Improve a Business

An interactive, resumable journey for a business that already runs but has stopped moving — real revenue, real customers, a flat line. Across eight phases it works the layers in order (strategy, execution, management, structure, motivation, metrics, margin, pricing), diagnosing before treating rather than chasing one tactic at a time. The agent asks before every decision and records each choice in docs/ so the turnaround survives across sessions.

Core Principle

A plateau is a diagnosis failure: diagnose before treating, and work the layers in order — strategy, execution, management, structure, motivation, metrics, margin, pricing. Poking one tactic at a time never breaks a plateau because the constraints compound. This skill sequences the phases, asks the decision questions, and records what was decided; the constituent skills carry the actual method — invoke them rather than improvising their frameworks.

Journey Map

Phase	Skill	Question it answers	Artifact
1	good-strategy-bad-strategy	What single challenge is capping the business?	Creates docs/STRATEGY.md — GATE
2	traction-eos	How do we execute the strategy every week?	Creates docs/OPERATIONS.md
3	high-output-management	How does the operator multiply the team, not bottleneck it?	Extends docs/OPERATIONS.md
4	team-topologies	Is the org structure itself the constraint?	Extends docs/OPERATIONS.md
5	drive-motivation	Why has the team quietly checked out?	Extends docs/OPERATIONS.md
6	lean-analytics	Which one number tells us the truth?	Creates docs/METRICS.md
7	negotiation	Where are we leaking margin, and what do customers truly value?	Extends docs/CUSTOMER.md
8	monetizing-innovation	Are we charging for the value we deliver?	Extends docs/OFFER.md

Operating Rules

1. Resume first. Before anything else, read docs/IMPROVE-BUSINESS-PLAN.md and every artifact in the Journey Map. If the tracker exists, summarize the journey state in 3-5 lines and ask which phase to enter. Done when the user has confirmed an entry point. A journey with a tracker is resumed, never restarted.
2. Intake on first run only. No tracker: run the Intake below, then create docs/IMPROVE-BUSINESS-PLAN.md with every phase statused pending | in-progress | awaiting-evidence | done | deferred: reason | skipped: reason. Done when the tracker exists and the user has confirmed the phase plan.
3. Phase entry. Announce: what the phase does, the decision it forces, the artifact it produces, rough effort. Offer proceed / skip / defer — phases marked GATE may be deferred, never skipped. Mark the phase in-progress on proceed. Done when the user chose.
4. Skill invocation and fallback. Invoke the phase's skill by its slug. If it is not available, offer: npx skills add wondelai/skills/<slug> --global. If the user declines, run the phase from its Brief — the minimum viable method. State which mode you are in.
5. In-phase decisions. Ask every question under "Decide with the user" — with concrete options and your recommendation. Record the choice in the tracker's Key Decisions. A decision made silently is a defect.
6. Phase exit. Present the draft artifact content for sign-off before writing. On approval: write or extend the docs/ files, update the tracker (status, Key Decisions, Next Actions). Done when the files are written and the phase row shows done.
7. Artifact discipline. Read before writing; create a file only if missing, otherwise extend — add or update your sections, preserve everyone else's. Files are UPPERCASE in docs/. Every recommendation lands as a checkbox or a table row with owner and priority. See references/artifact-templates.md when creating a docs/ file for the first time — create it from the full skeleton (all section headings), then fill the sections your phase names.
8. Diagnosis before treatment. No new initiative, reorg, or pricing change starts before docs/STRATEGY.md holds a written Diagnosis the user has confirmed. When asked to jump straight to Phase 4, 7, or 8, complete Phase 1 first (or confirm it is already done and recorded). Done when a confirmed Diagnosis exists before any treatment phase writes its artifact.

Intake

Ask these before creating the tracker:

1. What is the plateau — flat revenue, rising churn, a team working hard with nothing moving? (Frames the Phase 1 diagnosis and the Phase 6 OMTM.)
2. What passes for your current strategy today — a deck, OKRs, a goal list, nothing written? (Feeds Phase 1; if none exists the diagnosis work is larger.)
3. How big is the team and how is it structured? (Sizes the Rocks and 1:1 system in Phases 2-3; gates Phase 4.)
4. Do you run technology teams, or is this a services / brick-and-mortar business? (team-topologies is engineering-specific — skip Phase 4 for non-tech orgs.)
5. When was pricing last set, and on what basis — cost-plus, competitor, or value? (Gates Phase 8; cost/competitor pricing signals margin left on the table.)
6. What is your sales win rate, and do you hear price objections? (Near-100% win rate with no pushback signals minivation — Phase 8.)
7. Where does work get stuck — within teams or between them? (Between-team stalls point at Phase 4; within-team points at Phases 3 and 5.)

