Running a Scale AI Adoption Plan

• Source: Report framework library

Encodes a Revenue AI Report framework so the agent applies the published method instead of improvising one.

1. 1.Copy the SKILL.md text below, or download the raw file.
2. 2.Create a folder named exactly running-a-scale-ai-adoption-plan in your agent's skills directory.
3. 3.Save the file inside that folder as SKILL.md.
4. 4.Ask the agent one of the trigger requests below.
5. 5.Check the output against what you already know before it leaves your desk.

• We have AI budget for next year and no idea where to start. Give me a plan.
• Run SCALE on our plan to put AI into the mid-market sales motion.
• The board wants a five-week AI adoption plan they can approve, not another deck.
• We bought Agentforce seats six months ago and nothing changed. Rebuild the plan properly.

• Is our AI SDR project an Optimize or a Reinvent initiative?
• Score this running initiative out of five and tell me whether to kill it.
• Write a LinkedIn post about AI adoption in B2B sales.

---
name: running-a-scale-ai-adoption-plan
description: Turn an unfocused AI ambition into a five-week operating plan where every step ends in a deliverable and week five ends in a measured scale-or-kill decision.
---

# Running A SCALE AI Adoption Plan

Turn an unfocused AI ambition into a five-week operating plan where every step ends in a deliverable and week five ends in a measured scale-or-kill decision.

## When to use this skill

- A leader asks where to start with AI and has budget but no plan.
- Licenses were bought before anyone wrote down the business number they were meant to move.
- A pilot is running with no success metric, so nobody can say whether to expand it.
- Multiple teams are buying overlapping AI capability and there is no governance to stop it.
- A board or CFO has asked for an AI plan they can approve rather than a deck.

## Inputs to collect

- The one or two business targets the company already committed to this year, with the current number and the target number. Source: the annual operating plan or board deck, not the AI vendor.
- The workflow to be changed, end to end, with the owning team named at each handoff. Source: RevOps process documentation plus interviews with the people doing the work.
- Current cycle time, conversion rate, and cost at each step of that workflow. Source: CRM reports and finance system, not estimates.
- Data readiness for that workflow: field completeness, identity resolution, and who controls access. Source: RevOps or the CRM admin.
- Existing AI spend and contracts touching this workflow. Source: finance system and vendor invoices.
- Names of candidate vendors or internal builds already under consideration, with prices. Source: procurement.
- The pre-AI baseline for the target metric, captured before anything is switched on.

## Process

Run the five steps in order. One week per step is the working default, so five weeks to a controlled pilot with a measured result (https://www.therevenueaireport.com/frameworks/scale). Do not start a step until the prior step's artifact exists.

1. **Strategic Outcomes.** Pick one or two measurable business targets before touching a tool: revenue up, win rate up, cycle time down, or margin up. Write the specific delta and the gap between the current number and the target. Artifact: an outcome charter with a specific delta and a gap analysis.
2. **Chart Friction.** Map the workflow and find where work gets stuck. Separate people friction from process, technology, and data friction, then quantify what each bottleneck costs in time or money. Artifact: a prioritized friction map.
3. **Align Capabilities.** Match one AI capability to each real bottleneck and assess whether data, process, and people are ready. Score vendors on features, fit, security, and scale. Artifact: a capability match map and a scored vendor shortlist.
4. **Launch with Control.** Run one high-value pilot with a small team, clear success metrics, guardrails, and a feedback loop. Keep it time boxed and small in scope. Artifact: a running pilot with controls and written kill criteria.
5. **Evolve and Expand.** Measure impact against the pre-AI baseline. Scale what worked, kill what did not, and stand up a center of excellence for governance. Artifact: a scaling blueprint and a governance structure.

