Skip to content

Decides

Founder posts to ads

Picks the founder posts that earned real reach against the founder's own baseline, records the organic figures first, and sets them up as Thought Leader Ads.

Other tools

Add this skill to Claude

Claude takes skills as a ZIP file. Your download has started; if it didn't, use the button below.

  1. 1In Claude, open Customize, then Skills.
  2. 2Choose +, then Upload a skill, and pick founder-to-ads.zip.
  3. 3Ask for it by name, or let Claude use it when a task matches.

Skills need code execution turned on in Claude's settings. On a Team or Enterprise plan, an admin can share it with everyone.

Download the folder

Try it now, without installing

See the skill work on a made-up input before you add it. Nothing is installed, and you can swap in your own input.

Open in the Claude app

Opens a new chat in the Claude desktop app with the prompt filled in. Nothing is sent until you press enter.

Includes the whole SKILL.md and the sample input. Paste it into Claude, ChatGPT or any chat.

Influenced by
Thought Leader Ads
Needs
Post figures for the last 90 to 180 days · An ad account linked to your company Page
Works with
Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md
Version
1.0.0 · Last checked · License MIT
Maintained by
Sabih AhmedContent
SkillContent

Founder posts to ads

Example output

Which posts to promote
PostStatusWhy
3 CRM fields before a board meetingPickedTop-quarter reach, engagement held
Our first office (team photo)Left outPersonal milestone
One field fix moved creditBelow baselineReach without engagement

Picks the founder posts that earned real reach against the founder's own baseline, records the organic figures first, and sets them up as Thought Leader Ads.

The skill

The whole skill, in plain text.

Nothing is hidden. Read it, change it to fit your team, or take the whole folder.

Scripts
Optional Python scripts for the counting. It works without them and says which way it ran.
Network
Doesn't need the internet.

