Last Exit Westend

1. You fail to recognize that your web site is disconnected from your business.
2. You want all things different, but nothing to change.
3. You don’t know why your web site even exists.
4. You believe that your web site content attracts customers.
5. You hope you are buying the right technology with the redesign.
6. You think you don’t need a concept.
7. You are producing all content in-house.
8. You aim at outsourcing content maintenance.
9. You think your business is essentially the same as twenty years ago – just with computers.
10. You consider to establish a Web Analytics culture in your company through monthly reporting of web site visits, visitors, and page views.

Contrary to popular belief – and contrary to the recently started controversial discussions about “Catvertising” and “Kittywood” (more on that topic a bit further down in this post) it is a widely ignored fact that searches for <<dog>> or <<“dog video”>> outnumber the searches for <<cat>> or <<“cat video”>>.

So: why the hell is everybody talking about “Cats on the Internet” then?

Certainly: Lolcats. A tremendously successful Internet meme from 2007, featuring photos of predominantly cute cats with image captions that did bend the rules of orthography and syntax quite far (aka “lolspeak”. By now even academic papers are written about that topic). And this meme was quite successful, too: the search term <<I can has>> [this notation means: a search for all terms, no quotes involved] lists 8.740.000.000 results in Google. The same phrase search with quotes [notation: <<"I can has">>, you have gotten the idea by now anyway, haven't you?] brings up 44.300.000 results.

The according web site “I can has cheezburger” gets about 18.200.000 results , its dog equivalent “I can has hotdog” lists about 58.100.000 results in Google. Funnily enough, the domain name for the cat version is icanhascheezburger.com, while the dog equivalent lists under dogs.icanhascheezburger.com.

Nevertheless – as the recent talk about “catvertising”, “repurrters”, and “kittywood” properly illustrates (here’s an introductory read about the topic from no less than TIME Magazine (opens in a new window)), journalists enjoy talking about “Cats on the Internet”. And even the “Kittywood” vs. “Catvertising” copycat controversy [pun intended, really!] can be properly resolved in time (try it here: Google Trends: Kittywood vs. Catvertising.

The exact count difference for “dog vs. cat” is not depicted below, but as these graphs from Google Trends show, the difference in the relative amount of searches is quite significant:

One thing that is often mentioned when the question about the amazing popularity of cat videos on YouTube is mentioned/discussed is “But they are soooooooo cute”. Well – they certainly are. But that goes for puppy videos as well, as the following checks across Google for different search terms do show.

Combining the attribution of “cute” with the category searches of “puppy” or “kitten” repeats the findings for the initial category searches: around 5 million results for <<“cute puppy”>> vs. a bit over 3 million for <<“cute kitten”>>.
A side note: From the standpoint of a serious Web Analyst it is reasonable to assume that the success of the “cute [whatever]” search lies particularly in the exact match of a user’s intent with the delivered Search Engine results.

It is no surprise that search syntactics do matter a lot here in a stochastic sense: as we can expect, a search for a single search term returns a load of results: <<kitten>> produces 138.000.000 results, <<puppy>> produces 239.000.000 results. The kitten-to-puppy result ratio for the generic terms is quite similar to the one for the more specific terms <<cat movie>> and <<dog movie>>: 55.5% to 57.7%.
The term <<cute kitten>> returns 20.700.000 results (15% of the one-term search); the exact phrase search <<“cute kitten”>> returns 3.140.000 (2.2% of the one-term search).
For the puppies, it looks similar: 39.500.000 for <<cute puppy>> (that is 16.5% of the one-term-search) and 4.940.000 for <<“cute puppy”>> (that makes 2.1% of the initial one-term-search). So: the breakdowns are nearly identical, although the original volumes do differ quite a lot.

