Thanks Cyberian.
Measuring usage in a TLD is typically done with a web survey and analysis because it involves checking hundreds of thousands of websites and classifying their usage. It is a complex task as it is essentially building a search engine for each TLD and this is why registries have difficulties in doing it properly. The other aspect is that registries often only use a small 5000 site sample. But as Cyberian pointed out above, "developed" can mean different things to registries and some will consider Godaddy's PPC landing page for undeveloped domains to be a developed website or a redirect to a developed website. With some TLDs like .co or mobi, that single page can represent over 40% of the websites in the the TLD.
I've been running web usage surveys on the Irish domain/hosting market for years and generally survey about 350K sites each month. The July 2012 figures for the Irish market are here:
http://whoisireland.com/iecctldsurvey.html
http://whoisireland.com/irishsitessurvey.html
I have been working on shifting the site to a new server and haven't added the latest stats.
About a year ago, I ran a survey on .eu ccTLD that covered just over 2.3 million websites and posted the results here:
http://www.hosterstats.com/blog/2011/11/23/doteu-web-survey-results/
And this is a .co survey from July 2011:
http://www.hosterstats.com/blog/2011/07/27/website-usage-in-co-july-2011/
There are some trends that are present in most ccTLDs. If there's a strong local market and economy, then the level of developed websites will be strong (25%-50%). A repurposed ccTLD like .co may be highly speculative and have much more PPC/parked domains than developed websites. Development in such a TLD might be between 9% and 23% because the numbers of speculative registrations skew the overall figures and many of these speculative registrations are either PPC parked or holding pages.
Development in .com TLD is generally high on a country level market basis because of the global nature of .com and the fact that it predates most ccTLDs. As you move away from .com towards net/org/biz/info/mobi/asia, development rates fall.
In any well developed country level market, the ccTLD/com axis for domains will occupy 80% or more of that country's domain footprint and these two TLDs will have the highest level of development.
Traditionally, about 30% or so of domains had not been properly set up in DNS or had no website associated with the domain. This changed over the last eight years or so because of the way that many larger registrars and hosters now automatically park new and undeveloped domains on their PPC landing page for undeveloped domains. (Godaddy being the classic example of this).
Some registries publish their own limited surveys (5000 sites or less) but these are not really reliable given the way that web development varies. There's a timeline from a domain being registered and a fully developed and active site appearing on the domain. However there's also a level of website abandonment in TLDs where someone gets a great caffeine fuelled idea at zero dark thirty, registers the domain and has a Wordpress installation by zero dark fifty and then falls asleep leaving just a 'hello world' post and never bothers to update the site until the domain drops.
Nominet (the .uk registry) did a small survey (5K I think) of .uk websites in November 2011. It published its results here:
http://db.nominet.org.uk/page/domain-name-usage/
I'm not sure if SIDN (the .nl registry) did any surveys. Eurid did a survey in October 2011 but there are serious issues with the interpretation of results as they seem to show .eu with more development than the larger 2.3 million site survey shows. The .eu has really become just a gateway TLD where users are redirected to the relevant ccTLD or .com website.
http://www.eurid.eu/files/Insights_Cat3.pdf
http://www.eurid.eu/flipbooks/categorisation/index.html
Because of the wide diversity of usage and development, measuring and classifying website usage in a TLD is always going to be difficult and it is often in the best interests of the registries to choose certain categories and ignore others in order to present a TLD as being more developed or used than it is in reality. In terms of accuracy, COInternet's usage figures are not considered reliable.
Regards...jmcc