3 test cases

founder-to-ads/SKILL.md

In the folder

  • founder-to-ads/SKILL.md
  • founder-to-ads/references/setup-sheet.md
  • founder-to-ads/scripts/tla_pick.py
---name: founder-to-adsdescription: >-  Picks the founder's LinkedIn posts that earned real reach against the founder's own baseline (top-quarter  impressions with engagement that held at that reach), checks each is on-message, accurate and promotable, records  the organic baseline before any paid push, and writes the Thought Leader Ads setup sheet: permission, objective,  audience from company-profile.md, exclusions, the reader's budget and a flight with a stop rule. Returns PROMOTE, NOT  YET, NEEDS DECISION or BLOCKED. Use it when a founder posts regularly and the company wants paid reach behind what  already works. Not for planning all LinkedIn Ads (use linkedin-plan) or writing new posts (use founder-post).license: MITcompatibility: >-  Works in Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md. Works from a post analytics export or  figures pasted from each post. Never signs in to or changes an ad account. Python 3 is optional, for the baseline and  the picks.metadata:  author: SHAKEUP  version: "1.0.0"  stage: decide  plugin: content  homepage: https://www.shakeuphq.com/skills/founder-to-ads--- # Founder posts to ads Finds the founder posts strangers would still care about, and records what organic did before paid reach muddies it. ## Purpose - One job: decide which founder posts to promote as Thought Leader Ads, and set them up to be measured.- Stage: decide. It ends in a verdict from ordered rules, with the posts picked and the reason for each one left out.- It never invents a figure or a budget, never promotes a post without its baseline on record, and never touches the ad account: the reader builds the campaign from the sheet. ## Use when / not | Situation | Use || --- | --- || The founder has posted for a while and the company wants paid reach behind the best posts | this skill || A Thought Leader Ads flight ended and needs a keep or stop call | this skill, step 9 || Planning audiences, budgets and formats across all LinkedIn Ads | `linkedin-plan` || Writing the posts | `founder-post` || Checking a post's figures before it's promoted | `claim-check` (this skill runs it) | ## Inputs | Input | Required | Ways to provide, in order | If missing || --- | --- | --- | --- || Post figures: post, date, type, impressions, reactions, comments, reposts | yes | the founder's post analytics export; figures copied from each post's analytics; a table the reader pastes | `BLOCKED`: ask for the last 90 to 180 days of posts || ICP share of each post's engagers | no | the post's demographics; a count of engagers' titles the reader provides | audience fit `not checked`; coverage `partial` || The text of each qualifying post | no | links the agent can open; pasted text | ask for it once; a qualifier with no text can't pass the checks in step 5, so it's held || An ad account linked to the company Page | yes | `## Accounts and data` and `## Channels` in the profile; the reader's answer | ask once; `BLOCKED` if there is none || The budget for this test | no | `## Channels` (paid social budget); the reader's answer | `NEEDS DECISION`: picks are ready, the budget isn't | ## Profile read From `company-profile.md`: `## Company` (regions), `## Buyers` and `### Buying group` (the audience), `## Messaging house` and `## Buying moments` (on-message test), `## Claims and proof` with `### Never say`, `## Competitors` (mention policy, exclusions), `## Channels` (the paid social line and its budget), `## Accounts and data` (ad accounts), `### Founder voice` (main topics), and every `## Learnings` entry for `founder-to-ads`. No profile: ask in one message for the posts' figures, the buyer's roles and company size, whether an ad account is linked to the company Page, and the test budget. Say that `company-profile` saves this next time. ## Steps 1. **Apply learnings.** Past flight results (`fact`) and accepted rule changes (a different minimum, a different stop rule) override the defaults; name each one that fires. 2. **Check the channel.** If `## Channels` marks paid social on this platform `never`, or the reader confirms no ad account is linked to the company Page, stop with `BLOCKED`: say what would have to change. A profile line saying `testing` or `active` is enough to continue. 3. **Load the posts.** Take every post from the last 90 to 180 days. Mark the type (image, video, text, document, poll, event, article, reshare) and mark posts that are personal (a life event or a celebration, not a point of view). 4. **Set the baseline and pick.** Run:    ```   python3 scripts/tla_pick.py --csv posts.csv --today <YYYY-MM-DD>   ```    Without Python, work it out by hand and say so: engagement = reactions + comments + reposts; engagement rate and comment rate per impression; leave out posts under 7 days old, polls, events, articles, reshares and personal posts; take the median and the 75th percentile of impressions and the median engagement rate across the rest. With fewer than 10 posts left, stop at `NOT YET`. A post qualifies when its impressions are at or above the 75th percentile and its engagement rate at or above the median (and its ICP share at or above the median, when known). Rank qualifiers by comment rate, then engagement rate; take up to 3.    | Default | Reason |   | --- | --- |   | 10 posts minimum | a median of fewer posts swings on one outlier |   | 7 days old minimum | a younger post's organic figures are still moving, so its baseline would undercount it |   | 75th percentile of impressions | the post travelled past the founder's usual audience |   | Median engagement rate or better | people kept engaging at that reach; high reach with a low rate travelled on something other than the idea |   | Up to 3 posts | each gets enough of the budget to compare against its baseline within one flight | 5. **Check each pick.