AI Blocks

Putting the blocks together

This is taking longer than expected.

Try againBrowse the library
AI Blocks

AI Blocks › Track your numbers

Marketing budget scenario planner

Give it past spend and results by channel. Get low and high forecasts for a new budget, with diminishing returns built in.

Marketing budget scenario planner is a free AI skill for the job "Track your numbers". Give the instructions to an AI assistant that supports the required files and tools, then review its output.

Free · No account · MIT licence · Made by wizard7558 · Level: expert · Thinking: high

Start this task

Download the complete instructions and supporting files. Add the material the task needs in an AI that supports those files and tools.

Download all files

Made by wizard7558. See the original on GitHub; we copied it as it is, with its licence.

What it does

Low and high forecasts for a new marketing budget, based on your past spend and results, with diminishing returns built in. The model does not assume that doubling spend doubles results. You give weekly or monthly spend by channel and the outcomes at each funnel stage. A saturation curve is fitted to each channel. Missing spend data is treated as a gap, not as zero. For a new budget it projects outcomes, then runs many random simulations to give a cautious p10 and an optimistic p90. If you ask it to allocate money, it uses proper optimisation rules so it does not pile everything into one channel. It also backtests on past periods.

You give it

You get

Use it when

More for Track your numbers

Every track your numbers skill

Full instructions
  • Add it to your AI once: download it and add the folder to Claude, ChatGPT or Codex as a skill.
  • Ask for the job in your own words. Your AI opens the skill when it fits.
  • It works best when your AI can run code and read your spreadsheet.
Full instruction file

Budget Scenario Planning

Use this skill when analyzing marketing budgets, forecasting outcomes, fitting Hill-saturation curves, projecting funnel performance using Monte Carlo sampling, or backtesting models against historical scenarios.

This skill provides a mathematically rigorous approach to marketing budget allocation, replacing flat multiplier forecasting with methods that account for diminishing returns and variance.

Core concepts & workflow

  1. Calibrate: Ingest historical weekly/monthly spend by channel alongside funnel outcomes.
  2. Curve-fit: Fit a Hill-saturation curve f_c(x) = a_c * x / (b_c + x) for each channel using the a_c = 4 * μ_c * β_c heuristic to cap the upside of low-spend, high-return channels.
  3. Project: Given a total budget and timeframe, project the expected outcome across funnel stages.
  4. Sample: Run Monte Carlo simulations to generate p10 (conservative) and p90 (optimistic) bands.
  5. Backtest: Compare projected scenarios against historical hold-out periods to measure model error (e.g., RMSE, MAPE).

Mandatory constraints (Contract Boundaries)

  • Unknown ≠ Zero: If historical spend data is missing for a period, treat it as a missing data point (gap), not as $0 spend.
  • No allocator hallucination: When asked to "optimize" or "allocate" a budget, you MUST use the Karush-Kuhn-Tucker (KKT) conditions on the Hill curves. DO NOT use a naive linear programming solver that dumps 100% of the budget into the single highest ROI channel.
  • Strict framing: Projections are observational estimates based on historical correlation. Explicitly communicate to the user that these are forecasts, NOT causal incrementality guarantees.
  • Date alignment: Funnel projection periods must cleanly map to the historical fit window.

Using the included artifacts

  • Consult references/planning-math.md for exact formulas (Hill calibration, KKT allocation, Monte Carlo sampling).
  • Use scripts/run-budget-scenario.py to project budgets deterministically.
  • Use references/scenario-fixtures.json to mock inputs or test the script.
  • Check references/eval.md and references/eval-cases.json for validation criteria and evaluation prompts.