A D N Edwards
Honorary Fellow
Department of Computer Science
University of York
York
YO10 5GH
Preamble
Being retired does not stop one's mind working.
I think about stuff.
It does mean that I do not have access to any resources to undertake research, not even students to do projects.
But it also means that I can think – and write – about stuff without any concerns as to whether it's useful or publishable.
This paper is one outcome.
It is 'inspired' mainly by advertising on the sides of vehicles.
I have seen URLs, including long ones, on the sides of vans and trucks, and I have wondered what prompted the company to choose such long – and apparently difficult – names.
I thought I would write down my thoughts, and that has now turned into quite a long article.
It's way too long and trivial to be worthy of publication, but who cares?
Introduction
A URL is a Universal Resource Locator, or what most people probably think of as a Web address.
This paper is about the use of URLs in advertising.
Figure 1 shows the structure of a URL.
In this article, I am just interested in the domain element.
This is the main element that the website owner chooses.
(They also have a choice of the top-level domain or TLD.
That is they might choose .com or .co.uk or whatever, but that is a topic in itself, probably worthy of a separate article, and will not be discussed in detail herein.)
The main focus of this paper is subdomains which are long, for reasons which should become apparent.
Figure 1.
Structure of a URL, Structure of a URL, consisting of the protocol, subdomain, domain, TLD (top-level domain) and path.
Note that not all online authors agree on the nomenclature of these components;
I have chosen convenient names.
A website is a very valuable form of advertising.
If the owner can attract a potential customer to their site, then they have access to as much information as the owner wants to share,
and they can select precisely what information to present.
Contrast this with conventional advertising, on billboards, banners or magazine pages – or on the side of vehicles
(which is where I seem to have collected most of my examples).
The point behind this paper is, that you can use those conventional forms of advertising to direct customers to your site and hence to all that information.
It should be acknowledged at this point that by far the most frequent route to a website is by the user clicking on a link on another site.
That is to say that an advertising link might appear within search engine result, or on a Facebook page or wherever and that is the most likely way that the target site will be accessed.
Nevertheless, there is an apparent role of URLs in conventional advertising, and that is what we are going to discuss.
The point is that URLs are used in advertising, and that being the case, why would any website owner choose a long domain?
For instance, why would anyone choose the URL www.bluewaterbathroomsandkitchens.co.uk (a domain of 29 letters)?
'Resources' on the web are essentially files.
A file may contain text1, or an image or other material.
There are trillions of files – or resources – on the web, and each one has to have a unique address, and that is what the URL provides.
In inventing the Web, Tim Berners-Lee was incredibly far-sighted.
This is attested by the fact that so much of the web is still based on the essential technology he invented.
Yes, of course, there have been all sorts of extensions and new media included, but the fact that the very first webpage can still be viewed online2 demonstrates its future-proof technology.
The web and its underlying technology has survived its almost-unimaginable expansion to the vast and essential facility that it has become.
One element of that design is the URL.
The domain element is simple – and minimally structured.
A postal address, by contrast, contains a degree of redundancy.
There are famous examples of letters with minimal, syntactically incorrect addresses being nevertheless delivered.
This is not possible with a URL;
if the URL is wrong the page will not be delivered.
(This is setting aside, for the moment, the existence of search engines which may compensate for poorly spelled URLs, which will be discussed below.)
Accessing URLs
As a link (anchor)
There are a number of ways of accessing any URL.
Most common is as a link on another page.
For instance, Figure 2 shows a link on a
University of York webpage.
Figure 2.
A web link as it appears on another page.
The fact that this text is a link, and the URL to which it links is contained (but hidden) in the HTML text of the page which contains it,
as in Figure 3.
Figure 3.
The HTML text for the link in Figure 2.
Note that HTML refers to such links as anchors, hence the use of the <a> tag.
It should be apparent that the spelling of the URL in this case is irrelevant to the user, because it is (essentially) invisible to them.
This fact may be exploited for nefarious purposes by unscrupulous website owners.
In this case, for instance, it might not lead to information about Computer Science at the University of York.
Thus, the link in Figure 4 would look just the same to the user.
Figure 4.
The HTML text for the link in Figure 2.
Note that HTML refers to such links as anchors, hence the use of the <a> tag.
The destination of a link is not entirely invisible, because browsers generally do display the destination of a link.
If the user hovers over the link, they may see somewhere in the browser window the URL of the link, as in
Figure 5.
Figure 5.
The cursor is hovering over the link Department of Computer Science and in the bottom left-hand corner of the browser
(Firefox) is the URL of that link.