Using a more specific and complex search query showed some surprises, though: <<cute kitten sound>> brought up 30.800.000 results, the plural form <<cute kitten sounds>> returned about 1/3 of that (10.800.000), a search for the exact phrase/term combination <<“cute kitten” sound>> returned 3.740.000 results (12.1%), while the exact phrase <<“cute kitten sound”>> returned 18.000 results (0.05%). Interestingly enough the plural form <<“cute kitten sounds”>> gave only slightly more results: 26.500 (0.08%).
Testing the same with the “puppy” query: 1.290.000.000 results (gasp!) for the string <<cute puppy sound>>, but only 3.940.000 for the plural form (0.3%). 4.880.000 for the phrase/term combination <<“cute puppy” sound>> (0.4%), while <<“cute puppy sound”>> returned only 5.320 results (that is literally nothing that could be helpfully expressed as a percentage any more), <<”cute puppy sounds”>> returned 110.000 results (about 20 times as many as for the singular form of the search query. But still a value pretty close to nothing).

What do we learn from all this (besides that these numbers are huge)?
1. Puppies and dogs are more prominent on the Internet than cats and kittens. Except in articles about fluffiness and cuteness, and in relation to advertising agencies exploiting wordplay (“Catvertising”, “Repurrters”).
2. The neighbouring occurrence of the terms “cute”, “puppy”, and “sound” in documents on the Internet is A LOT bigger than that of “cute”, “kitten”, and “sound” (about 42 times as many returns for the former). This is heavily watering down relevant document retrieval for loosely coupled syntactic elements on target pages (worth a separate post).
3. The average amount of kittens depicted in “cute kitten(s)” images is close to 1.04. Approximately. I guess. Hm. Directly derived from that:
4. It seems that the predominant cultural meme of “cats = freaky, but fluffy and cute individualistic creatures” is mirrored on Search Engine Result volumes as well: about 1.1 million findings from an image search for <<“cute kitten”>> , but only 331.000 for <<“cute kittens”>> (that makes a singular-to-plural rate of 29.5%, as opposed to 70.9% for the “cute puppy” case).
5. The old idea that search term combinations would be primarily handled based on elementary Boolean Operators has become totally naught by now. Although the difference between a search for terms (the query is made without the quotes) and for a phrase (the sequence of search terms starts and ends with quotes) is still valid, the Google search is both a lot more capable of analyzing AND a lot more dependent on the existence of strong syntactic ties in documents on the web.
Elaborating on that thought surely is worth another post, scheduled for early 2012.

In case you couldn’t care less, or feel inclined to bridge the waiting time: why not go and watch some cute kitten videos in the meantime?

The term originally refers to landscape architecture, where it is describing a path in a park or green field that isn’t designed by the architects, but is created by people finding their own trails. They usually manifest as short-cut foot paths, eroding away the sward, and are often fought with fences and signs: “Keep off the grass!”

Looked at differently, these paths are documenting the perceived inadequacy and rejection of a plotted spatial design.

The article outlines the implications of “desire path” inscriptions for the field of web and service design. Let’s dive briefly into history first.

“In the beginning there was the homepage” could be a valid starting point for a history of web design. The homepage was a starting point for the user journey, a safe harbour for users surfing the deep waters of the Internet.

From there not much could go wrong originally. In the early Nineties homepages were barely more than linear text, with occasional headings and subheadings; a linear stream of words, only disrupted by underlined words in blue which marked connections to other pages, servers and sites. These connections were paraphrased as the auguries of a new type of non-linear text (“hyper text”).

Back then, all of these links on home pages were pointing to points somewhere in the unchartered terrains of the World Wide Web. For students pondering with markup language at the time they constituted a proof for their open-mindedness and connectivity within a multidimensional academic universe.

A lot has happened since then.
These days, commercial web site owners would rather change their web agency than to let any of the site visitors disappear from their carefully crafted home pages into any distant vault on a server far far away. The predominant idea today is to drag visitors deeper into a web site, comforting them with prominent unique selling points of products, with irresistible discount offers in web stores, and with positive customer testimonials – all supporting sales, or at least fueling an interest in a company’s products.