Phase-skip heuristics: skip Phase 4 when you don't run technology teams or teams already deliver cleanly; skip Phase 5 when morale is engaged and not the constraint; skip Phase 7 when no large renewals or vendor contracts are in play; skip Phase 8 only if pricing was recently repriced from validated willingness to pay.

Then create docs/IMPROVE-BUSINESS-PLAN.md from the template and confirm the phase plan. Done when docs/IMPROVE-BUSINESS-PLAN.md exists with every phase statused and the user has confirmed the plan.

Phases

Phase 1 — Diagnose the real challenge (good-strategy-bad-strategy) — GATE

Purpose: Name the single critical challenge capping the business, so every later layer works the right constraint. Phases 2-8 may not start before this Diagnosis is confirmed.

Brief (fallback): Good strategy has a kernel — an honest diagnosis of the critical challenge, a guiding policy, and coherent actions. Bad strategy is fluff, goals dressed as strategy, and forty "priorities" that concentrate nothing. Treat the funnel as a chain-link system and find the one binding link; investing in strong links is wasted until the weak one is fixed. Set a proximate objective close enough that the team can see how to hit it, and force a no-list.

Invoke: good-strategy-bad-strategy with the current annual plan, OKRs, or deck pasted in. Ask it to (a) audit the plan as strategy and score it 0-10, (b) write a proper kernel, (c) name the binding chain-link constraint and one proximate objective for the quarter.

Decide with the user: Does the diagnosis name the real obstacle, not a symptom? Which single link (lead-gen / activation / retention / expansion) is the binding constraint, and what is the proximate objective? What goes on the no-list — what will you deliberately stop doing?

Artifact: Create docs/STRATEGY.md with ## Diagnosis (critical challenge + binding constraint), ## Guiding Policy (including the proximate objective), ## Coherent Actions (Action | Owner | Due | Status), and ## No-List. Update the tracker.

Done when: STRATEGY.md holds a confirmed Diagnosis, a Guiding Policy, 3-5 Coherent Actions with owners, and a No-List — only then are Phases 2-8 unlocked.

Phase 2 — Install an execution rhythm (traction-eos)

Purpose: Convert the strategy into a weekly operating rhythm that survives Monday.

Brief (fallback): Vision without traction is hallucination. Set 3-7 quarterly Rocks — each with one owner, SMART, binary (done or not, no partial credit) — aimed at the proximate objective. Install the Level 10 Meeting: same day and time, 90 minutes, fixed agenda ending in 60 minutes of IDS (Identify, Discuss, Solve) driven to root cause. Never cancel it.

Invoke: traction-eos with STRATEGY.md's guiding policy and proximate objective. Ask for (a) 5-7 quarterly Rocks with owners and binary criteria, flagging any that are business-as-usual, (b) a Level 10 agenda with time boxes, (c) an IDS pass on the top recurring issue.

Decide with the user: Which 5-7 Rocks make the quarter (cap at seven)? Who owns each? What weekly cadence for the Level 10, and who facilitates?

Artifact: Create docs/OPERATIONS.md with ## Vision/Traction Summary, ## Rocks (this quarter) (Rock | Owner | Binary done-condition | Status), and ## Meeting Cadence (Meeting | Rhythm | Agenda). Update the tracker.

Done when: OPERATIONS.md holds 5-7 owned binary Rocks tied to the proximate objective and a scheduled Level 10 cadence the user has committed to.

Phase 3 — Manage for output, not activity (high-output-management)

Purpose: Multiply the team's output by getting the operator off the critical path.

Brief (fallback): A manager's output equals the output of their organization plus the neighboring teams they influence. Negative-leverage habits — meddling, waffling, being the approval bottleneck — multiply downward. High-leverage tools: subordinate-owned 1:1s at a cadence set by task-relevant maturity (TRM), and delegation-with-monitoring — delegate what you know best, monitor at task level, sample deeper when TRM is low, lighter as it rises.

Invoke: high-output-management with the operator's last two weeks of calendar and the tasks they keep grabbing back. Ask for a leverage audit, a redesigned week built around forecasted key events, and a 1:1 plus delegation system for direct reports.

Decide with the user: Which negative-leverage habits to cut? The 1:1 cadence per report (set by TRM, not seniority)? Which tasks to delegate-with-monitoring off the critical path?

Artifact: Extend docs/OPERATIONS.md: add ## Management Leverage (1:1 cadence per report, delegation decisions, monitoring depth). Update the tracker.