As is evident from Figure 5 the identity of the URL is not very prominent, but it is worth looking for if the user has any suspicions regarding the destination of the link.
It is worth saying at this point that in some ways most of this article is irrelevant.
That is to say, that as I have just explained, for the most part the spelling of a URL – including the root domain –
is hidden, and nothing that the web user needs to be concerned with.
However, if you stick with me, at least as far as the section on
Long URLs, the relevance should become clear.
A special case of getting a link on a page is provided by search engines.
For instance, if I search for Computer Science York
(in this case in DuckDuck Go), I get the return in Figure 6.
Figure 6
Result of searching for Computer Science York, in DuckDuck Go.
This is headed by a link, Department of Computer Science - University of York which refers to
www.york.ac.uk/computer-science/.
The link in the heading on this return refers to www.york.ac.uk/computer-science/, but again that is invisible to the user
(unless they check the box at the bottom of the window).
QR codes
Another way to access a particular URL is through a QR Code.
Scanning the QR code in Figure 7 will take you to the Computer Science home page, again.
Figure 7.
A QR code linking to www.york.ac.uk/computer-science.
Note, once more, that the spelling of the URL is irrelevant to the user, who cannot see it in the QR code.
The user just has to point their phone at the code to be taken to the site.
QR codes can be most powerful in (conventional) advertising.
A QR code can be printed on a magazine advert, a billboard or a banner.
All a (potential) customer has to do is to point their phone at the code and they will be taken to the site.
Of course, once again, this convenience has its downside, precisely because the underlying URL is invisible.
While the context might suggest that this code will link to the University of York Computer Science home page, it may not.
It might link to the rogue site in Figure 4, or this has also been used as a vehicle to spread viruses;
inquisitive phone users might scan a random QR code to find out where it leads, and inadvertently pick up a virus.
Address bar
The final way of specifying a URL to visit is via the browser address bar.
A typical address bar (this one from Firefox) is shown in Figure 8.
Figure 8.
A Firefox address bar, this one containing the URL https://www.york.ac.uk.
Address bars have developed a great deal in the history of the web.
Originally the user would have to type in a whole and correct URL to be taken to that site.
Two developments have largely changed the use of the address bar.
One is the use of Ajax4, which means that interactively as the user is typing in the bar they will be shown likely completions of their input.
For instance, Figure 9 shows that after typing just the first 17 letters of the URL
https://www.york.ac.uk (i.e. https://www.york.) the user is presented with the entire, correct URL – along with a number of other probable targets.
Figure 9.
An address bar with a pull-down menu which includes the target URL, along with 9 other potential targets.
The user has typed the substring https://www.york.
It is important to be aware that it is not necessary to fill in part of the URL, as in Figure 9 to get the appropriate address in the bar;
practically any combination of relevant words or strings – combinations of words such as 'york' and 'university' – will yield the desired result.
(There is more about this below.)
So it is that if user types something resembling a URL into the address bar, then that site will be visited,
otherwise whatever is typed is treated as a search term.
As is normal with any search engine, the immediate results are presented (by Ajax) in a drop-down menu.
The important point about the address bar in the context of this paper is that it is generally the only example where the user does need to know the (precise) spelling of the URL.
Choosing a URL
Bearing the above discussion in mind, how should someone setting up a new website choose its URL, or specifically its root domain?
There are a number of requirements.
There are rules, of course, as to what a domain may consist of.
It can have letters and digits, and dashes (minus signs), and that is it.
Thus, for instance underscores (_) and ampersands (&) are not allowed.
Also they are case-insensitive, so that, for example,
www.York.ac.uk and
www.york.ac.uk, and even
www.yORk.ac.uk will all take you to the same site.
Unique
The URL must be unique.
This is not a hard criterion to meet, since this will be checked before you register it.
It may be that the requirement for uniqueness drives some of the choices that are discussed below.
For example, the longest domain I have found (so far) is
bluewaterbathroomsandkitchens, and we see that this could not have been reduced to
bluewater because that site has already been registered, by a shopping mall.
We are concentrating on the domain.
It is possible to use an already-registered domain – but with a different TLD.
For instance, if you wanted to use the domain greatdomain and you found that greatdomain.com was already registered, but greatdomain.co.uk was not, then you could register it.
This is probably not a good idea, though.
For one thing, people looking for the .com site might alight on yours, and vice-versa.
Indeed, unscrupulous operators might set out to cause confusion.
For instance, if you have bought greatdomain.co.uk someone might deliberately register greatdomain.com and put some very different material on that site.
That might be pornography – or Rick Astley3 – and anyone coming across that when they are looking for your site might get a very poor impression of your company.
