The workflow before#
Nu Skin sells personal care and wellness products through independent affiliates in more than 50 markets, so its finance and supply chain teams need one view across very different regional businesses. The company had moved to AWS but brought its old patterns along. More than 150 overlapping ETL jobs took 6 to 8 hours to run and failed 15 to 20% of the time. Engineers spent about 120 hours a week keeping them alive. Business data arrived 24 to 48 hours late, and analysts spent 60% of their time preparing data instead of studying it. Meanwhile the cloud bill kept climbing, about $185K a month, much of it for capacity that sat idle.
What we built#
We rebuilt Nu Skin's data pipelines around a simple idea: fewer moving parts, one shared model of the business, and a cloud bill that matches the work being done. We consolidated more than 150 jobs into 40, gave analysts a global view they could explore on their own, and tuned the AWS estate until monthly spend fell from $185K to $128K.
Fewer, faster pipelines
Kafka now streams events from regional ERPs, e-commerce, supply chain and affiliate systems into an S3 data lake. Airflow runs the schedule, and Matillion reshapes the data and pushes the heavy work into Snowflake. We put Kafka between the source systems and everything downstream so a change in one regional ERP no longer breaks a dozen jobs that depended on it.
One global data model
We built a single model for customers, products, affiliates and the chart of accounts across every market, served through Tableau with role-based access. A finance lead can compare affiliate sales in Japan and the US without asking engineering to stitch the data together, and each team sees only what it should.
A cloud bill that matches the work
We reviewed the whole AWS estate against the Well-Architected Framework, then right-sized compute, moved steady workloads to Reserved Instances and flexible ones to Spot, and scheduled shutdowns for anything idle overnight. Cost allocation tags show each team what it spends, so the savings hold after we leave.
How we delivered it
The work ran over six months in four phases: assessment and quick cost wins, pipeline consolidation, the global data model and analytics, then governance and handover. We worked alongside Nu Skin's IT, data science and business teams throughout, with weekly workshops and hands-on training so the team owned the new setup by the end.
Before and after#
| Measure | Before | After |
|---|---|---|
| Monthly cloud cost | $185K | $128K |
| ETL pipelines | 150+ | 40 |
| Pipeline runtime | 6 to 8 hours | 1 to 2 hours |
| Pipeline failure rate | 15 to 20% | Under 2% |
| Pipeline maintenance | 120 hrs a week | 30 hrs a week |
What's next#
With one global data model in Snowflake, Nu Skin can move from reporting on the business to anticipating it. Next comes demand forecasting by market and an assistant that answers finance and supply chain questions directly from governed data, so planners get answers in minutes instead of waiting on a new report.
