On-Demand Webinar · 1 hr 1 min

The Celgene Story: Building a $1B Product Launch Success with DataOps

Rajesh Gill, Associate Director of Commercial Insights at Celgene, on how the team used DataOps to support the Otezla launch. Recorded June 2020; updated August 2026.

What you'll learn 6 points
  • Bringing a drug to market costs $2.6 billion, according to the Tufts Center for the Study of Drug Development. The first six to twelve months of a product launch are decisive, because how fast a product grows during launch determines its lifetime revenue.
  • The Celgene commercial analytics platform ran on Redshift feeding Tableau Online, with more than 1,000 dashboards serving hundreds of sales people plus marketing and executives. Inputs included syndicated data, sales data, Rx claims, specialty pharmacy, NPP events and campaigns, sales alignments, product hierarchies, and specialty mappings.
  • On the DataKitchen platform, Celgene integrated hundreds of data sets into a mastered, unified star schema with more than 20,000 automated tests, absorbed over 100 schema and data changes per week, and had very very few errors or missed SLAs at low total yearly cost for hardware, hosting, software, and staffing.
  • The engineering side of a new large data set followed six steps: build a scrappy star in a data mart, send questions to the data supplier while keeping analysts in the loop, add data tests, share the star with the analyst team for feedback, iterate over several Agile sprints, then release a solid star.
  • The analyst side ran in parallel: build scrappy dashboards, feed corrections back to data engineering, show early dashboards to users, hold active build and design sessions making as many changes live as possible, and publish production dashboards at 70 percent done.
  • The DataOps mindset shift is five substitutions: change fear becomes change velocity, manual operations become automated operations, hope for quality becomes integrated quality, hero mentality becomes repeatable processes, and perfection becomes 70 percent right the first time.

Slides

37 slides

Questions from this session

Why is the first year of a pharmaceutical product launch so important?

How fast a product grows during its launch determines its overall lifetime revenue, which puts the weight on the first six to twelve months. The stakes are set by what came before: bringing a drug to market costs $2.6 billion, according to the Tufts Center for the Study of Drug Development. The launch happens against increasing marketplace complexity, fierce competition, and continued pressure on cost.

What is a scrappy star, and how did the Celgene team use it?

A scrappy star is a first-pass star schema built quickly in a data mart when a new business question requires a large new data set. Data engineering built it, sent questions back to the data supplier while keeping analysts in the loop, added data tests to make speed safe, shared it with the analyst team for feedback, and iterated across several Agile sprints before releasing a solid star. Analysts worked the same way from the other end, building scrappy dashboards and showing them to users early.

Why publish a production dashboard at 70 percent done?

Because feedback from real users on a partial dashboard is worth more than a longer wait for a complete one, and because the remaining 30 percent is usually the part the team guessed wrong. The team ran active build and design sessions, making as many changes live as possible and updating through Tableau Online. Seventy percent right the first time is one of the five substitutions in the DataOps mindset, replacing perfection.

What results did Celgene get from DataOps?

The platform integrated hundreds of data sets, mastered them into a unified star schema, and ran more than 20,000 automated tests. It absorbed over 100 schema and data changes per week with very few errors or missed SLAs, at low total yearly cost across hardware, hosting, software, and staffing. The team supported ongoing production deliverables such as a weekly launch and performance tracker, ad hoc answers for business leaders, and resource allocation and predictive models.

What should an analytics leader keep under their own span of control?

Five areas: agile team management, a DataOps technology platform, data engineers with their tools and database, data analysts with their tools, and data operations. The underlying claim is that a leader accountable for delivering value to marketing, sales, and customers needs the data, the analytic database, the people, the tools, and the delivery process inside that span rather than dependent on another function.

Why does DataOps focus on process rather than on the analytics themselves?

Because what you do matters much less than how you do it. The what is the model, the algorithm, the pipeline, the visualization, the governance, and the data; the how is development, deployment, monitoring, iterating, collaborating, and measuring. Two outside arguments back it: Elon Musk on the greatest potential being in building the machine that makes the machine, and Deming's finding that 94 percent of causes are common cause, which means looking for a person to blame instead of fixing the process misses almost every time.

Where to go next