Done when: OPERATIONS.md records a 1:1 cadence for every direct report and at least one delegation decision that removes the operator as a bottleneck.

Phase 4 — Reshape teams around value streams (team-topologies)

Purpose: Remove structural friction when work keeps getting stuck between teams rather than within them.

Brief (fallback): Conway's law — you ship your communication structure. Reduce every team to one of four types: stream-aligned (the default), enabling, complicated-subsystem, platform. Cap cognitive load: at most one complicated domain per team; never split a single domain across two teams. Shed domains or remove extraneous load before adding headcount. Define each interaction mode: collaboration, X-as-a-service, or facilitating.

Invoke: team-topologies with each current team and what it does day to day. Ask it to classify teams, flag anti-patterns (shared-services ticket queue, teams split by function not value stream), diagnose cognitive load, and propose value-stream-aligned boundaries with explicit interaction modes.

Decide with the user: First — is the plateau structural (handoffs dominate lead time, recurring coordination meetings, everyone's-and-no-one's ownership)? If not, skip. If yes: which domains to shed, which extraneous load to platform-ize.

Artifact: Extend docs/OPERATIONS.md: add ## Team Structure (team topology, per-team type, cognitive-load notes, interaction modes). Update the tracker.

Done when: either the phase is skipped with a recorded reason, or OPERATIONS.md holds a four-type classification and a proposed value-stream structure with interaction modes.

Phase 5 — Re-motivate the people who checked out (drive-motivation)

Purpose: Rebuild intrinsic motivation instead of throwing money at disengagement.

Brief (fallback): For any task needing cognitive effort, if-then rewards degrade performance — they crowd out intrinsic motivation, narrow focus, foster short-term thinking. What sustains it is Autonomy (task, time, technique, team), Mastery (visible progress at a challenge between boredom and anxiety), and Purpose (serving something larger than the quarter). Pay enough to take money off the table, then stop using it as the lever; replace if-then bonuses with now-that recognition.

Invoke: drive-motivation with how incentives, reviews, and goals currently run. Ask for an AMP audit scored 0-10, the places controlling if-then rewards crowd out motivation, and specific autonomy / mastery / purpose fixes including a shift to now-that recognition.

Decide with the user: Which if-then rewards to retire? Which autonomy levers (the four T's) to grant? How to convert recognition from if-then to now-that?

Artifact: Extend docs/OPERATIONS.md: add ## Motivation (Autonomy / Mastery / Purpose) (Lever | Finding | Fix). Update the tracker.

Done when: OPERATIONS.md holds an AMP audit with at least one concrete fix per pillar, and the decision on if-then versus now-that rewards is recorded.

Phase 6 — Measure the one number that tells the truth (lean-analytics)

Purpose: Replace a dashboard of vanity metrics with one honest number tied to the riskiest assumption.

Brief (fallback): Good metrics are comparative, understandable, ratios or rates (not ever-growing totals), and behavior-changing — if a number won't change what you do next, stop watching it. The cumulative up-and-to-the-right chart is the top vanity tell. Pick one OMTM for the riskiest part of the business right now, guard it with a counter-metric so it can't be gamed, and draw a line in the sand: a target, a date, and a pre-committed miss response decided before results arrive.

Invoke: lean-analytics with the current dashboard and the business model and stage. Ask it to purge vanity metrics into actionable ratios, name the OMTM plus counter-metric, and build cohort retention tables to test the stickiness gate.

Decide with the user: What business model and stage are we in? Which single OMTM and counter-metric? The line in the sand — target, date, and pre-committed action if we miss?

Artifact: Create docs/METRICS.md with ## Stage & One Metric That Matters, ## KPI Definitions (Metric | Definition | Actionable ratio? | Owner), ## Baselines & Targets (including the line in the sand), ## Funnel, and ## Cohort Notes. Update the tracker.

Done when: METRICS.md names one OMTM with a counter-metric and a line in the sand, and the vanity metrics are rewritten as actionable ratios.

Phase 7 — Stop leaking margin in negotiations (negotiation)

Purpose: Protect price in renewals and vendor contracts, and capture what customers truly value.

Brief (fallback): The path to a good deal runs through making the other side feel understood; no deal beats a bad deal. Tools: the accusation audit (preemptively voice their objections), calibrated questions ("how am I supposed to do that?"), labeling ("it sounds like…"), and the Ackerman plan (decreasing increments ending on a precise non-round number plus a non-monetary sweetener). Hunt the Black Swans — secret constraints, hidden motivations, unknown context — behind any competing bid; it is rarely the real story.