Thus, if your site is going to be at all popular, you probably ought to register both the .com and .co.uk versions – and have them point to the same site.
Of course, there have to be limits.
Should you also register greatdomain.org, greatdomain.org.uk?
What about .net…?
As well as variations on the TLD, if your site is going to be really important, then you might feel the need to also register domains which are mis-spellings of your domain – again to avoid possible deliberate mis-direction.
Again, there is probably a limit to the number of close spellings you will want to register, and, at the same time, the search facility in the address bar is likely to direct users to the correct site, even if they have mis-spelled the URL.
There is more about mis-spellings below.
Memorable
This is a most important property:
in the context of advertising, ideally the website owner wants the (potential) customer to see and remember the URL – and to link it to their product or service.
The point is, why would you put your URL on advertising?
Presumably you hope that people will see the advert and be interested in your product.
They need to see that URL, remember it, and at some later time will type it into the address bar of their browser.
It is easiest if they already have a recognised brand;
if you are Ferrari, for instance, then the only question is how many TLDs are you going to buy?
A quick check suggests that they just have ferrari.com, but with different versions and different languages depending on the location of the user
(e.g. www.ferrari.com/en-GB).
If you are a new startup, then you probably start from the other end.
You can think of a name for your company or product which will make a good (and available) domain.
Some companies have clearly chosen their name – and domain – to be memorable.
Memory experts suggest that to make a memonic you should create a bizarre mental picture associated with the thing you are trying to remember – the more bizarre the more powerful.
I can attest that this works.
Over 50 years ago I had to go for a job interview and I was told to ask for a Miss Bartram.
How do I – notoriously bad at remembering names – remember that name?
It is because I constructed a picture of a tram on a pub bar.
Some, web-based, companies have skipped one step.
Their very names conjure bizarre (and memorable) pictures:
moonpig,
SmugMug,
Snapfish…
Long URLs
This brings us to the crux of this paper.
Why would anyone choose a long URL?
I have been collecting these.
Table 1 lists the longest ones that I have found, although I would be grateful for contributions of any more which the reader may spot.
Table 1.
A list of the 22 longest domains that I have been able to find.
As apparent from this table, domains are often constructed by concatenating words together, and this particularly applies to long ones.
With no separation of the words, these can be hard to read – and to remember, and to type.
Hyphens (or dashes) can be used within domain names, and these are sometimes used (e.g. protect-rural-peterborough), but not often.
Is this because users might be confused by them?
URLs are case-insensitive, so capital letters could be used to demarcate component words.
(Sometimes referred to as CamelCase.)
For instance, would BluewaterBathroomsAndKitchens be easier to use?
I came across one example where different colours were used in their advertising to distinguish the words.
(Figure 10).
Figure 10.
The puraplumbingandheating URL on the side of their van, with different colours to highlight the component words in the domain.
Long concatenations of words can be difficult to parse.
This can even lead to howlers, such as when the company Pen Island tried to register penisland.com.
bluewaterbathroomsandkitchens
Figure 11.
bluewaterbathroomsandkitchens
This is the longest domain that I have found …so far.
Why would anyone choose such a long one?
Remember the suggestion above that a user needs to remember the URL to type it into their browser at a later time.
How are they going to remember bluewaterbathroomsandkitchens – and spell it correctly?
The company owning this site is called Bluewater Bathrooms and Kitchens, so there is an evident connection to the site, but is there not a shorter, more memorable domain?
Bluewater Home Interiors was founded in 2010 by business partners Adrian and Pete.
Starting out as bathroom installers, they worked together for many years to learn the trade before opening their first showroom in York.
In the early days, Bluewater’s aim was to provide a dedicated and professional service to customers seeking a full bathroom or kitchen installation.
So, the company was founded in 2010, when there was less of an expectation that (small) companies would have a website.
It also had a shorter name at that time, Bluewater Home Interiors.
It appears that they have always fitted bathrooms and kitchens, so why they changed (and lengthened) the name is not apparent.
They do appear to have registered a number of similar domains, which all point to the same site, including
bluewaterhomeinteriors.co.uk
bluewaterbathroomsandkitchens.uk
bluewaterbathroomsandkitchens.com
No variations on bluewaterbathrooms (18 letters) appear to be registered, but perhaps they feel that domain would be mis-leading, that people might not realise that they do kitchens as well.
It is not possible to cut right down, to (variations on) bluewater, because there is a large shopping centre of that name in the south-east of England
(www.bluewater.co.uk).
So, it would appear that this company is aiming for specificity:
that is the company's name and they do fit both bathrooms and kitchens.
