Every member question answered, at any hour

The problem

ESOMAR receives a large volume of member queries every day, and the answers are spread across a website that is not easy to navigate: Directory, About, Events, Reports and Publications, plus the Research World blog, each its own tab to work through.

For a newcomer especially, finding one specific piece of information means skimming several pages. The FAQ covers the questions that come up often, and nothing else.

Conventional customer service has its own limit: nobody can be called at three in the morning, so a query raised outside office hours simply waits.

What we built

Chatbot Jacob answers a member's question directly and points them to the exact information they need, wherever on ESOMAR's platforms it happens to live.

It was built on the Xenioo platform and then extended with dev-UP's own AI models, so it handles natural language rather than only pre-programmed if-then rules, and copes with questions well outside the FAQ.

Natural language processing links different phrasings of the same query. "Where do I find my invoice?", "invoice where can I find it?" and "is there a way to find my invoice" all reach the same answer, so a member does not need the exact words to get help.

At the back end Jacob is connected through an API data hub to five or six separate data sources, which is what lets one question search across all of ESOMAR's platforms at once.

The result

• Members get the information they need without working through multiple pages themselves
• Questions that fall outside the FAQ get answered instead of going nowhere
• Different phrasings of the same question reach the same answer
• Support is available at any hour and from any location, and staff time is no longer spent answering the same questions by hand
• The AI model is reusable, and can be integrated into other platforms as a next step

Related case

Knowledge-base search without the ten-second wait