Case Study: Bigabid achieves real-time user profiling and sub-minute data freshness to supercharge mobile user acquisition with Upsolver

Building a Real-time Architecture to Supercharge Mobile User Acquisition with Amazon S3 and Upsolver

Bigabid, a mobile user‑acquisition and real‑time bidding company, needed to move from batch updates (user profiles refreshed 2–3 times a day) to real‑time user profiling to feed its predictive bidding algorithms. The company ingests TBs of data daily from multiple streams (bids, impressions, clicks, third‑party data) and required cleaning, enrichment and joins across streams, plus sub‑minute freshness — a project that Amit Attias estimated would take six months and four engineers to build in‑house before evaluating Upsolver.

Bigabid implemented a real‑time pipeline using Upsolver with Amazon S3, Kinesis Firehose, EMR/Spark and Athena: Upsolver performs the ETL, creates materialized views and writes compacted Parquet files for fast Athena queries, enabling joins, enrichments and advanced aggregates in real time. With an Upsolver proof of concept built in hours and a production system launched and maintained by a single user, Bigabid cut data latency from hours to under one minute, created continuously updated device profiles, and improved bidding decisioning and ad performance — making Upsolver a core part of their data infrastructure.

Bigabid

Amit Attias

Co-founder and CTO