## Decision rules

- Start with the business initiative, not the problem and not the tool. Jonathan Kvarfordt states the order directly: "So many people come out and say focus on your problem first. That's step B. That's not step A. Step A should be what are the business initiatives you're focusing on this year" (https://www.therevenueaireport.com/frameworks/scale). A plan that starts at the tool cannot be traced back to a number the CFO recognizes.
- Do not automate a broken process. Teams that skip friction mapping get faster broken work (https://www.therevenueaireport.com/frameworks/scale).
- Treat a pilot without success metrics as a demo, not a pilot. A demo cannot scale because there is nothing to measure against (https://www.therevenueaireport.com/frameworks/scale).
- Stretch Chart Friction, not the other steps, in a large organization. Workflow mapping is the slowest step and the site names it as the one that stretches (https://www.therevenueaireport.com/frameworks/scale).
- Capture the pre-AI baseline before the switch is flipped. Without it you cannot separate the AI effect from the market or from luck (https://www.therevenueaireport.com/frameworks/sling).
- Fund one fully attributed workflow rather than a portfolio of pilots. Attribution, not spend, is the constraint on the next budget, and only 22 percent of finance leaders can tie AI spend to a business outcome (CloudZero, n=260 finance executives including 135 CFOs, https://www.therevenueaireport.com/research/spend-vs-attribution).
- Include the verification and data cost line in the business case. Refinement work of checking, repairing, and reverifying is about 60 percent of an agentic task's cost, and it is routinely omitted (McKinsey, via https://www.therevenueaireport.com/research/spend-vs-attribution).
- Redesign the work, do not just insert the model. McKinsey tested 25 organizational attributes and found fundamentally redesigning workflows had the largest effect on EBIT impact (https://www.therevenueaireport.com/research/what-works).
- Budget for a reversal in step five. In a 10-country sample of 2,527 senior decision makers, 74 percent had already rolled back or shut down a deployed AI customer communications agent over a governance failure (Sinch, The AI Production Paradox, fielded January to February 2026, vendor research, https://www.therevenueaireport.com/research/rollback).
- The site publishes no numeric threshold for how large a friction cost must be to qualify for step three. Set that threshold with the team, state it in the friction map, and hold it constant for the rest of the run so the shortlist cannot be re-cut to fit a favored vendor.
- Stand up governance before scaling. Successful pilots without governance fragment into sprawl within a year (https://www.therevenueaireport.com/frameworks/scale).
- Separate reported adoption from operational adoption in step five. They are different numbers, and most of the gap is projects that skipped the outcome step (https://www.therevenueaireport.com/frameworks/scale).
- Run each phase as a standalone unit when the organization cannot commit five consecutive weeks. Each phase can be run on its own, and each phase's deliverable is the entry ticket to the next (https://www.therevenueaireport.com/frameworks/scale).

## Output requirements

Deliver a five-row plan table plus the five artifacts. Use this exact table shape, because each row is the entry ticket to the next step.

| Step | Artifact | Owner | Due | Gate to pass before the next step |
|---|---|---|---|---|
| Strategic Outcomes | Outcome charter with delta and gap analysis | | Week 1 | One or two targets stated with a current number and a target number |
| Chart Friction | Prioritized friction map | | Week 2 | Each bottleneck classified people, process, technology, or data, with a cost |
| Align Capabilities | Capability match map and scored vendor shortlist | | Week 3 | Every shortlisted vendor scored on features, fit, security, and scale |
| Launch with Control | Running pilot with controls | | Week 4 | Success metric, guardrails, feedback loop, and kill criteria written down |
| Evolve and Expand | Scaling blueprint and governance structure | | Week 5 | Post-AI number compared to the captured pre-AI baseline |

Also deliver: the pre-AI baseline value and the date it was captured, the kill criteria with a named owner who can trigger them, and a one-paragraph scale-or-kill recommendation with its caveat.

## Verification loop

Validate the plan before it goes to anyone.

1. Check every gate in the table. Each row must have an artifact, a named human owner, and a date.
2. Check that the outcome charter's metric appears unchanged in the step five measurement. If the metric changed mid-run, the plan measured a different thing than it promised.
3. Check that the pre-AI baseline has a capture date earlier than the pilot start date.
4. Check that every number in the plan carries a source: a CRM report, a finance record, an invoice, or a published URL.
5. Fix every failure, then re-run checks 1 through 4 from the top. Do not patch one line and declare the plan clean, because a changed metric usually invalidates the baseline as well.

Only proceed to present the plan when all four checks pass on the same version of the document. If any gate cannot be filled, present the plan with that gate marked unmet and name it as the open risk, rather than filling it with an estimate.