Yet web site visitors don’t really seem to appreciate the nicely laid out paths. Just as park visitors make their own trails by carving out footpaths offside the official tracks, web site visitors don’t follow the carefully crafted navigation paths either. More and more often we see users rather refining their search in Google than to bother with the navigation of a consumer web site.
As a result, web site visits are getting shorter and shorter, lasting for less than three pages on average, and with much shorter time-on-site values than two years ago.

The countless “Buy now” buttons that we can see on commercial home pages more and more often may make us believe that a quick and instant conversion is what people are looking for.
The underlying Web Analytics figures tell us a different story: Rates for purchase conversions stagnate well below ten percent across industries, Search Engine visits do most often end up on deeper navigational levels than the home page (which is a pity! Nobody sees the beautiful “Buy now!” buttons and teasers down there), and only occasionally these visits are traversing the site’s home page. And mostly, if they do: exit.

The “desire paths” web site visitors seem to have in mind surely are centered around taking the shortest possible path, but as Christoph Alexander has pointed out in his book “A pattern language” (in 1977!): “The layout of paths will seem right and comfortable only when it is compatible with the process of walking…” (pattern 120, Paths and Goals).

From here we dare to formulate a hypothesis: people promenading in a park are less likely to cut corners than people who try to reach their bus stop at the other side of it. We can tell the promenaders from those traversing by watching their behaviour. And we will see, undoubtedly: the “process of walking” has many faces.

In a similar manner we need to consider the web site promenaders as separate from the traversers and we need to make specific offers to accommodate both their needs.

Yet there is only one home page on each web site, and still the idea may prevail that there must be an ideal path layout. Sure – there are possibilities to prevent people from cutting the corners, but as Are Halland has laid out in his brilliant presentation “Core and paths” in 2007 the crystal clear design principles are too often obfuscated by the “Seven Deadly Sins of IA”, particularly by increasing the amount of choices.

In Alexander’s words: “To lay out paths, first place goals at natural points of interest. Then connect the goals to one another to form the paths. The paths may be straight, or gently curving between goals; their paving should swell around the goal.”

This approach, of course, presupposes a web site owner’s ability to cater for the most obvious short cut towards the goal. But on plenty of web sites we see the contrary: overcomplicated navigation patterns with multiple layers, flanked by category teasers, special product offers, obfuscated contact points, unclear pricing options hidden deep within the “place you order here!” funnel, etc.

Just as fifteen different brands of margarine don’t support easy decision making in a supermarket, offering five different ways to reach the same page on a web site doesn’t support easy decision making, either.

To design a web site from the home page may have been considered appropriate in the mid-Nineties. In 2011 you are wasting your money (as a business owner) and your time (or that of your agency) on this task.

Further reading:
Are Halland: Core and Paths (On Slideshare)

Christoph Alexander: A pattern language (New York, 1977). Available through a book store near you

If we compare the traditional industry corporation with a contemporary network organization we certainly will find more differences than similarities. If we, however, were to compare that same industry corporation with a traditional modern organization (like: an army), we supposedly would find more similarities than differences.

The classical way of tackling these comparison problems were centered around the identities that constitute the organization, thus aiming at their inner reality. Questions about hierarchy, corporate discipline, efficient decision-making methods, and corporate uniformity were in the very center of this description – and they still could be used to describe a successful company as defined by efficient means of hierarchical structure.

For our current economic environment (particularly in our saturated and volatile western markets) we need to concede a dramatic shift in the coordinate system a corporation needs to master when navigating the deep waters of successfully selling their products: a simple newsletter article about a company being in financial trouble can cause stock prices to plummet.

Which brings us to the core of the problem: it is no longer sufficient to look at the inner constitution of a producing company. Nor does it suffice to present or scrutinize a product portfolio. It is neither a company’s product range nor the inner organization processes (no matter how much the organization stresses its commitment to sustainability or Corporate Responsibility) that determines success from failure.

While for earlier times “control” was seen as both a justification for and an effect of a strictly hierarchical organization, we see a shift in these structures. “Control” has mostly become “indirect control” these days, and corporations aren’t controlled by their own decisions and their own Board of Directors any longer, but they are as well controlled by rating agencies and accounting firms, by their competitors, their customers, and the media.

