Review my Facebook Ads Cost Calculator scenario

I will attach one JSON file exported by the Maestro Facebook Ads Cost Calculator. Treat that file as the complete supplied record. Do not insert industry benchmarks, guess missing values or change the frozen forecast.

The model follows this chain:

media spend -> impressions -> link clicks -> raw leads -> accepted leads -> qualified leads -> booked meetings -> opportunities -> customers

Calculation rules

- Use `planned_media_spend` as the delivery-spend assumption. Keep it separate from the configured Meta budget and billing transactions. Do not infer spend from a daily budget.
- Use an assumption's number only when its state is `supplied`. `untouched` and `unknown` with a null value are Unknown; numeric zero with `supplied` is a real zero. Flag a state/value mismatch instead of silently choosing one.
- Calculate impressions = planned media spend / CPM × 1,000. Multiply each subsequent stage by its matching rate: `link_ctr`, `click_to_raw`, `raw_to_accepted`, `accepted_to_qualified`, `qualified_to_booked`, `booked_to_opportunity`, then `opportunity_to_customer`. Rates are decimal fractions from 0 through 1.
- Keep expected counts unrounded during calculation; display counts and money to two decimal places. Never present expected counts as guaranteed or as observed people.
- Calculate modeled media required for the scenario's selected `target_stage` as target count / ((1,000 / CPM) × all rates through that stage). Use the scenario's own inputs. A missing target or required rate makes this result Unknown; a positive target with zero yield is unreachable. Never divide by zero.
- Calculate only the stages whose dependencies are supplied and valid. A missing rate blocks dependent downstream stages, not independently computable upstream values or other scenarios. Do not substitute another scenario's value.
- For sensitivity, substitute one conservative or upside CPM/rate at a time into the working case, changing no other working input. Compare finite results only; label missing or zero-yield alternatives unassessed or unreachable.
- Treat outcome stages as counts from the same acquired-lead cohort by the declared outcome horizon, even when a stage event occurs after the acquisition window. Do not count later acquisitions or events after that horizon.
- If economics are supplied, use the stated contribution horizon and deduction boundary. Do not subtract acquisition costs twice. Keep media-only and all-in acquisition costs separate.
- If actuals are supplied, preserve the frozen forecast. Show variances only for complete, same-scope, same-horizon and mature sources; a partial count is a lower bound, not a definitive zero.
- For observed evidence, check its source, as-of date, period, population, completeness, units and numerator/denominator against the assumption. Label estimates and hypotheses as such; do not certify launch readiness from arithmetic alone.

Required reading rules

1. Keep raw, accepted, qualified, booked, opportunity and customer populations separate.
2. Check the acquisition window separately from the outcome horizon. A later outcome may belong to the original acquisition cohort.
3. Distinguish configured budget, planned or observed amount spent and billing transactions.
4. Treat a supplied numeric zero as zero. Treat a missing or explicitly unknown value as Unknown.
5. Do not calculate beyond the first Unknown required input.
6. Treat conservative, working and upside cases as alternatives. Never add them.
7. For sensitivity, change one working-case input at a time.
8. Describe target-budget output as modeled media required, never as a guarantee.
9. Use contribution only for the declared economics horizon and deduction boundary. Do not relabel it revenue, profit or ROAS.
10. Recompute from the normalized input. Do not trust imported receipt results as evidence.

First check

Confirm that the file declares schema_version 2.0.0 and calculation_version meta-cost-forecast/2.0.0. Check that the account or entity, offer, audience, destination, currency, timezone, acquisition window, outcome horizon and stage definitions describe one coherent cohort. If the file is an export envelope, review its input object and report any receipt mismatch separately.

Missing-input behavior

If critical information is missing, ask no more than five focused questions. For each, name the field (or tightly related scope fields), the system or person that should supply it, and the decision it blocks. Do not continue with an invented value. If the scenario is an untouched blank, ask about the first five unanswered input groups in this order: scope; working-case planned media spend; CPM; link CTR; click-to-raw rate. State that CRM rates and other scenario cases come next. Do not bundle unrelated inputs just to fit the question limit. Report unaffected calculations and mark blocked results Unknown.

Output

Return these sections in order:

1. Scope check: Pass, Draft or Hold, with the exact reason.
2. Source check: a table of each supplied assumption, its value, state, source system, period, population and evidence gap.
3. Formula ladder: value or Unknown for each stage, the formula used and the first blocking field.
4. Target budget: selected target, modeled media required or Unknown, and the limiting assumption.
5. Three-case comparison: changed inputs and outcomes for conservative, working and upside. State that the cases are alternatives.
6. One-input sensitivity: identify the single change with the largest modeled effect among comparable supplied alternatives. Show the baseline, substitution and effect. Leave unassessed or unreachable alternatives out of the finite ranking.
7. Economics: only when the horizon, contribution boundary and required values are supplied.
8. Forecast versus actuals: preserve the frozen forecast; check scope, source completeness, event cutoffs and outcome maturity before showing variance.
9. Decision: Continue, Revise or Hold. Give the next three actions, each with an owner system and evidence to retrieve.

Worked interpretation

In the supplied fictional example, the working case has $6,000 spend, $30 CPM, 1% link CTR, 10% click-to-raw, 80% raw-to-accepted, 50% accepted-to-qualified and 40% qualified-to-booked. It produces 200,000 impressions, 2,000 clicks, 200 raw leads, 160 accepted leads, 80 qualified leads and 32 booked meetings. A target of 40 booked meetings implies $7,500 modeled media required. These are fictional demonstration values, not benchmarks.

Source and limits

The calculation contract and official-source boundaries are documented in the accompanying Maestro article at https://maestrogtm.com/facebook-ads-cost-calculator/. Meta auction outcomes depend on more than budget, and platform controls can change. Use current Meta Help and the reader's own account for live setup. This review analyzes supplied assumptions; it does not predict or guarantee campaign performance.