UpSellit Advances SmartAgent Automated Chat Technology with Compound Recognition Engine

Share Article

UpSellit (, a leading provider of website abandonment solutions, today announced the release of the Compound Recognition engine for the SmartAgent chat platform. SmartAgent has recovered over 12 million online conversions for a wide range of online businesses with its proprietary automated chat technology. The addition of the Compound Recognition engine has proven to increase site conversions by allowing SmartAgent to better understand and address complex customer questions.

SmartAgent Automated Chat

Compound Recognition Advances SmartAgent Automated Chat

During our Compound Recognition BETA test, we noticed a 2.2% average lift in recovered conversions and about a 13% increase in SmartAgent response accuracy.

With a sophisticated Natural Language Parsing engine and an advanced knowledge base architecture, SmartAgent automated chat campaigns assist abandoning site visitors to reduce website abandonment and provide valuable consumer insight. Prior to the integration of the Compound Recognition engine, SmartAgent could only identify and respond to one customer question at a time. When a customer input contained multiple questions or topics, SmartAgent would simply identify the best match and provide a single automated chat response. The Compound Recognition engine enables SmartAgent to recognize and intelligently address a customer who has asked multiple questions within a single chat input. This increased efficiency has resulted in increased response accuracy, improved customer experience and an incremental lift in conversions.

"We're very excited about the addition of the Compound Recognition engine to our SmartAgent chat platform," commented Chris Wampler, UpSellit CTO and Co-Founder. "Over the years we've noticed a direct correlation between our SmartAgent's accuracy and the customer recovery rate. We are constantly expanding our clients' custom knowledge bases to improve accuracy, but the Compound Recognition engine has proven to instantly increase accuracy and enhance customer experience across all campaigns. During our Compound Recognition BETA test, we noticed a 2.2% average lift in recovered conversions and about a 13% increase in SmartAgent response accuracy."

After significant development and extensive BETA testing, SmartAgent can now effectively identify and address compound customer questions. When presented with a single customer input that contains multiple questions or topics, SmartAgent's automated chat engine will identify all appropriate responses. Marketers may now choose to either provide the single most relevant response followed by links to the additional responses or to have the automated chat agent send a series of responses to each of the customer's questions. As with all of SmartAgent's features and capabilities, clients are given complete control over how their SmartAgent will interact with their site visitors. Depending on the desired customer experience, marketers can choose the sequential approach to maintain the simulation of a live chat or the supplemental link approach for a transparent, automated chat experience.

For a demonstration of the Compound Recognition engine, please visit and click the chat link in the page footer. For more information about Compound Recognition and the SmartAgent chat platform, please write to info(at)upsellit(dot)com.

UpSellit was the first to market with a SmartAgent chat website abandonment solution in 2005. UpSellit has since lead the automated chat industry with significant advancements in both the Natural Language Parsing engine and the knowledge base architecture. UpSellit helps online businesses monetize their abandoning traffic with custom engagement solutions that leverage proprietary chat, email and survey technologies.

For more information UpSellit's automated chat release, please visit:

Share article on social media or email:

View article via:

Pdf Print

Contact Author

Jeremy Aaronson
Email >

Max Loewenthal
Email >
Follow us on
Visit website