Case Studies /
Case Studies /
CSEC
CSEC
CSEC · SPORTS & ENTERTAINMENT
CSEC · SPORTS & ENTERTAINMENT
Most data projects deliver a platform and call it a win. We did something harder. We started with a strategy, aligned people, processes, and technology around it, and trained the Flames’ team to run it. Fan data finally earned its keep.
Most data projects deliver a platform and call it a win. We did something harder. We started with a strategy, aligned people, processes, and technology around it, and trained the Flames’ team to run it. Fan data finally earned its keep.
The Flames had the data. We aligned the strategy, the platform, and the people so it could finally
earn its keep.
The Flames had the data. We aligned the strategy, the platform, and the people so it could finally
earn its keep.
HOW WE THINK
HOW WE THINK
One Method. Five Pieces.
One Method. Five Pieces.
Here's what each one meant in the CSEC work.
Here's what each one meant in the CSEC work.
People
We mentored while we built. By the end, CSEC's data analysts and IT team were writing code to our senior architects' standard and redeploying the platform themselves.
Data & Analytics
Eighty-two tables now flow through bronze, silver, and gold with documented lineage. Reports that used to time out return in seconds, at 45% lower cost.
Technology Platforms
We rebuilt the platform on Databricks, governed by Unity Catalog, with access tied to the corporate identity provider and ready for Power BI and Excel from day one.
How decisions get made,
how work moves, how teams learn.
The mechanics of flow.
Ways of Working
We built engineering habits that outlast us: CI/CD with pull request reviews, Terraform that rebuilds the platform the same way every time, and pipelines that recover from failure on their own.
Culture
Data stopped being one team's job. Ten workshops across four departments turned fan, ticketing, and customer data into a shared asset with named owners.
01 · Situation
A first cloud data platform, and a vision for what came next.
Calgary Sports and Entertainment Corporation, the organization behind the Calgary Flames, the Hitmen, the Stampeders, the Roughnecks, and the Wranglers, had stood up its first cloud data environment with a third-party vendor. It was a real foundation, and the team had built on it for years.
What they learned through that experience shaped what they wanted to build next. The business needed to shape shared data, not just consume it: fan data, ticketing data, customer data, all designed alongside the people who'd live with the consequences. Accountability for that data needed to live across IT and the business, owned at every layer, not centralized in a single team. And the internal data team needed to be able to support, extend, and evolve the environment themselves, without depending on outside vendors indefinitely.
That clarity was where the engagement with Domino started. Not "fix what's broken." Closer to: "we know where the next chapter needs to go, help us build it, and help our team learn to own it."
02 · Discovery (Sept – Dec 2024)
A technical assessment, ten workshops, fifty-five opportunities.
We opened with a focused technical assessment of the existing warehouse. Architecture review. Data pipelines. Data integration and security. DevOps practices. The output was six categories of recommendations, sized in weeks of work, and a clear architectural recommendation: migrate to Databricks.
But the warehouse wasn't the whole story. We ran ten investigative workshops across four departments to understand how the business actually used data, where the manual processes lived, and where automation would matter. By the time we were done we had fifty-five concrete opportunities mapped to people, process, technology, and analytics.
Two threads ran through everything we found. The platform was ready to be rebuilt to support what the business wanted to do next, and to give CSEC's team a foundation they could own and extend. And the business needed a way to integrate shared data across the source systems that ran daily operations, all working from a shared definition. The recommendations addressed both at once.
03 · Strategy & SOW (Jan – Feb 2025)
03 · Strategy & SOW (Jan – Feb 2025)
From recommendations to a signed sequence.
The strategy distilled into one sentence: modernize the platform on Databricks, start fan data master data management, and automate the highest-value manual processes, in that order. We didn't recommend the biggest change first. We recommended the move that made every subsequent decision obvious.
The signed Statement of Work landed in February 2025. Three deliverables: rebuild the data platform on Databricks, build initial integration workflows for the priority source systems, and run data governance and validation reports through CSEC's internal data analysts. A small senior team led delivery.
CSEC committed roughly half of their data analysts' time to the engagement so we could mentor while we built. Six-month engagement.
04 · Modernization Execution (March – Sept 2025)
04 · Modernization Execution (March – Sept 2025)
Twelve sprints. Two phases. One platform.
The work split cleanly into two phases. Each shipped on its own timeline. Each carried mentorship as a first-class deliverable, not a bonus.
Phase 01 · ~6 weeks · Build the platform.
Install and configure Databricks in CSEC's existing cloud tenant. Stand up dev, test, and production environments. Synchronize role-based access control with the corporate identity provider. Configure governance through Unity Catalog. Build a SQL Warehouse to support Power BI and Excel consumption. Deploy everything via Terraform Infrastructure-as-Code so the platform could be rebuilt the same way every time. Phase one delivered **100% on-time** at fifty-two story points. While we built, we mentored CSEC's IT infrastructure team on Terraform, most of whom had never used Infrastructure-as-Code before. By the end of the phase, they could redeploy the platform themselves.
Phase 02 · ~5 months · Ingest the source systems.
Four enterprise source systems into the new lakehouse, moved off legacy connections onto modern API-based ingestion, including a custom SDK we built where one system needed it. **Eighty-two tables in total**, all flowing through bronze, silver, and gold layers with documented lineage. We didn't ship the integration in bulk. We shipped it on incremental loads with recoverable pipelines, repair runs that resume from failure points, and audit-ready job history. Two hundred and ninety story points committed, two hundred and eighty-two completed: **97% on-time**. One hundred and thirty-six stories across five epics.
Throughout the integration phase, CSEC's data team rebuilt the existing Power BI reports against the new Databricks lakehouse, every dashboard the business depended on, repointed and validated against the new platform. We mentored on the new architecture and the SQL Warehouse patterns; their analysts did the rebuild work. By the time the old warehouse was retired, the reports the business looked at every day still worked, just faster.
The engagement landed one week past the original estimate, approximately four percent over budget. That's a strong outcome for a six-month modernization with mentorship written into every deliverable.
By the numbers:
12 sprints · 5 epics · 136 stories · 82 tables · 4 source systems · 3 environments ·
1 consumption layer rebuilt in flight.
By the numbers:
12 sprints · 5 epics · 136 stories · 82 tables · 4 source systems · 3 environments ·
1 consumption layer rebuilt in flight.
05 · What compounded
05 · What compounded
Faster queries. Lower cost. A team that runs the platform.
Six months of disciplined delivery produced results that were measurable on day one after the engagement closed.
Query performance compared the old warehouse against Databricks across eight reporting tables. **Every table improved.** The smallest gain was 97.52%, the largest 99.77%. The slowest reporting query, the full attendance table, dropped from **542 seconds to 1.19 seconds**. None of the eight tables landed below the 97% improvement threshold. Reports that used to time out before a meeting now load while you're saying their name.
The cost story compounded alongside the performance story. By August 2025, the full Databricks environment, dev, test, and production combined, was running well below what the old warehouse alone had cost before the migration, while delivering better performance: a **45% reduction in cloud cost** across the whole environment. The Databricks costs themselves landed within **seven percent** of the SOW estimate. Disciplined budgeting on a moving infrastructure target.
82
tables integrated
45% cloud
cost reduction
97% query performance improvement
4% variance to
original budget
