From Budget Spreadsheet to Rolling Forecast: How Canlak Coatings Took Microsoft Fabric Plan into Production

Success Story

From Budget Spreadsheet to Rolling Forecast: How Canlak Coatings Took Microsoft Fabric Plan into Production

under

4 weeks

kickoff to production

Plan and actuals on

1 Model

Reporting and planning on the

Same Capacity

Rolling Forecast

Refreshed monthly not annually

Canlak Coatings runs a product-level rolling sales forecast in production on Microsoft Fabric Plan, live in under four weeks.

Canlak had already made Microsoft Fabric the daily foundation of its business, with reporting running on a governed Power BI semantic model for close to two years. That head start is what made the next move quick. The company was adding a capability to a platform it already trusted, on data it had already governed.

Canlak Coatings is a North American manufacturer of wood coatings, serving customer across multiple markets. Its first Fabric story described how the company created an a alytics foundation across several ERP systems. This is the story of what that foundation made possible next.

The forecast a spreadsheet could support, but only so far

Canlak's planning had been a periodic budget exercise, and sales forecasting itself sat where it sits for most mid-market manufacturers, in a spreadsheet. The approach worked at the sales rep and customer level and served the business for a long time. Taking the same forecast down to the product level was a different matter. Every step toward more detail multiplied the spreadsheet: more tabs, more links, and more manual reconciliation against actuals that lived in a separate system.

The process carried the same weight. A draft went out by email, edits came back, and a new version went out again, with file names standing in for a single source of truth.

Meanwhile, the business already ran on daily, product-level actuals in Fabric. The forecast sat somewhere else, at a coarser grain and on a slower rhythm. Closing that gap, and moving from a periodic budget to a rolling forecast, was the purpose of the project.

Why Canlak chose to plan where its data already lived

When Microsoft introduced Fabric Plan, the planning capability in Fabric IQ, the design answered Canlak's question directly. Canlak already had an enterprise data warehouse on OneLake, and the forecast could draw on that existing investment rather than start beside it. Planning sheets sit natively on Power BI semantic models, so the forecast could sit on the same governed model Canlak already trusted for daily reporting. Same definitions, same dimensions, same security. No separate planning database to keep in step, and no external modelling engine to reconcile.

The security point carries real weight for a finance team. Canlak's semantic model runs in import mode, and the row-level security already defined there is inherited and applied directly in the planning sheet. The rules that govern who sees what in reporting govern the forecast as well, without being recreated or maintained twice.

For an organisation already working in Fabric every day, adopting Fabric Plan during preview was a measured decision rather than a leap. The data foundation was proven and the semantic model was mature. The open question was the planning layer itself, and the quickest way to answer it was to put it into practice.

Under four weeks from kickoff to a working forecast

Under Director of Finance and Treasury Rich Zielinski, Canlak stood up the planning model in under four weeks. Because the data, the model, and the security were already in place, applying Fabric Plan was largely a matter of configuration rather than a new platform to establish. The design reflects how the sales organisation actually thinks:

Organised by sales rep, cascading through the hierarchy. Segment, sales rep, customer, product. A figure entered at any level splashes up or down the hierarchy, so a product-level change flows into the customer and rep totals, and a total entered higher up distributes across the products beneath it.
Volume driven. The forecast is entered as volume, and the Plan engine calculates revenue as volume multiplied by the latest price.
Adjustable inputs, protected actuals. Both volume and price can be adjusted as the forecast evolves. Plan data writes back to a SQL database within Fabric, so the forecast never leaves the environment where the data already lives, and actuals remain governed and read only.
Security carried across. Row-level security from the semantic model applies in the planning sheet, so each sales rep works within the same boundaries that already apply in reporting.

The model spans five segments, two currencies, and more than forty sales reps, and it runs in the Canada Central region. Working in preview meant the occasional rough edge, as expected. One gap in how a feature filtered data was raised with the product team, and a fix is on its way to a future release. None of it held up the forecast, which went into production ahead of general availability.

Type once. Every level follows.

at product-level detail, this reconciliation is what a spreadsheet could not hold

A Rolling Forecast in Production, on the Same Capacity

Across finance, users enter forecast figures directly in planning sheets. The forecast cycle runs monthly, and full-year visibility is maintained as each cycle closes. What was a periodic budget exercise is now a rolling forecast that moves with the business.

“Moving from Excel spreadsheets to Fabric Plan transformed our forecasting process. What used to take three to four weeks of spreadsheets, emails, meetings and reconciliations can now be completed in about a week. By bringing planning and reporting into a single platform, we’ve simplified the process into a rolling forecast, improved visibility and given our team more time to focus on analysis rather than administration.”

Rich Zielinski,

Director of Finance and Treasury, Canlak Coatings

The efficiency of the approach shows in the capacity numbers. Canlak’s whole Fabric environment, the reporting the business depends on daily and now the planning workload, runs on the same capacity the business was already using, and adding Fabric Plan has produced no noticeable increase in consumption since the forecast went live. Enterprise planning, at a level of detail a spreadsheet could not hold, across forty sales reps and two currencies, without a step change in platform cost. How the foundation was designed is what keeps the cost flat as more runs on it.

Canlak’s first case study made the point that questions about whether Fabric is production ready were, in the company’s experience, largely theoretical. With Fabric Plan, Canlak makes the same point again, this time earlier in the product’s life.

What Canlak gained

A rolling sales forecast in production, live ahead of general availability
Forecast granularity down to the product level, cascading from segment to sales rep to customer to product
Plan and actuals on one governed semantic model, with no reconciliation between systems
Reporting security inherited by the planning layer, with no duplicate rules to maintain
Full-year forecast visibility refreshed on a monthly cycle
A working planning model in under four weeks, with no noticeable increase in capacity consumption

Why it worked

Four things carried the project:

A proven foundation first. Almost two years of Fabric in production meant the semantic model, governance, and team habits were already in place. Planning was an addition, not a change of course.

Ownership close to the work. Hands-on testing and iteration from Canlak’s team, led by Rich Zielinski, kept the model honest against how sales and finance operate.

A design that matches the business. Volume driven, price aware, and organised by sales rep. The forecast mirrors how Canlak sells.

Preview adoption as a strategy. Standing the model up during preview, with any rough edges raised as they appeared, turned launch day into a quiet one.

Technologies Used

What's next

Throughout, Canlak partnered with Data Crafters to design and implement the planning model, extending an analytics environment that has supported the business in production for close to two years.

With the sales forecast in production, the same pattern is available for whatever Canlak plans next. The natural direction is deeper financial planning on the same governed model, extending the forecast beyond sales into the wider finance function. Each addition sits on the model that already exists, which is what has carried the company from daily reporting, to daily operational insight, and now to forward-looking planning, all on one platform.

Early adoption, on a proven foundation

Canlak’s experience shows what early, committed adoption looks like when it rests on a working foundation rather than enthusiasm alone. By the time Microsoft Fabric Plan reached general availability, Canlak was already running its sales forecast on it in production.

There is a wider lesson here for finance and data teams whose forecasts stop short of the detail the business needs.
The distance between a governed reporting platform and a governed planning process is shorter than it looks. When the data foundation, the semantic model, and the security are already in place, planning is the next step rather than the next project. The forecast is no longer a set of rules living in a spreadsheet and in people’s heads. It is encoded on the same model that runs the business, which is what makes it dependable today and extendable tomorrow.