User Guide & Methodology

It is informed by the Phillips ROI formula and is compatible with planning for Kirkpatrick Level 3 behavior and Level 4 results. Importantly, this is a forecast planning model — it estimates ROI before a program runs; it does not observe evidence or measure delivered results.

The defining move is credibility adjustment, applied to each benefit individually:

• Population — only the learners who reach the right stage (attended → completed → applied) count toward a behaviour-change benefit.
• Isolation — the share of the outcome the training truly caused.
• Confidence — how sure you are of the estimate.
• Cashability — how much of the value becomes usable money (freed-up time isn't automatically cash).

The result is shown as a conservative–expected–optimistic range. It remains a planning range, not a statistical confidence interval.

Your data stays in this browser. Nothing you type is uploaded by the calculator. Inputs are auto-saved in the browser's local storage so they survive refreshes and closed tabs; use Reset to clear the saved local inputs.

1. Program name and how it's delivered (classroom, virtual, eLearning, or blended). For instructor-led: learners per cohort and cohorts per year. For eLearning: total annual enrollment.
2. How many years the program (and its benefits) will realistically last.
3. Average salary and benefits load of the employees being trained — used to price their time.
4. Build costs (one-time): analysis, instructional design, development, media, technology, and evaluation.
5. Delivery & running costs: facilitator, facilities, materials, hosting, administration, maintenance — plus, optionally, the itemized costs training cases usually miss (project management, accessibility, learner support, AI licenses, and more).
6. The outcomes you expect to move — fewer errors, faster ramp-up, lower turnover, more sales, etc. — with a rough size for each, and a sense of who realizes each one (everyone trained, only those who apply it, or an independent outcome count).
7. Optional but powerful: your participation rates (how many start, complete, and apply the training) and, per benefit, its isolation, confidence and cashability.

STEP 1 Scenario

Describe the program and the workforce it affects.

• Delivery modality comes first — it reshapes the cost inputs on the next tab. Pick how the program is delivered (see the four options below).
• Scale — for instructor-led modes this is learners per cohort × cohorts per year; for eLearning it's a single total annual enrollment instead.
• Lifespan is how many years the program and its benefits realistically last. Discount rate is used for Net Present Value; use your organization's finance-approved hurdle or discount rate rather than treating the example value as a benchmark.
• Burdened labor cost — salary × (1 + benefits %) ÷ working hours — is the fully-loaded hourly cost of an employee. The tool computes it and uses it to price training time and productivity gains.

The four delivery modalities — each changes which delivery costs and time figures apply:

STEP 2 Costs

Enter every cost, so the investment side is fully loaded (conservative):

• One-time build — needs analysis, instructional design, SME time, development, graphics, video, translation, technology setup, and evaluation & measurement. Counted once, up front — a one-time investment at launch (not spread across the years).
• Delivery (recurring) — this section changes with the modality you chose in Step 1. Instructor-led shows facilitator, facility, A/V, materials, travel, and hosting per cohort; eLearning replaces those with per-seat hosting. Fields that don't apply are hidden automatically.
• Maintenance — enter a direct annual amount (recommended), or estimate it as a % of your authored content only. Add a major content refresh ($ every N years) if you plan one.
• Administration / coordination — the annual overhead of running the program.

Advanced / itemized costs (optional) — expand this to add the costs training business-cases most often miss: project/program management, pilot & QA, accessibility, change management, fixed platform licenses, learner support, ongoing evaluation, reinforcement/coaching, backfill/overtime/temporary coverage, vendor delivery, facilitator prep & travel, assessment/certification, and a full set of AI-related costs. Each field is tagged with when it applies (one-time, annual, per cohort, or per learner) and what kind of cost it is (cash, opportunity, or overhead).

Advanced participation funnel (optional) — set the share of enrolled learners who start / attend, complete, and apply the training. These default to 100%. Lowering them scales the benefits you tie to each stage (Step 3) and prices non-completer time honestly.

STEP 3 Benefits

Switch on only the outcomes this program will genuinely move, and size each one. Options include error/rework reduction, productivity gains, faster new-hire ramp-up, retention, safety & compliance, reduced supervision, sales lift, customer retention, and support-ticket reduction.

Every benefit now carries its own credibility settings, so you're not forced to apply one blanket assumption to unrelated outcomes:

• Counts toward — which population realizes it: Enrolled, Started/Attended, Completed, Applied on the job, a Custom count you've already scoped, or an Independent outcome (like incidents avoided) that the funnel should not discount. Behaviour-change benefits default to Applied; outcome counts default to Independent.
• Attribution, confidence & cashability (a small panel on each benefit) — set that benefit's own isolation %, confidence %, a cashability class and realized %, and a source and owner for the record. Leave isolation/confidence blank to inherit the defaults from Step 5.

STEP 4 AI impact (optional)

Model AI on both sides of the ledger — honestly:

• Upside — lower build cost from AI-accelerated development, plus productivity from AI performance support and AI-adoption training. The development saving only reduces the AI-eligible authoring labor (instructional design, development, graphics, video) — never needs analysis, translation, technology setup or evaluation.
• Downside — if you claim AI benefits, add the AI-related costs in Step 2's advanced section (licenses, usage, integration, governance/security, human review, error correction/rework, monitoring, vendor implementation). The tool will flag a one-sided AI case that shows only the upside.

Leave all of this off for a classic, no-AI ROI case.

STEP 5 Timing, persistence & scenario

These settings shape how the forecast plays out over time and how wide the range is:

• Default isolation & confidence — the starting values every benefit inherits unless you override it in its own panel (Step 3).
• First-year realization and first-year cost ramp — how much of the benefits and of the recurring costs actually land in year one (a program rolling out across the year rarely hits 100% of either on day one).
• Benefit persistence after Year 1 — how much of each year's benefit carries into the next. Defaults to 100% (no decay); lower it, with a stated rationale, if skills fade without reinforcement. This only affects the multi-year projection — never the Year-1 number.
• Scenario swings — how benefit magnitude, application, isolation and volume-driven costs could move. The report turns these into a Conservative–Expected–Optimistic planning range. Confidence stays fixed because it is already captured per benefit.

STEP 6 Report

A clean, printable summary built for a finance audience:

• Year-1 ROI as the headline (the most defensible single number), with the multi-year range beneath it.
• "How each benefit is built" — the full equation for every benefit (raw × isolation × confidence × population × cashability = adjusted), in plain language, with each one's source and owner.
• Cost by type — cash outlay vs. opportunity (time) vs. allocated overhead, so nobody mistakes freed-up time for cash.
• Cost and benefit breakdowns, cumulative net economic value, your key assumptions, an evidence-readiness register, plausibility flags, and potential intangibles to investigate.

Use Print / Save PDF to share it.

• Save downloads a small file with all your inputs. Load reads it back later — perfect for revisiting or comparing scenarios.
• Your work is also auto-saved in this browser, so a refresh won't lose it.
• Report → Print / Save PDF produces a branded, paginated report for stakeholders.
• Reset clears everything; Example reloads the built-in sample.