** A pick that fails one of these is held, with the reason:    | Check | Holds the post when |   | --- | --- |   | On-message | it maps to no pillar in `## Messaging house` and no moment in `## Buying moments` (paid reach should buy attention for what the company sells) |   | Accurate | `claim-check` (or its test inline) finds any `unsourced`, `unsupported`, `expired` or `forbidden` claim; rerun it on posts older than 90 days |   | Promotable | never held here (polls, events, articles and reshares were left out in step 4); a type other than image or video stays picked as `confirm type`, and the budget split says where its share goes if Campaign Manager won't promote it |   | Still true | it mentions a price, a feature, a date or an offer that has changed |   | Mention policy | it names a competitor and `## Competitors` mention policy doesn't allow it in ads | 6. **Record the baseline.** For each post still picked, fill section 1 of `references/setup-sheet.md`, dated today. No baseline, no promotion. 7. **Write the setup sheet**: permission request, objective, audience and exclusions from the profile, the reader's budget split, a flight of 2 to 4 weeks, campaign names (sections 2 and 3). 8. **Set the stop rule** (section 4) so the keep-or-stop call after the flight is decided before money is spent. 9. **After the flight** (when the reader returns with results): compare each post with its baseline using section 4, make the call, and write each result as a learning. ## Check and verdict | Verdict | Condition (first match wins) || --- | --- || `BLOCKED` | No post figures; the channel is marked `never`; or no ad account linked to the company Page || `NOT YET` | Fewer than 10 posts old enough to count; or no post qualifies; or every qualifier is held in step 5. Say what to collect or post next || `NEEDS DECISION` | Picks pass step 5 but the budget isn't known, or a pick needs the reader's call (a competitor named under a `facts only` policy, a post older than 180 days) || `PROMOTE` | 1 to 3 posts picked, each passing step 5, with its baseline recorded and a budget set | Named stops: `BLOCKED` asks for the missing input. `NEEDS DECISION` stops with the picks and the one decision needed. No other stops. ## Learn Append to `## Learnings` when: - a flight ends: each post's paid engagement rate against its organic rate, ICP share and the call made (a `fact`, with the source and dates);- the reader promotes a post this skill held or skips one it picked (a `proposal` with their reason);- results across 2 or more flights suggest a different threshold, such as a lower stop line for video (a `proposal`, applied only once `accepted`). Example: `- 2026-10-24 | founder-to-ads | fact | Teardown post kept 71% of its organic engagement rate when paid; contrarian post kept 38% and was stopped | Why: Campaign Manager and post analytics, flight 2026-09-15 to 2026-10-10 | Status: noted` ## Output ```Verdict: <PROMOTE | NOT YET | NEEDS DECISION | BLOCKED> · Coverage: <complete | partial (no ICP share; post text missing for n)> · Baseline: <script | by hand>Baseline rules: <n> posts counted · median impressions <n> · 75th percentile <n> · median engagement rate <x%> · median ICP share <x% | not available> | Post | Date | Type | Impressions | Engagement rate | Comment rate | Status | Reason |(every post; status from step 4, then holds from step 5) Picked: <post, why, pillar, moment> (one line each)Organic baseline (record now): <section 1 fields per picked post, dated>Setup sheet: <sections 2 and 3, filled>Stop rule: <section 4, with this flight's dates>Decision needed (NEEDS DECISION only): <the one decision>Learnings applied: <list, or none>``` Then the handoff block. ## Hand off ```handoffskill: founder-to-adsversion: 1.0.0verdict: PROMOTEcoverage: partial (no ICP share)for: <founder name> posts, <date range>outputs:  picks: [<post id>, <post id>]  baseline_recorded: <YYYY-MM-DD>  flight: <start> to <end>  stop_rule: paid engagement rate under half of organic  budget: <the reader's figure, or not available>next: [linkedin-plan]open: [confirm the document post can be promoted]``` `linkedin-plan` reads `picks` and `budget` to fit Thought Leader Ads into the wider paid plan. After the flight, rerun this skill with the results. ## Failure modes | Problem | Do this || --- | --- || The reader wants to promote first and take the baseline after | Refuse the order: once paid delivery starts, organic and paid mix and the baseline is lost. Record it first; it takes minutes || The export has impressions but no comments split | Use total engagement only; comment-rate ranking falls back to engagement rate; say so || The founder's best post is a personal milestone | Leave it out; suggest a point-of-view post on the same theme instead || All picks are the same archetype | Keep them, and note it: the flight then tests one shape, not the founder || Figures are pasted with rounded impressions ("4.2K") | Use them as given and say they're rounded || The post is by someone outside the company | Allowed when they approve the request; record their approval as an open item | ## Influenced by - LinkedIn's Thought Leader Ads guidance: sponsoring posts from employees and others with their approval, the objectives they run under, and saving the post's organic performance before the campaign starts so its impact can be measured: https://business.linkedin.com/advertise/ads/sponsored-content/thought-leader-ads- SHAKEUP's paid social practice: posts picked against the founder's own median, not an outside benchmark, and a stop rule set before the money is spent.