Invoke: negotiation with the specific renewal or vendor contract (numbers, the threat, your instinct). Ask for an accusation audit, five calibrated questions to surface the Black Swan, and an Ackerman plan with a non-monetary concession — never concede without getting something back.

Decide with the user: For each big negotiation: the target and walk-away, the concessions you will trade (never for free), and the real driver (Black Swan) behind the price threat.

Artifact: Extend docs/CUSTOMER.md: add to ## Competing Alternatives (the competing bids customers cite and their real weaknesses) and ## Interview Evidence (concrete facts and Black Swans surfaced in the renewal — what the customer actually values). Update the tracker.

Done when: the negotiation is prepped with an accusation audit, calibrated questions, and an Ackerman plan, and CUSTOMER.md records the real drivers surfaced.

Phase 8 — Reprice around delivered value (monetizing-innovation)

Purpose: Recover margin by repricing from validated willingness to pay, not cost or competitors.

Brief (fallback): Price is a diagnostic of value. Name the monetization failure: feature shock, minivation (right product priced too timidly — signature: near-100% win rate, no price pushback), hidden gem, or undead. Segment by willingness to pay, not demographics. Class every feature leader / filler / killer; never give the leader away in the cheapest tier or bundle a killer that makes buyers reject the package. Validate with acceptable / expensive / prohibitive probes before shipping any change.

Invoke: monetizing-innovation with current pricing, tiers, win rates, and deal data, plus the WTP signals from Phase 7. Ask it to diagnose the monetization failure, run leader-filler-killer on the tiers, and design a WTP conversation to run before any price change ships.

Decide with the user: Which monetization failure fits the evidence? The tier restructure (which tier you actually want to sell, the premium anchor)? Whether to run WTP conversations before repricing.

Artifact: Extend docs/OFFER.md: add ## Offer Stack, ## Willingness-to-Pay Evidence (Segment | Acceptable | Expensive | Prohibitive | Source), ## Leader / Filler / Killer Features, ## Tiers (Good / Better / Best), and ## Price Metric. Update the tracker.

Done when: OFFER.md names the monetization failure, a leader-filler-killer classification, and a repriced tier structure backed by WTP evidence (or a scheduled WTP conversation).

Optional Phases

Skill	Add when	Artifact
blue-ocean-strategy	the market is commoditized and competition is on price	Extends docs/STRATEGY.md (## Strategy Canvas & ERRC Grid)
hundred-million-offers	the offer itself is weak, not just the pricing	Extends docs/OFFER.md (## Offer Stack)
jobs-to-be-done	customers churn and nobody knows which job the product lost	Extends docs/CUSTOMER.md (## Job Statement, ## Job Dimensions)
obviously-awesome	the market misunderstands what the business is	Extends docs/POSITIONING.md
37signals-way	the org overplans and underships	Extends docs/OPERATIONS.md

Optional phases follow the same operating rules; insert where the Add-when condition first becomes true.

Common Mistakes

Mistake	Fix
Jumping to tactics before the diagnosis	Complete Phase 1 first — the chain-link constraint tells you which single link to work; effort on strong links is wasted.
Treating the strategy offsite as the finish line	Install the Level 10 cadence (Phase 2) and never cancel it — the weekly rhythm is what changes behavior.
Confusing being busy with creating output	Apply Grove's equation (Phase 3): delegate what you know best, monitor by TRM, and get off the critical path.
Reorganizing when the problem isn't structure	Run Phase 4 only when handoffs dominate lead time; if teams already deliver cleanly, skip it.
Throwing money at disengagement	Pay fairly, then rebuild autonomy, mastery, and purpose (Phase 5); replace if-then bonuses with now-that recognition.
Repricing from cost or competitors, not value	A high win rate with no price pushback signals undercharging — reprice from validated WTP and segment by it (Phase 8).

Completing the Journey

Exit checklist:

• docs/STRATEGY.md holds a confirmed Diagnosis, Guiding Policy, Coherent Actions, and No-List
• docs/OPERATIONS.md holds owned Rocks, a Level 10 cadence, management leverage, and (where structural) team structure and motivation fixes
• docs/METRICS.md names one OMTM, a counter-metric, and a line in the sand
• docs/OFFER.md is repriced from validated WTP, and docs/CUSTOMER.md records the real drivers surfaced in renewals

Close the tracker: every phase done or skipped: reason, and carry each open Next Action into the relevant artifact as a checkbox or table row with an owner. Then route forward:

• When the fundamentals work again and the goal shifts to expansion, continue with grow-business.
• When the product itself is what drags the business down, continue with improve-app.