At the same time new dimensions of control have been enabled for producing companies themselves. Customer Relationship Management systems, real-time logistics systems, and Web Analytics systems have increased corporations’ possibilities for knowing when, and where purchases, customer contacts, increased or reduced demands, and peaks in publicity do happen.

With the words of the German sociologist Dirk Baecker a control surplus is to be conceded: the increased possibility for control brings the problem of failure in control as its reverse side, and control projects of one stakeholder can turn into control problems for another.

The present discussions about the dominance of social networks, and particularly the challenges of controlling both privacy issues and sharing horizons show the bimodal character of the emerging control patterns.
Digital channels, it seems, are both well equipped for establishing control projects and likely to produce control problems.

For corporations (particularly those selling from business to business), the need for controlling and qualifying their prospect contacts (leads) has by now even superseded the importance of product development and portfolio management.

In practice, we see a variant of the same double-faceted control character at work: while the ever same content structure and distinction into “Our products” and “Our Corporation” is still present on most B2B web sites in 2011, we often find a lack of tailored means for entering structured communication with that company on these very same web sites.

Often, particularly with global organizations, “Contact” links often lead to a global phone book, where users have to specify their location (or the location of the department they want to get in touch with), their intent for the contact (sales inquiry, product inquiry, request for proposal, request for contact) as well as their potential interest in one or several of the company’s products.

For the potential buyer, this standardized procedure is at least inconvenient, as all the requested information could easily be seen as already known – if an inquiry is started from the pages of a particular product or service and is coming from a particular IP address (which can be attributed to the user’s geographic location), the only remaining unknown is the purpose of the inquiry.

While this contact process demands a potential buyer to comply with a control project started by that company, it at the same time shows that this very same company has a control problem as it proves unable to leverage the information already at hand. We can assume that corporations tend to focus their control projects on things and patterns said to be in their control: decisions about products and product portfolios seem to be handled more easily than decisions about subsidiary principles allowing local sales organizations to handle their inquiries on their own.

The question raised by the generic double structure of control projects and control problems is: “Which level of ignorance can still be considered harmless?” With the shift from direct to indirect control we have to add the question: “And from which perspective?”
For the given example the question appears in the form of: “Is it really just an inconvenience to make people repeat all their selections already made on a web site just as the department shifts?” “Does this ignorance make prospects turn away from our web site and our company entirely?” – or, in other, and more relevant terms for people with a business background: “Does this ignorance cost our company money?”

What we see here is yet another control problem: if the company could only figure out which of the alternatives are true – and in the terms of the topic discussed here, that is in consequence: if the company started another control project to validate the adequacy of their chosen control approach, which subsequent control problems would emerge from that?

As we have seen for the case of Web Analytics so far: Web Analytics tools offer means for starting and handling control projects. Any Web Analytics project creates further control problems: “What if we don’t like the data?”
Before that, even: “Can we make sure we even can utilize the data we are collecting?” And, if so: “Who owns this data and who has the right to make decisions about altering the data capturing methods, or internal business processes?”

The reason why we haven’t seen Web Analytics change the business culture of companies (except for a selected handful of digital marketplaces) is that any control project involving Web Analytics data capturing creates a set of subsequent control problems for the organization. And the strange thing is: both the data capturing and the problem that shall be solved with it seems to be imposed on the organization from the outside, but the consequences from this control project need to be carried and controlled internally.

With the given organizational constitution for most larger corporations, a culture of change is an utterly impossible thing, as the predominant corporate culture doesn’t provide enough means or resources for handling secondary and tertiary control problems caused by any new control project.

Companies still tend to re-assure themselves primarily by looking at what they can directly control – and that most often means their own inner realities: If sales decline, stronger sales efforts need to be made, lead conversion times are thoroughly looked at, and close scrutiny of the volumes of pending opportunities are the natural reflexes of a company with decreasing sales. The existence of a new competitor, a new competing product, or a new powerful sales channel utilized by other players in the market is not usually in the focus of an organization in trouble.