How it thinks

Inputs in, a verdict out, and a stop only where you decide.

What we built

Posts are picked against the founder's own median, not an outside benchmark, because what counts as good reach differs by audience. The stop rule is set before the flight, so nobody decides after the fact that a weak result was fine.

It reads your company profile if you have one, or asks for only what it needs. Then it works through its steps, checks its own work and ends in a verdict.

The steps

  1. 01

    Check the channel is allowed

    If your profile marks LinkedIn Ads as never, or there's no ad account linked to your Page, it stops and says what would have to change.

  2. 02

    Set the founder's own baseline

    Its script takes the median and the top quarter of the founder's posts, leaving out polls, events, reshares, personal milestones and posts under a week old.

  3. 03

    Pick what earned reach

    A post qualifies when its reach is in the founder's top quarter and its engagement rate held at that reach. Comment rate breaks ties, because a comment takes more effort than a reaction.

  4. 04

    Check each pick

    It has to support a message you sell, pass a claim check, still be true today and follow your competitor policy. Paid reach shows every figure to many more people.

  5. 05

    Record the baseline first

    The organic figures for each pick are written down, dated today, before any spend. Once paid delivery starts, organic and paid mix and the baseline is gone.

  6. 06

    Write the setup sheet and the stop rule

    Permission request, objective, audience from your buying group, exclusions, your budget split and a 2 to 4 week flight. The keep or stop call is set before the money is spent.

What you need

  • Required

    Post figures for the last 90 to 180 days

    The founder's post analytics export → Figures copied from each post's analytics → A table you paste

  • Optional

    Share of each post's engagers who fit your buyer

    The post's demographics → A count of engagers' titles you provide

  • Optional

    The text of each qualifying post

    Links your agent can open → Text you paste

  • Required

    An ad account linked to your company Page

    The Channels and Accounts sections of your company profile → Your answer

  • Optional

    The test budget

    The Channels section of your company profile → Your answer

What it can return

  • PROMOTE

    1 to 3 posts are picked, each passes its checks, the baseline is recorded and a budget is set.

  • NOT YET

    Fewer than 10 posts are old enough to set a baseline, or no post beat it. It says what to collect next.

  • NEEDS DECISION

    The picks are ready but the budget isn't set, or a pick needs your call.

  • BLOCKED

    There are no post figures, the channel is marked never, or there's no ad account linked to your Page.

Where it stops for you

  • When there's no ad account or the channel is marked never
  • When the picks are ready but the budget isn't

Example output

What you get back.

A run for a made-up company. Your output depends on your inputs.

Here's a run for a made-up B2B SaaS company whose CEO posts weekly. The reader set a test budget for a 3-week flight.

Verdict

PROMOTE

Coverage: complete

Baseline

10 posts counted, median 5,600 impressions, top quarter from 8,125, median engagement rate 1.9%.

PostTypeImpressionsEngagement rateStatusReason
The 3 CRM fields to check before a board meetingImage9,4002.7%PickedTop-quarter reach, engagement held
A 7-point attribution self-checkDocument10,2002.1%Picked, confirm typeConfirm a document post can be promoted
Our first office (team photo)Text12,1004.2%Left outPersonal milestone
Which attribution model do you use?Poll8,8002.1%Left outPolls aren't promoted
Paid social lost credit after a field fixImage8,3001.4%Below baselineHigh reach, weak engagement
Organic baseline, recorded today

impressions, engagement and comments for both picks, before any spend.

Stop rule

a post whose paid engagement rate falls under half its organic rate stops.

The handoff block the next skill reads
skill: founder-to-ads
version: 1.0.0
verdict: PROMOTE
coverage: complete
for: CEO posts, July to October
outputs:
  picks: [P02, P08]
  baseline_recorded: 2026-10-24
  stop_rule: paid engagement rate under half of organic
next: [linkedin-plan]
open: [confirm the document post can be promoted]

What good looks like

  • Every post shown with its status and the reason for it
  • Personal milestones and polls left out of the baseline
  • The organic baseline recorded with today's date
  • Your budget split across the picks, never a made-up figure
  • A stop rule written down before the flight starts

Install

Use it in the tool you already have.

Pick your tool. Each tab has the shortest path we know of.

  1. 01
  2. 02
    In Claude, open Customize, then Skills.
  3. 03
    Choose +, then Upload a skill, and pick the ZIP. Code execution must be on in your settings.
  4. 04
    Ask for it by name, or let Claude use it when a task matches.

Questions

Before you add it.

Why record the organic figures first?

Once a post is promoted, its figures mix organic and paid. Without a baseline taken before, you can't tell what the ads added.

Why the founder's own median?

Reach depends on the founder's audience, so an outside benchmark says little. A post that beats the founder's own normal travelled further than usual.

Can every post type be promoted?

Polls, events, articles and reshares are left out. LinkedIn's page lists single image and video formats, so other types are kept but flagged to confirm in Campaign Manager first.

Will it touch our ad account?

No. It writes the setup sheet, and you build the campaign. The founder approves each post in Campaign Manager before anything runs.