AWS Identity Resolution using Zingg at DMG Events

DMG Events is a global exhibitions and publishing company with 13 offices across 10 countries and a portfolio of over 115 annual events, attracting more than 650,000 visitors worldwide. Contact data flows in from event registrations, paid attendee records, email sign-ups, and third-party system integrations. A small data engineering team centralises this data and serves it to BI dashboards and downstream marketing tools.

Contact data was fragmented across dozens of sources, making contact deduplication impossible and leaving the team with no unified view of who their customers were. The same person appeared as a registrant in one system and a paid attendee in another, with no entity resolution layer to link them. AWS FindMatches handled contact matching until AWS decommissioned it, leaving the team with a model locked to a single account that could not be exported, and no path forward without retraining a contact deduplication model from scratch.

DMG Events deployed Zingg Community as their entity resolution engine on AWS Glue using PySpark. The team spent a month training the Zingg model to match and better FindMatches results, delivering a golden record view of their entire contact base with full in-house ownership for the first time.

• 22M contact records from event registrations, paid attendee records, email sign-ups, and third-party integrations unified into a single identity layer
• Incremental entity resolution built using Zingg's linking phase, deduplicating historical data and daily deltas separately, then linking both together
• Model trained locally and ported freely across dev, test, and production AWS accounts, something impossible with FindMatches
• Stop words and column exclusions configured to handle null-like values and keep irrelevant fields out of matching
• Min and max confidence score thresholds tuned with business stakeholders, ensuring conservative matching critical because merged records cannot be unmerged