Finding the Relevance in Relevance Engines
"A relevance engine is designed and built on the foundation of application, device and network neutrality -- because it must connect all people, processes and information, in a variety of locations, using all types of technologies," explains Troy McAlpin, CEO of xMatters, a relevance engines vendor that services seven out of ten of the largest Global 2000 companies.
This adds a new dimension to business intelligence. If you know Customer X never checks his voicemail but always uses text, then you know to text rather than call him. If Customer Y hates sports news but often buys team apparel for her children then you know to send her offers on children's licensed products, but don't bother her with game scores.
Relevance engines can, at least theoretically, tell you this information and more on every individual in a group no matter how large the group is. This is highly desirable because customers are more likely to respond favorably to communications that actually fit their needs or desires. Cha-ching!
Buzz and bust
While the term relevance engine is just now emerging on the buzz front, it incorporates older technologies albeit with some newer Web coding languages such as RDF (resource description framework), OWL (ontology Web language) and a few others that help search engines understand "natural" language. But, at its core, the technology is much simpler -- a push technology applied to segmented recipients.
"The basic idea is not new," said Kate McArdle, vice president of Product Strategy and Management at Varolii, a relevance engine vendor. "The piece that's new is being able to use detailed transaction history as part of the overall segmenting process."
McArdle said the two most powerful aspects for relevance engines are integration and self-learning. "Self-learning is key: if a company has to keep a team of scientists on staff to continually tweak business rules and algorithms to ensure the technology keeps working, it's self-defeating from a cost-containment standpoint."
Indeed, the most sophisticated relevance engines do self-learn. As they get more data from each subsequent interaction, "they get smarter and smarter about how to communicate with each individual in the database and that's really what the promise of a relevance engine is all about."
Sorting the relevance
"The assumed benefit to the enterprise organization is mass personalized and targeted customer management resulting in augmented sales with happier clients -- CRM on steroids," said Erika Brown, executive vice president of Corporate Strategy at Frost & Sullivan.
But "[u]nless there is a firm, hard wall somewhere to lever the customer, a pay wall, a controlled piece of hardware, and firm chargeable applications, it might not work," warns Brown.
In the years before the now, the underlying technology in relevance engines was called by a variety of names ranging from semantic search and contextual engines to "decisioning engines." For example, Bing is advertising itself as a "decision engine" in the hopes of differentiating itself from competitors that merely deliver search results by matching key words. Google and Yahoo, not waiting to be outdone by Bing, quickly launched site redesigns that offer more structured data and credible sources (as opposed to merely spewing a long list of popular links) and a relevance engine setup of their own.
This collection of different but related names has clouded the relevance concept exactly as the words "hosted" and "software-as-a-service" cloud, well, the cloud. "The terms are more or less interchangeable," said Eric Bryant, director at Gnosis Arts Multimedia Communications. "The purpose of a relevance engine is to discern the semantic intent of the searcher, in order to deliver a 'more relevant' result. It's the whole, 'all semantic search engines are relevance engines, but not all relevance engines are semantic search engines' idea."
For instance, a search engine like Topsy, which is based on Tweets (in Twitter), is a kind of relevance search engine "but it is not a semantic search engine," he explains.
First adopters are using relevance engines in creative and individual ways. Certainly the leading search engines are trending hard in favor of delivering search results that are both credible and tailored to the searcher. But relevance engines are relevant to more than the search industry. xMatters' CEO Troy Alpin offered these examples from his own client base as a sampling of what companies can accomplish with the engines:
- Sprint Nextel was able to consolidate a legacy notification system with more than 56 individual and geographically distributed application servers, improving operational effectiveness and automating manual processes.
- Denver International Airport uses a relevance engine to send 14,000 messages in just minutes; notifying key airport personnel of a sudden security breach, severe weather, etc.
- A large grocery chain uses a relevance engine to send out critical recall notices to all store managers around the country in a couple minutes.
- Thomson Reuters uses a relevance engine to deliver internal and external messages to over 25,000 customers, notifying them of disruptions to the real-time data that can potentially cost their business their revenue and reputation. For example, the Financial Markets Exchange might send incorrect or suspect data to Thomson Reuters, which may impact data retransmitted to customers. A relevance engine would assist in sending information to the relevant traders notifying them of the suspect data before it impacts their customers. The system will also be used to inform subscribing clients of system maintenance and data changes, ensuring clients only receive information that is of interest to them.
The uses vary as much as the companies that deploy the engines do. Indeed, there seems to be no natural limits for how relevance engines can be used. This of course, adds to their appeal. "While the technology is still in the early adopter phase, it is expected to move rapidly to mainstream adoption as awareness of its benefits grows," said Drew Kraus, research vice president at Gartner.
A prolific and versatile writer, Pam Baker's published credits include numerous articles in leading publications including, but not limited to: Institutional Investor magazine, CIO.com, NetworkWorld, ComputerWorld, IT World, Linux World, Internet News, E-Commerce Times, LinuxInsider, CIO Today Magazine, NPTech News (nonprofits), MedTech Journal, I Six Sigma magazine, Computer Sweden, NY Times, and Knight-Ridder/McClatchy newspapers.She has also authored several analytical studies on technology and eight books. Baker also wrote and produced an award-winning documentary on paper-making.Baker is a member of the National Press Club (NPC), Society of Professional Journalists (SPJ), and the Internet Press Guild (IPG).