"This training program cost us $50,000. What did we get back?"
You'll hear that question eventually. And the person asking it isn't looking for "the feedback was really positive"—they want a number.
Training ROI is the capstone of Level 4 evaluation. But most L&D professionals avoid ROI calculations because they seem complicated, imprecise, or politically risky. The truth is, the math itself isn't hard. What's hard is collecting the right data and making defensible attribution claims. This article breaks down the full process.
The Phillips ROI Formula
Training ROI comes from the Phillips model—an extension of Kirkpatrick's four levels, often called "Level 5." The formula is straightforward:
Where:
- Net Benefit = Training Benefits − Training Cost
- Training Cost = Direct Costs + Indirect Costs
- Training Benefits = Monetary value of business metric improvement attributable to training
Simple formula. Each component has gotchas. Let's break them down one at a time.
Step 1: Account for the Full Training Cost
Most people calculate training cost as facilitator fees plus venue rental. The real cost is bigger.
Direct costs:
- Facilitator fees or external program purchase
- Venue (including catering, refreshments)
- Materials and participant workbooks
- Assessment tool costs (survey platform, testing software)
Indirect costs:
- Participant time away from work (daily salary × headcount × days)
- Program administration labor (planning + on-site + evaluation)
- Travel expenses (if applicable)
The indirect costs are what people forget. A two-day workshop with 20 participants earning an average of $400/day means $16,000 in lost productivity alone—often more than the facilitator fee. Leave indirect costs out, and your ROI is inflated beyond credibility.
One cost people forget: Participant time isn't just salary. It's also opportunity cost—the value they would have produced if they'd been working instead of training. For revenue-generating roles like sales, this is estimable: average daily sales × days in training. It's not pocket change.
Step 2: Estimate Training Benefits
This is the hardest step and the heart of ROI calculation.
Training Benefit = Improvement in business metric × Monetary value per unit of improvement. But how do you know the improvement was caused by training? Three methods, in order of reliability:
Method 1: Trend Extrapolation
Compare business metrics before and after training. If monthly sales averaged $500K in the three months before training and $600K in the three months after, the monthly gain is $100K—$300K over three months.
The problem: How much of that $100K gain is attributable to training? Maybe the market improved. Maybe a new product launched. Maybe the sales team got a better CRM. You can't tell from trend data alone.
Method 2: Management Estimation
Ask the business manager to estimate what percentage of the improvement came from training. The manager says "I think about 50% is the training." So training benefit = $300K × 50% = $150K.
The problem: Subjective. A manager who likes training overestimates; one who doesn't trust L&D underestimates. But it's better than claiming 100%.
Method 3: Comparison Group (Most Reliable)
Find a group that attended training (treatment) and a group that didn't (control), with similar baseline metrics. After training, compare the difference between the two groups. The gap is the training effect.
This is the most scientific method because it controls for other variables. The comparison group design article in this track covers the methodology in detail.
Step 3: Confidence Adjustment
Even with a comparison group, you can't claim 100% certainty that the benefit is entirely from training. Phillips recommends a confidence adjustment—discount the benefit based on how confident you are in the data.
If you estimate $300K in benefits but are only 80% confident, the adjusted benefit = $300K × 80% = $240K.
This feels conservative, but it makes your ROI number defensible. Leadership would rather see a conservative but credible number than an inflated one that collapses under scrutiny.
A Worked Example
Let's walk through a sales skills training program:
- Training cost: Facilitator $10K + venue $3K + materials $1.5K + participant time $12K (20 people × 2 days × $300) + admin $1.5K = $28K
- Benefit estimate: Post-training 3-month average monthly sales for the treatment group increased by $24K compared to pre-training; control group increased by $6K. Net difference: $18K/month × 3 months = $54K
- Confidence: Comparison group data, ~80% confidence. Adjusted benefit = $54K × 80% = $43.2K
- Net benefit: $43.2K − $28K = $15.2K
- ROI: ($15.2K ÷ $28K) × 100 = 54%
ROI of 54% means every dollar invested returned $1.54. That's a number leadership can work with.
Don't chase a precise ROI—chase a defensible one. A ±10% margin with rigorous methodology is worth more than a number that looks exact but has holes in its logic. What leadership challenges is rarely the number itself; it's the reasoning behind it.
When to Calculate ROI
Not every training program warrants an ROI calculation. ROI measurement itself has costs—data collection, comparison group management, analysis and reporting. If the training budget was $5K, spending a week calculating ROI doesn't make sense.
Calculate ROI when:
- The program is a flagship annual initiative ($50K+ investment)
- Training is directly tied to leadership's core KPIs (sales, production, customer service)
- You need to justify continued budget for next year
- It's a new training format being piloted (need evidence before scaling)
For routine small-scale training, L1 + L2 is sufficient. Save your ROI effort for the programs that matter enough to justify it.
🛠️ Support ROI Data Collection in FormLM
The most labor-intensive part of ROI calculation—data collection—can be streamlined with FormLM:
- Use scale fields to collect pre-post capability and behavior data as evidence for benefit estimation
- Use statistics overview to track participation rates, completion rates, and average duration as cost-accounting references
- Use data export to pull Excel data and integrate with business intelligence systems for trend analysis
- Use AI reports to auto-summarize evaluation data into a leadership-ready training impact summary
✅ Key Takeaways
- ROI = (Net Benefit ÷ Training Cost) × 100, from the Phillips model (Kirkpatrick Level 5)
- Account for the full cost—direct + indirect. Participant time is often the biggest line item
- Three benefit estimation methods: trend extrapolation (weakest), management estimation (medium), comparison group (strongest)
- Confidence adjustment makes ROI defensible—conservative beats inflated
- Not every program needs ROI—focus on high-investment, KPI-aligned programs
- Leadership challenges the logic behind the number, not the number itself
