Information Squid

In Everything is Miscellaneous, David Weinberger’s thesis is that digital objects aren’t stuck with one type of organization. Instead of an item being on one particular shelf in a given store, items can be found by many different characteristics. Last weekend I was reminded that’s one reason I rarely shop in physical stores - but the online ones haven’t solved my problems either.

I was in Macy’s, looking for a coat. Since it’s fall, I’d have expected there to be an outerwear section in the store where I could look at all the possibilities. No such luck. Instead the coats were scattered by designer: a clerk explained that the Ralph Lauren coats are over here, the Nautica ones are over there, and there are a few more scattered on the floor.

Certainly that can be the right organization. If I’m shopping for myself, I often look first at brand, then color, then size. But in this case, when I wanted an overview of all the options within a particular category, the fact that the store hadn’t sorted by that category made it nearly impossible to shop effectively.

At macys.com, I can see all the coats. Sorting by type of apparel is even listed in the left navigation above sorting by brand - bricks-and-mortar stores, take note. Once I’m looking at coats, I can narrow my view further by brand. Some stores (i.e. Nordstrom) will let me sort by brand also, so I can still see everything but can easily compare within each line. But I’m still not happy with my shopping experience.

Why? I still haven’t seen a department store that will let me search only for “brands I wear”. Simple customization, right? Checkboxes in my profile when I register, and if I’ve filled out that section, offer me a personalized search. Then the store can also target its email marketing to me, meaning I’ll enjoy receiving useful emails (branding) and will buy some of the “five Nine West items on sale in our shoe department!” (direct response). Is anyone out there doing this? Am I just not registered at the right online stores?

Following up on the concerns about how to use Yahoo SmartAds, the answer is probably to use a souped-up creative optimizer, ideally hooked into the ad serving system itself. That way you can analyze your results and then show your most effective ads most often (or modify the less effective versions). Creative optimization has been around for a while for search campaigns, trying to improve performance for certain keywords. Advertising.com in November 2006 released a study of keyword performance using three optimization techniques.

It’s harder to find information about optimization solutions for advertising, perhaps because everyone is trying to keep their methods proprietary. WPP’s (Ogilvy’s) mOne developed a tool called mEuclid and publicized it in 2005. That seems likely to be the same tool mentioned in The Quest for the Perfect Online Ad (Business 2.0, 4/3/2007). The Atlas Institute of aQuantive has done some writing about creative optimization (also in 2005), though the referenced white paper says to worry more about ad placement than about exactly what’s in your creative.

Still, given the option to have potentially thousands of combinations of ad components, agencies that take advantage of SmartAds will need a way to manage and analyze the data they collect. Erick Schonfeld’s The Next Net picks up the same concern about how many items marketers will need to juggle. Seems like Ogilvy has a good setup already - I wonder whether the other agencies are behind or just quiet?

Separate from all the hoopla this week about the iPhone, advertising has also made a leap forward. Yahoo has announced a system called SmartAds for on-the-fly composition of display ads - meaning a firm could set up an ad for an item the user has recently searched for, with its price in their location, as well as images/colors appealing to their demographics. The system can use behavioral, geographic, and demographic information to pick ad components.

This isn’t the first system to use behavioral information, but it seems to be the first offering customization at the level of parts of ads rather than at the level of the whole image. So then the problem is figuring out how to use it.

The NYTimes’s Bits blog brings up the point that most companies don’t know yet how to take full advantage of behavioral targeting. Thousands of versions of an ad are hard to manage - how do you tell what’s working? Tracking the clickthrough information will be as much work as tracking web analytics for a good-size website - especially as companies move toward microsites in display ad spaces. Will this become a new specialty within marketing/advertising companies?

NYTimes article (registration required)
NYTimes’s Bits blog
Wall Street Journal (subscription required)
CNet
Editor and Publisher
ClickZ
Search Engine Land
PC Magazine’s AppScout blog
Red Herring
New Media Age (UK)
NewsFactor.com
Marketing Shift blog