Budget Reallocation Proposal

Variables

Replace each placeholder before you run the prompt.

• {{portfolio_data}}Per-account: name, monthly budget, MTD spend, CPL, conversion volume, ROAS or close-rate.

  See attached CSV — 12 accounts, total $145K/mo budget.

• {{constraints}}Hard constraints — total budget cap, min spend per account, accounts that cannot be cut.

  Total budget cannot increase. Account "Industrial-A" must keep $25K/mo for relationship reasons. No account below $5K/mo.

Propose a budget reallocation across the portfolio below. Optimize for total return (revenue per ad dollar) within the stated constraints.

PORTFOLIO:
{{portfolio_data}}

CONSTRAINTS:
{{constraints}}

OUTPUT:

1. A reallocation table: per-account, current budget → proposed budget → delta (in $ and %).
2. Per-account justification (2–3 sentences each): why are you cutting / increasing / holding this account?
3. A top-line summary: total expected change in conversions / revenue if proposed allocation runs for 30 days.
4. Risk callouts: which accounts in the proposal have weak data (low conversion volume, recent strategy change, etc.) and need a smaller move.

Do NOT cut any account by more than 30% in one move unless the data is overwhelming.
Do NOT increase any account by more than 50% in one move — Smart Bidding can't absorb that without going off the rails.
Stay within constraints exactly. If constraints make optimal allocation impossible, flag it explicitly.

Expected output shape

A reallocation table, per-account justifications, top-line projection, and risk callouts.

Why we wrote it

Budget reallocation is one of the highest-leverage things an agency lead can do, and most teams do it by intuition. This prompt forces explicit reasoning and risk gating.

How to use

1. Open Claude or ChatGPT. The recommended model for this prompt is claude-opus-4-7 — opus when the prompt requires deep reasoning, sonnet for the rest.
2. Replace every {{variable}} with content specific to your account. The examples above are starting points, not templates to ship as-is.
3. Paste the prompt and run.
4. Read the output against the expected shape above. If the model produced a structurally different response, re-prompt rather than accept the drift.