September 13, 2006
Case Study

Results from CafePress.com Tagging Test -- Clicks & Conversions Increase

SUMMARY: Tagging is the new hip Web 2.0 buzzword. Seems like everyone is testing an aspect of it, from Amazon to Burpee. But few have results to share yet.



So, you should find this Case Study particularly interesting. Discover how CafePress.com invented a new banner ad system based on merchandise tags that's improving clicks and conversions, especially for ads run on blogs:


CHALLENGE

Last year, the marketing team at CafePress.com were frustrated by the ecommerce site's own search engine. Like most merchants’ sites, the search engine surfed words in the title and copy for products to decide which SKUs to show consumers who are searching.

The problem was that when you write descriptive copy -- or in CafePress.com's case, when any one of hundreds of thousands of participating merchandise creators wrote their copy -- it's awfully hard to include every keyword or phrase that a consumer might use when searching.

So, the team rolled out a new internal search engine and related content management system tweaks that would allow copywriters to add a list of tags (descriptive words and phrases) to each SKU -- all 40 million of them.

It was a gargantuan project. The team began to wonder, how could they get more ROI out of the merchandise tags than simply better site search?

CAMPAIGN

CafePress.com had launched another big initiative last fall -- an affiliate program. Sounds crazy, most ecommerce sites launched affiliate programs eons ago.

But in CafePress.com's case, they already had thousands of shopkeepers across the Web operating ecommerce stores powered by the service so that these acted as de facto affiliates.

Why launch affiliates in addition to shopkeepers? "Bloggers," explains Maheesh Jain, VP Business Development. After seeing how Google's AdSense program expanded across the blogging community, Jain's team bet that thousands of bloggers might be interested in monetizing their hobby by carrying some CafePress.com ads as well.

The affiliate recruitment home page on CafePress.com was even headlined, "Do you have a Web site or blog?" instead of the more traditional "Join our affiliate program."

At first, the team handed out fairly traditional banner ads featuring product shots for each of just over two dozen general categories, such as "dogs." (Link to sample below.) But, when the merchandise tagging initiative hit a pre-defined SKU level four months after launch, the team invented an all-new type of contextual ad they called TopicAds.

"There needs to be a critical mass of content for it to work," notes Jain, "so that no mater what you type in you will find relevant offers." In effect, you're taking advantage of the long tail.

The new ads pulled up product offers based on tags. The blogger or Web site would type in a list of topics they wanted their display ads to feature products on -- for example "cocker spaniels" or "California Democrat politics." Then the ad would dynamically pull similarly tagged merchandise offers from the database. (Link to sample below.)

Because they were based on human-typed tags, rather than software that decided on contextual relevance, these ads might be far more relevant to the blog's subject than a Google AdSense ad would be.

Plus, because CafePress.com has thousands of new SKUs being tagged daily, it was likely that the product would be available for even highly newsy topics that a typical blogger might write about.

Last but not least, the tech team created a few apps for popular blogging software that made it easier for bloggers to pick up and run the ads. In addition, they created a separate format for MySpace users because MySpace was blocking outbound Flash-based links.

RESULTS

Since launching the affiliate program last fall, roughly 19,000 affiliates have signed up. A small percentage are professional affiliates who prefer to create their own stores and campaigns. The largest chunk are bloggers.

These bloggers responded highly favorably to the new TopicAds format. "They obviously had lots of suggestions looking at it in beta for ways to make the interface better, but overall the feedback was this will make us more money," Jain says.

And, indeed, it does. Results so far show that on average TopicAds get 23% higher clickthrough rates than traditional banners and those clicks convert 12% better than other banner clicks.

Jain strongly suspects this trend will continue to improve slightly over the next year as both communities -- merchandisers and bloggers -- learn how to tag more effectively.

Although CafePress.com has lots of instructions on this topic -- in particular trying to stop people from using too general tags ("dogs" instead of "cocker spaniels") and from tag spam (where you tag with phrases that don't really pertain directly to the product), it takes consumers a while to learn.

In the meantime, Jain is thrilled with the results so far. "The biggest surprise is the diversity. When I'm bored, I'll start typing in random words into search to see what shows up and now it's just crazy. It's the long tail in action."

Useful links related to this article:

Creative samples from CafePress.com's tagging initiative
http://www.marketingsherpa.com/cs/cafepress/study.html

Flickr -- photo sharing site based on tagging
http://www.flickr.com

del.icio.us -- Web bookmark sharing site based on tagging
http://del.icio.us/

CafePress.com
http://www.cafepress.com

Improve Your Marketing

Join our thousands of weekly case study readers.

Enter your email below to receive MarketingSherpa news, updates, and promotions:

Note: Already a subscriber? Want to add a subscription?
Click Here to Manage Subscriptions