 + popularityOfHost/30) ratio = (p1+p2+p3+p4+p5/2+1) / (n1+n2+n3+n4+n5/2+1) weightedRatio = ratio ^ reliability SCORE = weightedRatio / (1 + weightedRatio) * 100

With that out the way, we can continue with a description of the table elements. First off, you'll see numbers in 'faded' in most columns, as well as normal/black. The faded numbers indicate the results from two years back (near 2007) so you can compare. The numbers under the Negative/Positive web comments are fairly self explanatory: The bold black numbers reflect the filtered amount of hits for that term. The number in the brackets reflects the unfiltered amount (which may be lower than expected for some hosts, since I ran the results twice for certain hosts like "liquidweb", which could also be called "liquid web" - with the space).

Key Google searches: The phrases and colours below correspond with the numbers in the main table. Basically, the numbers represent the number of times a phrase appeared in Google. The number in brackets is the total number of results before filtering.
Negative N1 = "hate hostx" N2 = "hostx sucks" N3 = "hostx sux" N4 = "avoid hostx" N5 = "problems with hostx"	Positive P1 = "love hostx" P2 = "hostx rocks" P3 = "hostx is great" P4 = "hostx is * great" P5 = "recommend hostx"

Next onto the basic specs. I only have room to give a very rough indication of the type of web hosting plan you'll get. If you wish, you can investigate the site in question to find out if extras include PHP, CGI/perl, MySQL, ASP, ASP.NET, number of email accounts, email POP / SMTP Access, Windows and Linux/Unix hosting, FTP access, number of domains hosted, access to raw logs, free domain registration/transfer, and other special discounts. I'll leave that up to you as there's no way I'm including all that data in the table :) If you see '^^ GB' or '^^ MB', that means there's supposedly no limit to the disk space and bandwidth (unmetered), but in reality, there will always be. Always read the terms & conditions and smallprint from each website for the full story (look at the harsh TOS limitations on Hostultra's free web hosting for example!). Finally, most web site hosting services are based on low cost 'shared hosting', but if you see MBhd or MBvps instead of MB in the table below, that means it's a dedicated server or virtual private server (and is therefore going to be much more expensive!).