Skill v1.0.0
currentLLM-judged scan95/100version: "1.0.0" name: abm-program-setup description: >- Run this first, before any other ABM & Scale OS skill. Use when the user says "start an ABM program", "set up ABM", "new ABM engagement", "build the program charter", "we want to do account-based marketing". Interviews the user on objective, motion, segment, budget, team, and constraints, checks whether the company is ready for ABM at all, and writes workspace/program/charter.md plus stack.md. Every other skill reads these files.
ABM Program Setup
Establish the program charter that every other skill reads: what the program is for, who it targets, what motion it runs, what it can spend, who does the work, and what stops it.
When to use
- A new ABM program, a new client engagement, or a reboot of a program that drifted
- "We want to do ABM" with nothing written down yet
- Before
icp-definition,account-sourcing, or anything downstream. Those skills route back here ifcharter.mdis missing
Inputs
- Reads:
workspace/program/charter.mdif one exists (updates rather than overwrites) - Needs from user: the interview below. Do not guess any of it
Workflow
1. Readiness check, before anything else
ABM is expensive and slow. Run these four questions and say plainly if the answer is no.
| Question | If no | |
|---|---|---|
| Is the ACV high enough to justify per-account effort? Roughly $25k+ for 1:few, $75k+ for 1:1 | Recommend a segment or demand-gen motion instead and say why | |
| Has anyone closed a deal in this segment without founder heroics? | The company is pre-product-market-fit for this segment. Recommend founder-led validation first | |
| Are there sales resources to work the accounts? | ABM with no seller is a content program. Say so | |
| Can the target list be named, or is it "anyone in the category"? | Do icp-definition as a standalone project first |
Refuse to run a Series B playbook at a company that has not proven the segment. State the constraint, offer the smaller thing that fits.
2. Interview, in one batch
Ask all of this at once, then wait. Do not drip questions.
Objective and scope
- What is the single primary objective this quarter? (New logos in a segment, expansion in the installed base, competitive displacement, a specific product's adoption, entry into a new vertical or geography)
- What number would make this a success, and by when?
- What is explicitly out of scope?
Segment and motion
- What segment: industry, size band, geography, business model?
- Net-new acquisition, installed-base expansion, or both? (Both means two programs. Say so)
- Sales-led, product-led, partner-assisted, or founder-led today?
- Typical ACV and sales cycle length
Team
- Who owns the program? Who are the sellers? Who does list building, who does creative, who does ops?
- Is there an executive sponsor on the sales side? Named?
- How much of a marketer's week does this get?
Budget and stack
- Program budget for the quarter, split between media, data, and events if known
- Current tools by layer (see
docs/stack.mdfor the layers) - What is already committed and cannot be changed
Constraints
- Brand, legal, or compliance rules that affect outreach
- Regions with data restrictions (GDPR, CCPA, sector rules)
- Accounts that are off limits: partners, competitors, existing escalations, board relationships
- Anything that failed before and why
3. Set the sprint
Ninety days, one primary objective, three gates. Fill in the sprint table with dates. Nothing gets planned past sprint one.
4. Write the files
workspace/program/charter.mdfromtemplates/program-charter-template.mdworkspace/program/stack.mdfrom the tool answers, using the table shape indocs/stack.md
5. Name the open questions
List every answer the user could not give, with who owns getting it and by when. Do not fill them with assumptions. An unanswered question in the charter is more useful than a confident guess.
Output
- Writes:
workspace/program/charter.md,workspace/program/stack.md - Prints to chat: the readiness verdict, the one-line objective, the sprint gates, and the open questions list
Rules & quality bar
- One primary objective. If the user gives three, make them rank and put two in "later". A program with three objectives has none
- Say no when the readiness check fails. Recommending ABM to a company that cannot run it is how a program burns a quarter and the marketer's credibility
- Budget caps are written before plays exist. It is the only time they can be set unemotionally
- Never invent the numbers. ACV, cycle length, and budget come from the user or are marked unknown
- Both motions means two programs. Acquisition and expansion have different lists, different messaging, and different metrics. Do not let them share a charter
Related skills
- Hands off to:
icp-definitionnext, always - Read by: every other skill in the repo
- See also:
docs/methodology.mdsection 10 for the sprint model,docs/stack.mdfor the tool layers