By QUENTIN HARDY

The world of Web marketing is based on the idea of search engine optimization, which means building Web pages that search engines can find and then drive readers to. But what if the right pages could come to you instead?

BloomReach, a company based in Mountain View, Calif., claims it has such a method. Staffed by former executives from Google, Cisco, and Facebook, the company has spent three years developing a way to look at one billion Web pages a day, divine what kind of products and services they might have, and then by looking at a broad range of customer interests, deliver Web pages that have just the right items and descriptions to suit an individual consumer.

“There are 10 to the 30th power different ways just to describe flat panels,” Raj De Datta, co-founder and chief executive of BloomReach, said in an interview. “What is the optimal way to describe that to a specific consumer?”

As big data problems go, this one is pretty daunting. The core of their business is a “Web relevance engine,” which uses machine learning and search techniques to gather data on content and user behavior. It then adapts Web sites to show what the algorithm concludes is the relevant content for that viewer. While looking to deliver a personalized experience, Mr. De Datta says, the engine treats the data as anonymous content. The software used to control for variations in the descriptions, he says, may eventually be open sourced.

“Adaptive Web pages dynamically insert information, then watch the behavior and fine tune,” Mr. De Datta says. “We have driven 80 percent increases in traffic for some customers,” which include Williams-Sonoma and Oodle, an online classifieds listing site.

The initial products are aimed at search, online advertising, and social media marketing. The company works on a pay- for-performance model, based on things like improved customer traffic and conversions to an online sale. The company is backed by, among others, Bain Capital and Lightspeed Ventures.