Home/Paid ads/Claude Ads Skill
Keep ad spend on buyers who respond
A Claude ads skill packages instructions that Claude loads for ad work, such as keyword research or copy. Our paid ad consultants use live account data to run and review campaigns each week. You decide spending with them as the tests show where buyers respond.
What is a Claude ads skill?
A skill gives Claude a method it can use whenever the relevant job comes up. Anthropic's Agent Skills documentation describes packages of instructions, metadata and optional resources, loaded when needed. For advertising, a skill might hold the method for researching keywords, organising ad groups or writing copy. That guidance stays available across conversations, so each session can use the same advertising method.
A repeatable method helps the account receive the same checks each time. The instructions can tell Claude how to organise ad groups and what standard the copy must meet. The campaign also needs live account data, a schedule, review steps and a project memory that later runs read. On Tin's platform, a command turns a skill into a workflow with saved inputs, scheduling and an approval step.
You can run a skill by hand in Claude as each task comes up. An in-house team can turn it into a scheduled workflow on Tin's open-source platform. Our paid ad consultants can run the campaigns using those workflows, from the initial assessment through the weekly review. Compare those ways by who will read the results and decide the budget each week, including when the campaign should stop.
- Use current account data. Live campaign data lets the skill's checks examine what has happened since the previous run.
- Schedule the work. Saved inputs and a schedule make the checks part of the campaign's regular operation.
- Read each week's results. A person interprets the findings and chooses what the next run should do.
Ads without a stop rule spend budget on empty clicks
When the skill depends on someone remembering to open Claude, the campaign can spend between checks with nobody reading the account.
- Unproductive tests continue. If a test brings no qualified signups, spending continues until someone checks it and makes a decision.
- Clicks leave little evidence. Without a next action on the page, clicks reveal little about whether people want the product.
- The economics go unchecked. An assessment can show when a product’s economics cannot carry the cost of search ads.
We use scheduled checks and a cap, with a person reading the account to decide when spending should end.
Our paid ad consultants run the campaigns
We assess the product, prepare the campaign and read the account each week. The skill supports the consultants doing those jobs.
Assess whether ads can pay
The assessment gives you a verdict, the constraint behind it, a scorecard and a rough campaign shape.
Prepare and launch campaigns
Our consultants run Google, Bing and LinkedIn campaigns with landing pages for the needs behind the searches.
Connect ads to conversions
We set up conversion tracking and audiences for the campaigns, including remarketing to visitors who have already encountered you.
Tune spending each week
Campaign check-ups show our consultants what happened in the account before they adjust the next budget.
We agree the budget cap before launch
We assess the product's economics with you, then agree how much to spend and what would bring the test to an end.
First
Assess the business case
We examine the product, keyword data and site to judge whether Google Search ads can pay for themselves.
Then
Agree the spending limits
We decide the cap and stop rule with the founder, making each spending decision explicit before the test runs.
Next
Review before launch
A person on our team reviews the ads and pages, and the campaign launches after the founder's approval.
Each week
Decide from campaign checks
On the call and in Slack, we use the account findings to choose what to continue, change or stop.
Results from each run shape the next one
The next run can read the earlier campaign findings in project memory. Our consultants use the same record when they decide which campaigns should receive more budget and which should end.
- Carry findings forward. Project memory keeps earlier campaign work available when the skill runs again.
- Move the budget deliberately. Weekly decisions connect the account's response to the campaigns we continue, adjust or end.
- Keep the campaign knowledge. Your ad accounts, campaign data and the Claude workflows we build remain yours.
It fits founders already using Claude for marketing
Teams using coding agents
We run campaigns for teams that already use Claude Code, Codex or Cursor in their work.
Startups testing paid channels
Pre-seed to Series A B2B companies can work with our consultants on their first Google, Bing or LinkedIn tests.
Founders with an ads skill
We schedule the skill a founder has written and put a consultant on the job of reading the campaign each week.
Talk to Tin
Let's discuss whether search ads can pay for your product and what an initial test should check. Working with our consultants involves no long-term commitment.
FAQ
Can we bring the ads skill we already wrote?
Yes. A command can turn your skill into a workflow on Tin's platform. That adds saved inputs and a schedule to the instructions. We can discuss the method you have written and how our consultants would use it while running and reviewing the campaigns for your company.
Do we need Claude Code to work with you?
No. Tin also works with Codex and Cursor, and projects are accessible in the browser at app.tin.computer. Our team runs the workflows during an engagement. You can use the browser to follow the work and connect integrations without operating the campaigns yourself.
What do you need from us before launch?
We start with short conversations, access to the ad account and site, and the product's economics. Those let us assess whether search ads can pay and set up the work; Google Ads connects natively. Founder conversations and approvals generally take about two hours a week, with the founder approving the important steps.
What happens to the skill if we stop working together?
The skill and the workflows we build for your company stay with you, along with your accounts and campaign data. Tin's platform is open source under Apache 2.0. Your team can keep the instructions and the record of earlier work when deciding how to run campaigns afterwards.
Talk about the ads your product could run
Talk with paid ad consultants and engineers about the campaigns you have in mind, whether a skill already exists or you are starting from the product. We can discuss what the account would need for a useful test.