Burn Rate Monthly Forecast
About This MicroSim
A combination chart showing daily LLM spend (blue bars) up to "today," with a solid green cumulative line over the actual data and a dashed green forecast line extrapolating to day 30. The red horizontal budget line is the reference; the status banner says ON TRACK or OVER BUDGET based on the forecast intersection.
How to Use
- Read the default state (today=18, burn=1.0, budget=$30K). Note the status banner.
- Slide "today" earlier to day 5. Forecast becomes much noisier — small changes in early-month spend project huge differences by month-end.
- Bump "burn multiplier" to 1.5. The dashed line steepens. Watch the status flip to OVER BUDGET.
- Toggle seasonality. Forecast bends upward on weekdays (×1.3) and dips on weekends (×0.6).
- Adjust budget. Lower budgets fire OVER BUDGET earlier in the month — useful for setting alerts.
Bloom Level
Apply (L3) — calculate a monthly cost forecast from a partial-month burn rate.
Iframe Embed Code
1 | |
Lesson Plan
Audience
Engineering managers and platform-team members responsible for monthly LLM budget reporting.
Duration
10–15 minutes inside Chapter 3.
Prerequisites
Chapter 3 sections on Monthly Token Spend, Forecasting Token Cost, and Burn Rate.
Activities
- Calibrate the default (3 min). Read the status. Note the math: cumulative-to-date / today = avg daily burn → × 30 = projected end-of-month.
- Early-month volatility (5 min). Slide today to day 3, then day 18. Note how the forecast confidence improves with more data.
- Seasonality lesson (5 min). Toggle seasonality. Note that a flat-burn forecast over-predicts spend in seasonal workloads when the projection date lands on weekends.
- Practice scenarios.
Practice Scenarios
| # | Today | Burn × | Seasonality | Budget | Status |
|---|---|---|---|---|---|
| 1 | 18 | 1.0 | off | 30K | ? |
| 2 | 18 | 1.5 | off | 30K | ? |
| 3 | 18 | 1.0 | on | 30K | ? |
| 4 | 5 | 1.0 | off | 30K | ? |
| 5 | 28 | 1.0 | off | 30K | ? |
Assessment
Learner has met the objective when, given a partial-month spend total and a target budget, they can compute the projected end-of-month spend and identify whether the team is on or over budget.
References
- FinOps Foundation — Forecasting cloud spend — adjacent literature on partial-month projection.
- Anthropic Documentation — Usage and billing dashboard.
- Chapter 3 of this textbook — Monthly Token Spend.
Senior Instructional Designer Quality Review
Reviewer perspective: 15+ years designing engineering and FinOps curricula for adult professional learners.
Overall verdict
Approve as-is for Chapter 3. Score: 88/100 (B+). A clean Apply-level (L3) calculator with the right primary visualization (linear extrapolation) and a proper status banner that translates math to action.
What works
- Bloom alignment correct. L3 "calculate" is exactly what this is.
- Seasonality toggle is the right pedagogical bend. Naive linear forecasting over-predicts seasonal workloads; the toggle makes that visible.
- The status banner is the load-bearing UX element. Without it, the chart is just data; with it, the chart is a decision.
- Budget slider, not a fixed value. Lets the learner explore "what budget would I need?" rather than just "did I exceed mine?".
Gaps
- Linear extrapolation is too simple. Real teams use moving-average or trailing-window forecasts. A toggle for "use last 7 days only" would teach a more honest method. Score impact: −3.
- No alerting threshold. A real burn-rate alert fires before end-of-month, not at it. A "alert when projected > budget × 1.1" line would teach the operations side. Score impact: −2.
- No way to see prior months. A "compare with last month" overlay would teach trend analysis. Score impact: −2.
Accessibility
Color-blind safe (blue/green/red verdict text in status). Slider labels show numeric values. Chart legend is at bottom for clarity.
Cognitive load
4 controls. Tractable. Status banner provides natural-language summary.
Recommendation
Approve. Open follow-up for moving-average forecast option (gap 1).