Perhaps it is the case that they do not expect people to remember the spelling of their domain, but publicizing it does demonatrate that they do have a website,
and they can hope that people will remember enough about it to find it through a search engine, even if they forget the details.
That may well be the case for some of the other examples discussed.
YSSParentDiscount
An interesting example is shown in Figure 12.
The first thing to note about this example is the use of a QR code.
As advocated above, this is the best way to get users from conventional advertising to a website.
However, in case for some reason the viewer does not want to use the QR code, the URL is also given.
In this case it is not the domain that is long, but the path (YSSParentDiscount, 17 letters).
Indeed, the domain in this case is a very short one, (bit), and this is a web host (bit.ly) specifically designed to provide short domains.
Website owners submit their URL to this site and it generates a short one.
In this instance, if you go to bit.ly/YSSParentDiscount you are in fact taken to
yorksport2012.wufoo.com/forms/m18kp78u0dpw8s9/.
It is immediately evident why the shortening through bit.ly was chosen – but were the 17 letters of YSSParentDiscount really needed?
Figure 12.
Corner of a poster showing a QR code, and also the URL 'bit.ly/YSSParentDiscount'.
Discussion
Most of the long domains listed above are constructed from multiple words.
It seems that in many of the cases the website owners are tyring to be very precise:
Bluewater fits bathrooms and kitchens.
I am unsure of the motivation to include the definite article;
is there more than one Beverley events company, so that they have to emphasize that they are the one?
Three letters of and could be reduced to a single ampersand (&), if only they were allowed in domains,
but they are not – and anyway, if you were allowed mawremovals&storage you would probably want to register mawremovalsandstorage as well.
As discussed above, separating component words with dashes or CamelCase might make domains easier to use – although dashes would also increase the letter count.
It would seem that long domains are a means of tying the website clearly to the specific identity of the company.
Mis-spellings
Choosing a long domain increases the chances of the user mis-spelling it.
Suppose the user (tries to) memorise a URL – from the side of a truck or where ever – and then to type it into an address bar, it is easy to mis-type it
Domain.com suggests,
'It’s recommended to register at least 5 to 10 domain variations, including common misspellings, abbreviations, and relevant extensions.
This helps capture more traffic and protect your brand from competitors or errors.'
Given the power of modern browsers with search/autocorrect facilities, might a mis-spelt URL work regardless?
I carried out a small experiment to test this.
I wrote a program which would randomly change one randomly selected letter in a domain at a time.
I then clicked on the resultant link to see if the scrambled URL might nevertheless be effective.
It was not.
For instance, at one extreme, every letter of bluewaterbathroomsandkitchens could be substituted, giving www.zyiteczcsehkzdgfkbentgcrqavyl.co.uk.
Not surprisingly, this mangled URL yielded a 404 (using both DuckDuck Go in Firefox, and Chrome).
However, even a single change, www.bluewaterbathroomsandkitcheys.co.uk, was unreachable.
Problems of mis-spellings can be turned to the website's advantage.
The comparison website comparethemarket.com exploited potential confusions by playing on the idea that customers were going to
comparethemeerkat.com by mistake.
They riffed on this, creating such a site, which naturally re-directs to their real site.
This evolved into a large – and very successful marketing campaign.
Conclusions
Sometimes I write things in order to sort out my thoughts about the topic.
This is one of those instances.
However, in this instance I do not think anything has become clearer.
I have presented some ideas, but I still do not know why companies choose long domain names.
The ideal follow-up would be to find out if they work, are users more or less likely to visit websites with long URLs?
However, I cannot see any way of getting access to the necessary data, the logs of these sites.
I did supervise a student project on a related topic in 20035, but I no longer have access to that report, and as I recall it did not really address this question;
the student was more interested in whether an emotional connection to a URL would make it more memorable
So, I do not know what I conclude.
1
Pages which consist of text are generally in the format of HTML documents, where HTML stands for
hypertext markup language.
This will be referred to frequently in this article, but it is beyond the scope of this article to explain HTML.
Any reader needing more clarification on this language might refer to the
W3 Schools tutorial.
2
To be honest, this is a re-creation of that page, but the point still stands.
3
To anyone who is puzzled by this, there was a period when (benign) hackers would take any opportunity to modify URLs such that they linked to a video of Rick Astley singing Never Going to Give You Up.
This practice was known as Rick rolling.
4Ajax is another technical term which need not be explained further herein.
Interested readers might consult this Ajax Tutorial.
5
Cuerpo-Sanchez, A (2003)An experimental assessment of the memorability of web addresses (URLs)
University of York, Department of Computer Science, unpublished Third-year project report.
