Submit your work Try your firsta scenario

Ads and creativeWeekly

By Moby Hayat

Weekly check for conversion campaigns

Judges every Lead Gen Form and Message ad against your goal cost per lead: which ads to pause, which winners to copy into new versions, and what to check when every ad in an ad set is expensive. It gives recommendations to approve and changes nothing.

Get the install commandSource on GitHub

Using it? Join the community

Runs with
  • LinkedIn Ads export
  • DemandSense (optional)
You need
The Ad Performance export with a daily breakdown (from each ad’s first day, or the last 37 days), the Campaign Performance export with audience penetration and frequency, and a short note per ad set: goal cost per lead, cold or retargeting, bid. Optional: your ICP with the top 100 companies of the saved audience, and last week’s table. Or the DemandSense MCP.
You get
A headline count of ads to pause, a table of flagged ads with the numbers behind each, winners to copy with their headlines, a diagnosis for ad sets where every ad is expensive, and a table to paste next week.

What you get

What to pause, what to copy. Every ad against your goal cost per lead

  • The weekly check: Every ad against its goal cost per lead: pause, keep or too early
  • Pause and replace: The winners to build new versions from, with their headline and intro text
  • The ad-set diagnosis: Delivery, audience size and audience quality where every ad is expensive
  • Your decisions: Six recommendations to approve, and how last week’s decisions worked out

Get it

Paste one prompt. It recommends and changes nothing

Your exports 
Load the Weekly Check for Conversion Ads scenario from https://github.com/demandsense-labs/linkedin-ads-skills/tree/main/conversion-ad-weekly-check and run it on the exports I’ve attached.
DemandSense 
Load the Weekly Check for Conversion Ads scenario from https://github.com/demandsense-labs/linkedin-ads-skills/tree/main/conversion-ad-weekly-check and run it on my DemandSense workspace. Pull the data through the connector rather than asking me for exports.

Rather not paste it? Install it instead

Add it to Claude Code once and call it by name on any export

Claude Code 
/plugin install linkedin-ads-skills --marketplace demandsense-labs/linkedin-ads-skills

Read the source on GitHub Skill folder on GitHub

How it runs

6 steps, in order. Each one applies a rule of the Standard, and stops the run when the data can’t carry it

  1. Scope

    Lead gen and Message ad sets that spent in the last 7 days. Any without a goal cost per lead is skipped.

  2. No leads

    Flag for pause once the ad has spent twice the goal cost per lead.

  3. Expensive leads

    Flag for pause when cost per lead reaches 1.5 times the ad set’s own baseline.

  4. Winners

    Ads at or under goal. The top two become the source for new versions with the same offer.

  5. Whole ad set expensive

    Check delivery (spend against daily budget, manual bid), audience size (penetration under 10% means the budget is too small or the audience too big; 20% and up is where you want it) and audience quality (share of engaging companies that fit the ICP).

  6. Approve

    One recommendation per flagged ad or ad set.

The rules it applies

2 rules of the Standard. Each step above leans on one of them. Open a rule to see what it checks and why

Hi there, here is myskill

Run a check we don’t have? Put it in the Standard

Send a skill or a whole scenario as you run it today. We package it, test it with you and publish it under your name, with your company and links on every page it appears on

Bring your work