Coffee, Sun & Technology

Competition online is fierce, and many online retailers do not want to talk to their direct competition about new marketing programs or general techniques to increase the bottom line. A few of my customers have taken a different approach for exchanging cool ideas - buddy’ing up with other sites which aren’t directly competing. If you sell wine, find someone who sells travel - if you sell shoes, find someone who sells musical instruments. It’ll make you feel confortable when talking about your site and will give you another perspective on your own business.

And maybe that’ll help solve the bigger issue of sites starting to look all the same, within a given vertical (see earlier post “Templated Experiences”)

Web sites are getting cleaner and easier to navigate. That’s good. But they are also getting more and more templated, making user experience’s more uniform across sites. Take a look at these screenshots - 1800flowers and Proflowers – Dell and Gateway – MSN and Yahoo. Even Saks Fifth Avenue and Crate and Barrel look alike, even though they are not competitors.

Top 5 reasons for this happening:
1 - The amazon.com syndrome. Everyone wants to be an Amazon. Same cool ideas same dumb mistakes.
2 - Fear of the competitor. How much time are we spending looking at our competitors’ site? Again, same cool ideas, same…
3 - Risk adversity. We don’t want to take any risks, no funky homepage or innovative product presentation.
4 - Best practices. We hired a consultant who told us that checkouts should be 3 steps. And 3 steps it is.
5 - Resources. It’s easy and cheaper to maintain a templated site. No money for innovation.
…And there are probably 15 other reasons.

By making sites undifferentiated, online businesses are making pricing become the determining factor. If my user experience at Proflowers is identical to my experience at 1-800-Flowers then price is all that is left to convert the undecided buyer.

I have no data to support what I’m saying, but I think there is much to be gained if sites claimed their right to be different — and executed on it.

Jupiter Research published a solid report on cookies today - see this press release.

99+% of commercial sites require user to have cookies turned on for making a purchase, such cookies are typically session-based and expire automatically after 30 minutes or so. Another type of cookies, called persistent cookies, are stored for a longer period of time and are used to identify returning users. If you’ve been to a site recently and have seen the message “welcome back Paul” without login to the site, it is probably because a persistent cookie was set on your computer.

Jupiter polled thousands of end users on cookie usage and the results are appalling: up to 40% of users delete all their cookies on a regular basis, as often as weekly or monthly. The consequences are significant for a number of applications: personalization, Web analytics, advertising, etc. If persistent cookies are deleted we cannot accurately track users over multiple sessions for long periods of time. We cannot target anonymous users based on past behavior on the site. We cannot deliver ads which are more relevant and targeted. Everyone’s Joe Blow.

For Web analytics, a few questions arise (assuming this research is reliable). If 40% of users frequently delete their cookies it means Marketers are underestimating all deferred conversions resulting from older marketing campaigns. They are also overestimating the number of new prospects (or first time users) who are visiting, since some of them might be returning users who cleared their cookies.

Arguably the research is seriously denting any hopes of using the Web to track individual users for long periods of time. Lifetime value can still be measured down to individual users, but it’s getting harder and harder to know how users are behaving between 2 purchases: what products they looked at, how many visits before a purchase, initial entry point to the site, etc. Many of us analytics experts have dreamed of using individual click-stream patterns to drive more personalized, relevant marketing campaigns, at the right time. “If user X has visited the DVD section of the site 3 times in the last 12 weeks, send him an email with a coupon”.

On the other hand maybe it is time for all of us to acknowledge how perishable and fragile Web data is.
- Perishable since most modern sites update their site and content at least every 2 weeks, since product selection constantly changes, since marketing campaigns are short, extremely seasonal and extremely dynamic. As an example, does it make sense to run a home page path analysis query for December 18th today March 10th? Are you hoping to find out something interesting?
- Fragile since data collection and accuracy issues have been present in the Web analytics space since day 1. Whatever the technologies (logs, ASP, in-house), it is hard to get a real accurate assessment of online activity. Period.

So if we look back at today’s announcement, yes we are learning that maybe our returning users reports aren’t as accurate. But the damage is relatively contained and we should be able to work around this. For example, think of marketing campaigns. Most of my customers track their campaigns down to individual users within 7 days of the initial entry, no more. Deferred campaign data will be off, but not by much since few people delete their cookies every week, according to the research.

Overall, this research will inevitably influence the future of Analytics. But if it forces us to better understand and use Web analytics data, this could in the end be a good thing.

I was at a JupiterMedia conference this morning, attended a presentation delivered by Eric Peterson on Web analytics as they apply to Search Engine Marketing. Eric made a number of interesting points about vendor selection - but most significantly he truly believes that “people matter” . It’s neither technology nor features.

I wholeheartedly support the idea that people are the essential component of a successful analytics “program”. Who will use the application and how - will they look good in front of their boss when they show a report - why should they care about analytics in the first place.

As interfaces get easier to use, naturally more will dare using analytics. My question back to Eric however is…: 90% of us are mathematically challenged. However sexy you make these apps, they remain business tools, with decisions to make based on numbers.

We geeks would love everyone else to be geeks. I go crazy trying to explain my mom how to use Yahoo Messenger. Democratization of analytics. It’s a good concept but “can” it happen?

1 - Infinetly fast disks and I/O, for very cheap
2 - A way of tracking users anywhere, i.e. when they browse, when they visit a regular store, when they open a catalog, etc.
3 - A set of tools to act immediately on Web analytics information and reach users for cheap (today this can be email - but email is becoming rather ineffective these days)

Some guy in Australia has found a way to uniquely identify computers connecting to a site without the use of cookies - just by monitoring their system clock: See this article..

OK. Since for a given computer, you can expect (in general) just a few real people using it, we’re not too far away from being able to uniquely identify people visiting a site. Scary.

Not sure what to think about it yet. I’m inclined to believe that such technology will never be commercially available. If it does however, I will be the first one to buy a blocking software against it. I like the internet for it being anonymous - or at least giving me the option of being anonymous.

Here’s a pretty good overview of guided navigation, a new hot eCommerce feature, and some good tips on product presentation. Take a look: Presentation PDF. This was presented by Kevin Kearney from Avenue A/RazorFish. I like it also because he talks about quite a few Fireclick customers I personally know. A few good points there:

First, on guided navigation, Kevin talks about how complicated and disconnected categories can confuse users. Other flaws (according to him) include: Truncated list of attributes (users need to hit the button “More” for finding their category), limited categories, complexity of the navigation, hidden functionnality and poor page layout. He gives a few real life examples.

Second, on product presentation. Again, a few very good points, including the fact that you need to be clear in your product pages about pricing and purchasing options. But the most interesting point to me is about cross sells. Many of my customers are currently working on better cross sells - which in many ways are an extension of guided navigation.

Way too often, as explained in Kevin’s presentation, cross sells are often a list of product in the same category, nothing else. I highly recommend you use your analytics reports to figure out which products sell well together, and start pushing those. It works really well.

In a pretty good report published today, Jupiter Research estimates the Web Analytics market to be at around $500M this year (2005). This number seems inflated but overall I agree with some other key points they’re making:

“We believe that the available market for applications like yours is only about 15% penetrated”. This seems correct, based on my own market and competitive research. But I also think the market will saturate at 30% penetration max. Lots of businesses out there are not willing to spend lots on analytics. Period.

“We predict strong but slowing growth in ASP analytics and a move back to software solutions, driven primarily by the need for data privacy.” I agree. For this to happen however, a strong software vendor has to emerge. ASPs currently offer superior products.