Campaign Budget
by
Daily budget decisions for ecommerce teams: read each campaign on contribution after ad cost, then set the new daily budget, hold it, or pause it.
Budget decisions get made on ROAS because ROAS is the number the platform puts in front of you. It is revenue over spend, and it has no idea what the product costs to make, ship, or take back. So the campaign selling a 62% margin product at 3.1x and the campaign selling a 24% margin product at 3.4x look like the second one is winning, and the budget moves the wrong way. This skill reads each campaign on contribution after ad cost, compares the marginal return of the next dollar across campaigns, and sets a new daily budget, holds it, or pauses it.
What It Does
- Recomputes each campaign on contribution margin rather than on platform ROAS.
- Subtracts returns, shipping and payment fees from the revenue the platform reports.
- Estimates the marginal return of the next dollar, not the average return of the last thousand.
- Checks whether spend has already passed the point where the curve flattens.
- Holds budgets that lack the volume to support a change yet.
- Sets the new daily budget, or holds, or pauses, per campaign.
Use Cases
- Move budget toward the campaign that earns more per dollar rather than per impression.
- Catch the high ROAS campaign selling your thinnest margin product.
- Stop scaling a campaign that has already flattened out.
- Pause spend on a campaign whose contribution is negative once returns are counted.
Example Session
Which campaigns should get more budget tomorrow and which should I cut?
Your best ROAS campaign is your worst contribution campaign.
- Campaign | Spend | ROAS | Gross margin | Contribution | Action
- Prospecting, bestsellers | $1,400 | 3.4x | 24% | -$261 | Pause
- Prospecting, full price | $900 | 3.1x | 62% | $829 | Raise to $1,300
- Retargeting, cart | $420 | 6.8x | 51% | $1,036 | Raise to $560
- Broad, new customer | $650 | 2.2x | 58% | $179 | Hold
The bestsellers campaign returns $3.40 per dollar on products that keep 24 cents of each of those dollars, and the returns rate on that set finishes the job. The full price campaign is the one to fund. Retargeting earns most per dollar and gets the smaller raise, because its audience is finite and pushing past $560 buys impressions you already own.
Why This Beats Prompting It Yourself
ROAS vs Contribution After Ad Cost: The Hidden Risk
A generic prompt reads the platform dashboard, where ROAS is the headline and margin does not appear. It recommends scaling the 3.4x campaign, which is the correct answer to the question the dashboard asks and the wrong answer to the question you asked. The budget moves toward your thinnest products and the monthly number gets worse while every campaign metric improves.
When You Still Need The Marginal Dollar
Average return tells you what the last thousand dollars did. It does not tell you what the next hundred will do. This skill prices the marginal dollar per campaign, which is why retargeting at 6.8x gets a $140 raise and not a doubling: the audience runs out before the budget does. Scaling on average return is how a campaign stays green on the dashboard while the incremental spend earns nothing.
Known Limitations
Marginal return needs enough recent spend variation per campaign to estimate, so a campaign held at one budget for months comes back with a wide band and the output says so. Contribution depends on product-level cost being available, and SKUs missing a cost are excluded from the margin calculation rather than assumed at the category average. It sets budgets and does not write them into the ad platform.
Install It
- Download the ZIP.
It is free and there is no account to create. - Unzip it into your agent's skills folder.
Claude Code reads ~/.claude/skills/, which is hidden by default: the command in the folder block opens it. Other agents scan their own directory, so drop the same folder there instead. - Ask your agent to use it.
Restart the agent if it was already running, then it picks the skill up with no config.
skills folder
Copy
~/.claude/skills/campaign-budget/
SKILL.md
references/margin-map.md
decisions/
# one file per day, with the reasoning
# macOS: create the folder and open it in Finder
mkdir -p ~/.claude/skills && open ~/.claude/skills
# Windows: paste in the Explorer address bar
%USERPROFILE%\.claude\skills\Frequently asked questions
How is this different from the platform's own budget recommendations?
The platform optimizes the metric it can see, which is revenue over spend. It does not know your cost of goods, your returns rate, or your shipping cost, so it cannot tell a profitable campaign from a busy one. This reads contribution after ad cost and recommends against the platform regularly.
Which agents does the skill run in?
Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision.
What does it need connected to work?
Campaign spend and revenue, product cost of goods, returns by product, and shipping cost. Without returns data it still computes gross contribution and labels the result as pre-returns, which usually overstates the thin margin campaigns most.
Does it ever recommend cutting a profitable campaign?
Yes, when the next dollar into it earns less than the next dollar somewhere else. A campaign can be clearly profitable on average and still be the wrong place for tomorrow's budget, and the marginal view is what separates those two cases.
How quickly does it react to a bad day?
It does not. Daily noise on a campaign spending a few hundred dollars is large enough to reverse the ranking, so the skill requires enough volume before it moves a budget and holds otherwise. The hold is a decision, not an absence of one.
What is included with the skill?
The skill itself, the margin map reference, the decision log format, and the marginal return model. It is free and security scanned. We re-publish the ZIP when the model changes, so download it again if your recommendations start shifting.
A budget change is only as good as the CAC behind it.
Polar computes blended CAC across 45+ sources, so increase, decrease or pause is set against the acquisition cost your team actually reports.
<script type="application/ld+json">{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How is this different from the platform's own budget recommendations?","acceptedAnswer":{"@type":"Answer","text":"The platform optimizes the metric it can see, which is revenue over spend. It does not know your cost of goods, your returns rate, or your shipping cost, so it cannot tell a profitable campaign from a busy one. This reads contribution after ad cost and recommends against the platform regularly."}},{"@type":"Question","name":"Which agents does the skill run in?","acceptedAnswer":{"@type":"Answer","text":"Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision."}},{"@type":"Question","name":"What does it need connected to work?","acceptedAnswer":{"@type":"Answer","text":"Campaign spend and revenue, product cost of goods, returns by product, and shipping cost. Without returns data it still computes gross contribution and labels the result as pre-returns, which usually overstates the thin margin campaigns most."}},{"@type":"Question","name":"Does it ever recommend cutting a profitable campaign?","acceptedAnswer":{"@type":"Answer","text":"Yes, when the next dollar into it earns less than the next dollar somewhere else. A campaign can be clearly profitable on average and still be the wrong place for tomorrow's budget, and the marginal view is what separates those two cases."}},{"@type":"Question","name":"How quickly does it react to a bad day?","acceptedAnswer":{"@type":"Answer","text":"It does not. Daily noise on a campaign spending a few hundred dollars is large enough to reverse the ranking, so the skill requires enough volume before it moves a budget and holds otherwise. The hold is a decision, not an absence of one."}},{"@type":"Question","name":"What is included with the skill?","acceptedAnswer":{"@type":"Answer","text":"The skill itself, the margin map reference, the decision log format, and the marginal return model. It is free and security scanned. We re-publish the ZIP when the model changes, so download it again if your recommendations start shifting."}}]}</script>