Maybe this is what has to change in corporate culture: the willingness to see the sequence of control problems and control projects as a continuously repeating pattern. The best lesson to learn from the current economic discussions is not to call for more or less freedom in the markets. Control, or lack of it, is not an ideal state, nor is it indicating a market equlibrium. As soon as a control problem is perceived, a control project is likely to be started to compensate. This will undoubtedly create new control problems which call for new control projects. And so on.
Any corporation or organization ignoring this simple mechanism isn’t very likely to last for long.

A lot of wise stuff has been written about the appliance of the EU Directive 2009/136/EC, often commonly referred to as the “cookie law”.

One of the most thriving discussions I have followed happened and happens at Brian Clifton’s blog (with a follow-up post summarizing the discussion – here).

A very catchy and interesting contribution to the discussion was thrown in by Vicky Brock recently.

A short note on the background: In anticipation of the enforcement of this new law (which was postponed in the UK to May 2012) the ICO (UK’s “Information Commissioner’s Office”) has put up an according opt-in disclaimer on their web site (Steve Jackson nicely attributed it as a “teletext ad”, see the ICO page here), asking the visitors for their consent to accept cookies.

Vicky has then made an FOI request to see the before/after effect on the tracked visits on that site. The results are non-ambiguous and look really dramatic. See for yourself here. Bottom line: the graph depicts “how traffic measured in the web analytics tool (GA) has fallen by 90% since their explicit cookie opt in request”.

Leaving aside the poor visual design of the opt-in message on ICO’s web site (I believe this to have been done on purpose) we can derive a couple of thoughts from this easily:

(1) If the loss in tracked visits (90%) is representative for other sites data accuracy for click stream data will suffer greatly from the opt-in obligation. Dramatically decreased sample sizes will increase error margins for analysis tremendously. To be a bit more dramatic on the matter: it will render web analytics data irrelevant.

(2) Tracking of users on the server side may seem to become a suitable method of capturing data – but only technically. The aim of the directive is to prohibit the collection of individually attributable data without users’ prior consent, regardless of the underlying technology. This is a tiny, but important difference and may even restrict server log analysis.

(3) As so often in history the legislation seems to be trying to sanction the tools, not the appliance of tools. To use a very manifest analogy: if you hand a knife to a surgeon and a murderer, they both will use it in their very own way. The current legislation seems to prohibit the purchase of knifes in the future to bring down the figures for committed murders. “Oh, these good intentions…”, you may say.
On a second thought, the directive is more subtle. To stay in the “knife” analogy, its aim is “to prevent people from being cut”, in other words: from the effects of the appliance of the tools – for whatever reason.

(4) The huge fines which come along with any breach of the “cookie law” legislation makes the usage of contemporary Web Analytics tools questionable for anything but commercial organizations (preferably: large e-commerce sites with large average order values). As I suppose that the fines will be imposed on a per-case basis, this could become as expensive as p2p file sharing has become for some recently.

(5) Alternative ways of tracking stuff could be found – but will/might still be considered illegal. It will be interesting to see how data on front end actions tracking (esp. Ajax calls and element tracking in general) can be replaced in the future, particularly if you consider the rapidly increased demand for properly sessionizing click stream data in recent years for syntactic analyses.

No doubt: We’re still talking about strictly anonymized click stream data here – and as clever folks have shown, it only requires smart combinations of filters to decrease redundancy in semantically and syntactically rich data. Combining rich data from various sources (catchword: profiling) may create personalized data that is forbidden to COLLECT without consent. But what about CREATING it?

Consider purchase history data from a web shop (anonymized) with credit card purchase history data (anonymized) from a credit card company. Imagine you could get your hands on telecommunications data (which is covered and collected through the Directive 2006/24/EC) – don’t we have the potential for aggregating data sets which can be attributed to individuals?
Depending on: which demographic markers come with the “anonymized” data?

