Fixing Forecasting: The Metrics That Made the Difference at Catalyst

At Catalyst Therapeutics, we knew our forecasting was broken. Constant shortages, excess inventory, and endless adjustments weren’t just frustrating—they were costly. When we shifted from reactive, sales-based forecasting to a structured, install-base-driven approach, we didn’t just hope for improvement—we measured it.

Here’s how we tracked our success using key forecasting metrics.

1. Forecast Value Added (FVA): Proving Our Process Worked

(Reference: "Forecast Value Added Analysis: Step-by-Step" by SAS Institute.) (sas.com)

According to SAS Institute, "The key to improving forecasting performance is not just adding complexity but eliminating waste in the forecasting process. Forecast Value Added (FVA) analysis helps identify where adjustments actually improve accuracy and where they do not." We needed to know if our new approach was actually helping. Forecast Value Added (FVA) became our guiding metric, comparing our new install-base-driven forecast to our old sales-driven method and a naive baseline.

What we saw:

• Before: Constant manual overrides, but no real improvement in accuracy.
• After: FVA showed positive gains, proving our new approach was consistently better.

If your forecast isn’t adding value, why keep doing it the same way?

2. Forecast Bias: Eliminating Overcorrection

(Reference: "Measuring Forecast Accuracy: The Complete Guide" by RELEX Solutions.) (relexsolutions.com)

As RELEX Solutions states, "Consistently biased forecasts lead to excess costs and poor service levels. Companies that monitor and correct forecast bias see significant improvements in inventory optimization and customer satisfaction." Before our transformation, we were constantly reacting—either overestimating demand, leading to bloated inventory, or underestimating it, causing stockouts and emergency shipments.

By shifting to an install-base approach, bias dropped significantly. No more overcorrections. No more second-guessing.

3. Mean Absolute Percentage Error (MAPE): Seeing the Accuracy Gains

(Reference: "MAPE, WMAPE, and Forecast Bias" by Demand Planning Net.) (demandplanning.net)

Demand Planning Net highlights, "MAPE remains one of the most widely used metrics for forecast accuracy, but its effectiveness depends on the variability of demand. Companies must balance MAPE with other KPIs like weighted MAPE and forecast bias for a full accuracy assessment." Forecast accuracy is more than just gut feel—it’s measurable. We tracked MAPE to see how much our forecast was deviating from actual demand.

The result?

• Our previous method had high error rates due to reacting to noisy sales trends.
• With install-base forecasting, MAPE improved dramatically, showing a stable, predictable process.

Forecast accuracy means nothing if it doesn’t drive better business decisions. Our ultimate goal was right-sized inventory—enough to meet demand without excess waste.

Tracking inventory turns, we saw:

• Higher turns = No more unnecessary stock tying up cash.
• Fewer stockouts = Service levels improved without firefighting.

The Futility of Arbitrary Forecast Accuracy Targets

(Reference: "Overcoming Five Major Challenges in Biopharma Forecasting" by The Dedham Group.) (dedhamgroup.com)

The Dedham Group notes, "In biopharma, rigid forecast accuracy targets create a false sense of control. Instead, companies should focus on continuous improvement and reducing variability in key demand drivers to improve planning outcomes."

One of the biggest pitfalls in forecasting is setting rigid accuracy targets—like demanding 90% accuracy across all forecasts. These targets often create more problems than they solve:

? They ignore the reality that different products have different demand patterns.

? They incentivize teams to manipulate numbers rather than improve the process.

? They discourage focus on metrics that actually drive better decisions, like FVA and bias reduction.

At Catalyst, we learned that a forecast is only as good as its ability to support better supply chain decisions. Instead of chasing arbitrary accuracy goals, we focused on reducing waste, improving alignment, and making forecasting a value-added process.

The Impact of Artificial Volatility in Biotech Forecasting

(Reference: "Lab to Patient: Overcoming 5 Key Challenges in Life Sciences Supply Chains" by Oracle.) (blogs.oracle.com)

Artificial volatility in the biotech industry often stems from frequent manual forecast adjustments, overreactions to short-term sales fluctuations, and misaligned incentives across departments. These internal behaviors amplify demand variability, leading to the bullwhip effect—where small changes in demand cause significant fluctuations upstream in the supply chain.

According to Oracle, "Biopharma companies often struggle with demand planning because internal stakeholders introduce artificial demand volatility. Addressing this requires cross-functional collaboration and process discipline."

The impact of artificial volatility includes:

• Inventory Imbalances: Stockouts and overstock situations become more frequent, leading to increased holding costs or lost sales opportunities.
• Operational Inefficiencies: Erratic production schedules and procurement plans increase overall supply chain costs.

The Value of Forecast Accuracy Improvements (Reference: "What’s the Business Impact of 10% Extra Forecast Accuracy?" by Nicolas Vandeput.) (nicolas-vandeput.medium.com)

A study found that a 10% improvement in forecast accuracy can lead to:

• 4% reduction in inventory levels
• 6% decrease in shortages

By addressing artificial volatility and improving forecast accuracy, Catalyst Therapeutics achieved more stable operations, better customer service, and improved financial performance.

The Bottom Line: A Forecasting Process That Actually Works

By measuring our transformation with these key metrics, we turned forecasting from an unreliable guessing game into a strategic advantage:

? FVA proved our new method was better.

? Bias was eliminated, leading to consistent, reliable forecasts.

? MAPE improved, confirming increased accuracy.

Want to improve forecasting at your company? ?? Read Lean Forecasting Demystified to learn how to take control of your demand planning.

?? Stop reacting. Start leading.

Let’s Discuss! ?? Have you measured FVA or bias in your forecasting process? Drop a comment below—let’s talk!

#SupplyChain #Forecasting #FVA #ForecastBias #DemandPlanning #LeanForecasting #OperationalExcellence