Demo Fest, now in its fifth year, is a way for the researchers within Microsoft’s adCenter Labs to get their ideas and prototypes out to the company’s product groups. The adLabs group showed four prototypes to a handful of reporters in Redmond this morning. That’s a small sample of the 25 demonstrations and 10 poster presentations it will have for full-time employees to see later today.
The demos I saw were not jaw-dropping. The concepts seemed to echo things we’d seen elsewhere or things I would have thought were already table stakes for advanced online advertising (understanding whether a user is searching for local information, such as a dentist in Seattle, for example).
AdLabs general manager Eric Brill was asked whether other demonstrations pointed more clearly to Microsoft’s strategic direction for online advertising.
Brill would not go into specifics, but talked about the “high-level” view of three sets of things the company wants to work on:
- Known problems where the solution is known, but more solid engineering is required.
- Known problems that don’t yet have solutions, such as improving search ad relevance.
- Still unknown problems and user desires.
“It’s kind of the process of exploration,” he said, expanding on the third area. Microsoft aims to improve its process for getting products in front of users, taking feedback and quickly evolving them. “… That’s one of the big keys and that’s one of the core competencies that we’ve really been building up.”
Here’s a rundown of what adLabs shared publicly, with the caveat that these are still in the idea/concept stage and may not turn into actual products:
Gift Matching. A shopping tool designed to help you find a gift that will both surprise its recipient, but also be something they actually want — a tough balance to strike. The user inputs the intended recipient’s sex, age, location and interests. The tool brings back recommendations based on Live Search query data. For example, if one of the interests is chess, the tool may recommend a backgammon set because people who search for chess also search for backgammon, said Rohan Shetty, an adLabs program manager. He emphasized that the tool is designed to help find the unexpected after the tool recommended oolong tea for a 30-year-old Nascar fan.
Microsoft Gaze. Already in a private beta test, this tool is designed to amp up in-text advertising for Web site publishers. The tool allows publishers to embed a few lines of Java script on their site that launches a pop-up gadget presenting information about select keywords (celebrities’ names, for example). The content, currently limited to what Microsoft can license, includes photos, videos, career timelines and other data displayed in the pop-up window. The gadget, which is highly customizable, includes advertising. Mario Esposito, another project manager, said it can double the number of contextual ads displayed.
Display ad tool. Aimed at cutting the cost of creating online display ads for small businesses, this tool searches through a database of images to automatically generate several ad layouts. The advertiser writes the text of the ad and selects a category, such as entertainment or clothing. The system takes that information to search its database and select an appropriate image and automatically create an appealing arrangement of text and image. Advertisers can also upload a logo.
Location Platform. The tool tries to better predict whether someone performing an Internet search is looking for local information, or information about a travel destination through a machine learning algorithm that analyzes search queries. It also cross-references a searcher’s current location based on their IP address. For example, a search for “Seattle pizza” from a computer based in Seattle is likely to be a local search. A search for “Las Vegas hotel” from the same computer, may indicate intent to travel. Advertising can be sold along side the search results accordingly.