## Quality checks

- The business target was written before any vendor name appears in the document.
- Friction is classified into all four categories, not only technology.
- Every shortlisted vendor carries all four scores: features, fit, security, scale.
- The pilot has a written kill line with a metric, a threshold, and a date.
- The pre-AI baseline has a value and a capture date.
- No number appears without a source.
- Governance is a named structure with an owner, not a line saying governance will follow.
- Reported adoption and operational adoption are reported as two separate figures.
- The scale-or-kill recommendation names which one it is, with its caveat.

## Limitations

- SCALE produces a defensible plan, not a guarantee. It reduces the chance of unmeasurable spend; it does not make an unready data estate ready.
- The five-week default assumes one workflow. Multi-workflow programs run the five steps once per workflow rather than compressing them.
- Attribution stays hard. The method separates what can be defended from what cannot, and it does not manufacture certainty.
- The framework does not publish thresholds for friction cost or vendor scores. Those are team-set, and a team that sets them loosely will produce a loose shortlist.

## Example input

A 400-person B2B software company. Committed target from the annual plan: cut sales cycle time from 74 days to 60 days this year. Two AI vendors already in procurement. No pilot running. No baseline captured. Illustrative and synthetic, provided to show output shape.

## Example output

Outcome charter: reduce median sales cycle from 74 days to 60 days, a 14-day delta, measured on closed-won opportunities in the mid-market segment.

Friction map: security review is the longest step at a median of 19 days and is process friction, not technology friction. Proposal drafting is 4 days and is people friction. Data friction is present because 31 percent of opportunity records have no stage-entry timestamp, which makes any cycle-time claim unverifiable until fixed.

Capability match: the proposal drafting bottleneck matches a drafting capability. The security review bottleneck does not match any AI capability on the shortlist, and it is the larger cost. Recommendation is to fix the timestamp gap and re-sequence security review first, and to run the drafting pilot second.

Pilot: eight reps, four weeks, success metric is median days from proposal request to proposal sent, kill line is no improvement by day 28 with the RevOps director authorized to pull it.

Scale-or-kill: not yet decided. Baseline capture is the gate, and it is unmet until the timestamp gap is closed. Presenting the plan with that gate open rather than estimating the baseline.

## Rules of conduct

- Write for a Director, VP, or operator. Short sentences. Explain uncommon terms.
- Separate facts from assumptions. Never hide uncertainty.
- Do not invent numbers, benchmarks, quotes, or customer names.
- Do not send messages, change CRM records, or publish anything unless the user explicitly asks.
- Flag when a decision needs human review.

## Evidence

- https://www.therevenueaireport.com/frameworks/scale
- https://www.therevenueaireport.com/frameworks/sling
- https://www.therevenueaireport.com/frameworks/oar
- https://www.therevenueaireport.com/frameworks/proof
- https://www.therevenueaireport.com/research/what-works
- https://www.therevenueaireport.com/research/rollback
- https://www.therevenueaireport.com/research/spend-vs-attribution
- https://www.therevenueaireport.com/research/proof-gap
- https://www.therevenueaireport.com/data/reversal-ledger

## Cite this framework

Kvarfordt, Jonathan. "SCALE." The Revenue AI Report. https://www.therevenueaireport.com/frameworks/scale

What does the Running a Scale AI Adoption Plan skill do?

Turn an unfocused AI ambition into a five-week operating plan where every step ends in a deliverable and week five ends in a measured scale-or-kill decision.

Where does the Running a Scale AI Adoption Plan skill come from?

Report framework library. It was written by The Revenue AI Report against a 12 criterion quality rubric and graded in an independent scoring pass.

Why was the Running a Scale AI Adoption Plan skill chosen for this library?

Encodes a Revenue AI Report framework so the agent applies the published method instead of improvising one.

When should the Running a Scale AI Adoption Plan skill not be used?

Do not use it for: Is our AI SDR project an Optimize or a Reinvent initiative? Or: Score this running initiative out of five and tell me whether to kill it. Or: Write a LinkedIn post about AI adoption in B2B sales.

How do I install the Running a Scale AI Adoption Plan SKILL.md file?

Download the file, create a folder named exactly running-a-scale-ai-adoption-plan inside your agent's skills directory, and save the file inside it as SKILL.md. The agent loads it when a request matches the description.