In the comparison shopping space, however, IAC is pursuing a build vs. buy strategy; In March 2005 IAC launched Giftts.com and Pronto.com was launched in November 2006.
Dan Marriott is the founding Chief Executive Officer behind Pronto.com. A ten year veteran of IAC, Marriot has spearheaded Corporate Strategy and Development, overseeing Mergers & Acquisitions. He also held senior operating positions at Ticketmaster-CitySearch.
Aggregated “targeted shopping services” are only capturing about $1 billion of the $250+ billion market; according to Marriott.
Marriott’s discussion of the new social shopping services competing in the online comparison shopping space—Stylehive, Kaboodle, Jellyfish.com—highlighted the challenge of gaining critical mass, and loyalty, among online comparison shoppers.
What is a long-term shopping aggregation winning strategy? I spoke with Marriott last week to get his first-hand take.
The IAC mission statement indicates its operating units seek to “make daily life easier and more productive for people all over the world.” Pronto.com aims to differentiate itself with such a technology-driven value proposition.
Pronto.com says it offers the “best comparison experience on the Web” and Marriott underscores the Pronto.com technology and business model support the claim:
Most other comparison shopping engines only display products from paying advertisers, products that they receive in data feeds directly from those merchants... Pronto.com deploys a patent pending technology to scour the Web and bring consumers every possible option of where to buy a product from all qualified merchants.
The Pronto.com technology is developed in-house out of Boulder, Colorado, where a team of about a dozen scientists apply data mining, artificial intelligence and decision theories to “structure and organize product information for fast and easy comparison.”
By capturing product information from “all” merchants, not just those that pay for inclusion, Marriott believes comparison shoppers will choose Pronto.com because it is unique in providing the lowest price product option.
The Pronto.com product-inclusive data model also aims to include items typically unavailable through other shopping search engines, such as out-of-season and less popular items carried by major retailers.
Pronto.com’ technology employs information extraction, information retrieval and semantic analysis to create a targeted product index from an unlimited number of unstructured and semi-structured sources, according to the company:
Crawling the web and extracting product information.
Directed crawling via a goal-directed, probabilistic model to determine: “Given the content of the page you are on and the content of a link to follow from the page, what is the probability the link leads to 'product oriented' information?" High probability = Crawler follows the link with an end result that the crawlers generate less load on internet web servers but fetch more "shopping relevant" information than any other crawl platform.
Information extraction algorithms with model fitting approach to transform semi-structured web pages into highly structured database records and determine: "Given a set of product-oriented web pages, automatically learn layout models that explain the placement and relationship of information elements on similar pages." A layout model captures information such as where the product titles are on a page, what prices go with which product and what images go with which each product.
Distributed Crawling with the Pronto web crawler operating as a massively distributed, broker based system to crawl and refresh 100s of millions of web pages in an efficient manner. Pronto's crawl cluster runs 100s to 1000s of crawlers in a fully distributed, yet fully coordinated architecture to keep data fresh.
Automatic product classification and contextual feature extraction.
Pronto uses a set of probabilistic text classifiers with artificats to automatically assign topical context to every product encountered. Lexical, linguistic and rule-based feature extraction is implemented with contextual bias to identify 20 types of "product features" and then group and filter products.Product clustering algorithms for automated product normalization enable discovery comparable products from disparate sources and allow for separating the products from the accessories for any product search.
Pronto Shopping Messenger
The Pronto.com value proposition also includes “Pronto Shopping Messenger” browser plug-in enabling shoppers to browse third party retail websites and receive real-time price alerts and promotional messages about the products they are viewing “so they don't overpay for an item.”
PRONTO.COM BUSINESS MODEL
To complement its customer facing value-proposition, Pronto.com defines itself to prospective retail partners as “a new distribution system that enables merchants to sell their products more intelligently.”
A core differentiator of Pronto.com’s comparison shopping engine is its no-fee required product and merchant inclusion strategy. Pronto.com aims to “up sell” merchants with enhanced listings and priority placement.
Pronto Merchant Solutions:
Get premium exposure for your product inventory through Pronto’s sponsored listings, similar items, and outlet modules
Elevate your brand through permanent logo placement whenever your products are seen
Increase conversion on select products through exclusive offer and coupon functionality
Participation in Pronto.com’s merchant program is based on a variable Cost-Per-Click fee schedule.
While merchants may self-enroll at Pronto.com, the Pronto.com business model includes a robust direct sales effort.
Marriott told me Pronto.com does not aim to rely on any other company for its revenues.