Search Term
by
Search term hygiene for ecommerce teams: join every query to the revenue and margin it actually produced, then return the negatives worth adding and the bids worth moving.
Search term reports are reviewed when someone remembers, and the review is usually done on clicks and cost. That finds the obviously irrelevant queries and misses the expensive ones: the terms that convert perfectly well into your lowest margin products, or that buy a customer who returns half the order. Meanwhile broad match keeps widening the surface faster than anyone can read it. This skill joins every search term to the orders it produced, carries product cost and returns into contribution margin per query, and returns two lists: the negatives worth adding, ranked by the margin they are draining, and the bids worth moving.
What It Does
- Pulls every search term with its spend, clicks, conversions and revenue.
- Joins each query to the products it actually sold.
- Carries product cost and return rate into contribution margin per term.
- Ranks wasteful terms by margin drained rather than by spend alone.
- Separates a term that never converts from one that converts unprofitably.
- Returns a negative keyword list and a set of bid changes, each with its reason.
Use Cases
- Run the search term review weekly instead of whenever someone has an afternoon.
- Catch queries that convert well into products you barely make money on.
- Keep broad match campaigns from widening faster than they are cleaned.
- Give a paid search manager a ranked list rather than a full term export.
Example Session
Review last month's search terms and give me the negatives and the bid changes worth making.
1,840 terms ran last month. Forty are worth acting on, and only half of those look bad on spend alone.
- Term | Spend | Conv | Revenue | Contribution | Action
- cheap running shoes | $1,240 | 31 | $1,910 | Minus $180 | Add negative
- waterproof hiking boots | $890 | 18 | $2,540 | Plus $610 | Raise bid 20%
- shoe repair near me | $410 | 0 | $0 | Minus $410 | Add negative
- trail shoes womens | $620 | 9 | $1,180 | Plus $95 | Hold
The first term is the one a spend review misses: it converts at a healthy rate and still loses money, because it sells almost entirely into the clearance end of the catalog and returns above average. The third is the obvious one. Adding both as negatives frees $1,650 a month, and moving that to the hiking boot term is the highest confidence bid change in the account.
Why This Beats Prompting It Yourself
Cost Per Click vs Margin Per Query: The Hidden Risk
A generic prompt reviews a search term report the way the report is laid out, on spend and conversions. Ask which terms to cut and you get the ones that never converted, which is the easy half. The term that converts at a good rate into your thinnest products and comes back at a 30% return rate looks like a winner in every column the report shows, and the prompt has no way to see the margin underneath it.
When You Still Need The Product Behind The Click
A query is only worth what the products it sells are worth. This skill states $1,910 of revenue on that term and minus $180 of contribution, with the product mix and the return rate that produced it, so the negative gets added for a reason a paid search manager can check. The bid increases come from the same calculation, which is why they tend to hold up a month later.
Known Limitations
Margin per query needs the order line detail that connects a click to the products it sold, and where attribution only reaches the order level the skill works at that grain and says so. Low volume terms are reported without a recommendation rather than judged on a handful of clicks. It produces the negative list and the bid changes as a plan and does not push them into the ad account itself.
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/search-term/
SKILL.md
references/match-types.md
reviews/
# one review per run, negatives and bids
# 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 reading the search terms report?
The report ranks on spend and conversions. This joins each query to the products it sold, carries cost and returns into contribution margin, and ranks on the margin being drained. The terms that change hands between those two views are usually the expensive ones.
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?
The search term report from the ad account and order line detail from the store, so a query can be traced to the products it sold. Without line detail it runs at order level, which still catches most waste and is labeled clearly in the output.
Will it add the negatives for me?
No. It produces the list with a reason against each term and the bid changes alongside. A negative applied in error is tedious to find later, so the plan is written for a human to apply after a read. Everything comes with the arithmetic that produced it.
How does it handle low volume terms?
It reports them without a recommendation. A term with four clicks and no conversion is not evidence of anything, and treating it as evidence is how accounts end up with negative lists that block queries nobody ever tested properly.
What is included with the skill?
The skill itself, the match type reference, the review template, and the margin rules used to score a query. It is free and security scanned. We re-publish the ZIP when the ad platforms change their report formats, so download it again if a review starts missing columns.
A wasteful search term costs more than it returns.
Polar joins search spend to net revenue and contribution margin, so the negatives you add are the ones draining margin rather than just clicks.
<script type="application/ld+json">{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How is this different from reading the search terms report?","acceptedAnswer":{"@type":"Answer","text":"The report ranks on spend and conversions. This joins each query to the products it sold, carries cost and returns into contribution margin, and ranks on the margin being drained. The terms that change hands between those two views are usually the expensive ones."}},{"@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":"The search term report from the ad account and order line detail from the store, so a query can be traced to the products it sold. Without line detail it runs at order level, which still catches most waste and is labeled clearly in the output."}},{"@type":"Question","name":"Will it add the negatives for me?","acceptedAnswer":{"@type":"Answer","text":"No. It produces the list with a reason against each term and the bid changes alongside. A negative applied in error is tedious to find later, so the plan is written for a human to apply after a read. Everything comes with the arithmetic that produced it."}},{"@type":"Question","name":"How does it handle low volume terms?","acceptedAnswer":{"@type":"Answer","text":"It reports them without a recommendation. A term with four clicks and no conversion is not evidence of anything, and treating it as evidence is how accounts end up with negative lists that block queries nobody ever tested properly."}},{"@type":"Question","name":"What is included with the skill?","acceptedAnswer":{"@type":"Answer","text":"The skill itself, the match type reference, the review template, and the margin rules used to score a query. It is free and security scanned. We re-publish the ZIP when the ad platforms change their report formats, so download it again if a review starts missing columns."}}]}</script>