Indeed a good point to get a closer focus on what’s so tricky about the case: just look at the EU Directive 2009/136/EC (strengthening an individual’s right on their personal data) and contrast this with the EU Directive 2006/24/EC (strengthening the authorities’ rights on an individual’s telecommunications data for 6 to 24 months) – don’t we see a slight contradiction here?

Undoubtedly: the cookie law will put pressure on all Web Analytics system providers and on the online advertising industry in Europe, as the effects on the predominant tracking paradigms are not yet clear.

But if you insist on looking at the whole thing with a more moralistic (but genuine European) perspective:
Of course the real battleground in European legislation is the applied distinction between volonté particulaire vs. volonté générale, and particularly the oscillation between the two sides. One directive sets the particular interest first (privacy policy), one directive sets the general interest first (telecommunications data). I am not a lawyer, so I only can ask a categorial question about it: what is the common underlying principle? The “averting of a danger for the society”?

And if so: is it reasonable to assume that smoking is so heavily restricted in most European countries I travel to for the same reason? Are my rights as a smoker restricted so I wouldn’t harm innocent children?
And if that is: why is smoking prohibited in so many areas, but not the sale and possession of cigarettes? So that everybody in society (including myself) would be spared from inhaling dangerous smoke, while walking the dog in the morning along a six-lane inner-city main traffic route?

I’m exaggerating, of course. But as you see: as soon as you can put yourself on the side of the general interest, you are in the possession of a moral wildcard.

For the ongoing discussion the analytics industry (as well as the online advertising industry) considers itself to be on the side of the general interest, too.
The “improvement of user experience/improvement of the service” which is stated in so many web sites’ usage terms as a reason for having (anonymous!) tracking in place might be regarded a site owner’s vested interest with regard to the general audience of the web site.

It’s somewhat hard to understand why this should be restricted. But it is equally hard to understand why a web site owner should be allowed to collect (potentially) personal data in order to maximize profits.
Even harder for me becomes to clearly draw the line between data collected as-is, and data that is created by being matched with other data from other sources.

Remember: The battle between “particular interest” and “general interest” is the very same battle that has already bestowed copyright laws on us which nowadays protect content distributors rather than content creators, and a battle that has allowed governments to treat uncomfortable information as “classified” at will.
It is the same battle that has weakened the idea of political immunity for particular individuals who allegedly have committed crimes. So: which general principle is at work here?

There is no right or wrong – there is only a picture puzzle that changes appearance, depending on your own starting point.

Returning to the topic itself: Taking the commercial interests of all the entrenched players in the online industry into account, the most likely scenario for me is that that the whole “cookie law” directive will be dragged to courts across Europe with a demand to scrutinize and exploit the text as well as the loop-holes in the directive itself. Any of the big players may take the first bullet on this (a multinational company thus would be my best bet for starting the law suit cascade).

Most countries in Europe don’t clearly distinguish between “written law” and “common law”. In those countries this distinction is to be applied by Supreme Court decisions, which nevertheless have to ground their decisions on basis of the written law. Judges have to decide, in other words, whether a case is covered by the written law or not.

The distinction between “rightful” and “just” is not supposed to be made (as “just” can be considered rightful only avant la lettre, which shall clearly not be the guiding principle).
Nevertheless the public opinion and discussion is as moralistic as in other countries, and the decision between “particular interest” and “general interest” is only implicitly (if at all) drawn in public discussions in Europe.
What is heavily present in these discussions (as well as in European jurisdiction, as far as I can tell as a layman) is the principle of the individual need to be protected against forces which are said to be more powerful or said to be more ruthless than the individual. And as smoking individuals are considered more ruthless than average individuals, protection doesn’t apply for them, morally seen.

Bad cards for the web analytics/online advertising industry from this perspective, I am afraid.

And just in case you haven’t been following the discussion about the “cookie law” too closely so far – the folks from Silktide have collected valid thoughts (and an excellent, witty, and partly polemic ebook about the nitty-gritty) on the matter (here).