A growth plan for WOWCube: hold CAC at or below the ceiling while unit volume climbs.
Everything in the rest of this plan is one of these three objects. Nothing else gets in.
Seven failure modes. Every one is triggered by spending more, which is why the answer is a standing system and not a one-off fix.
| ID | Disturbance | Mechanism | Trigger | Leading indicator | Absorbed by | Gate |
|---|---|---|---|---|---|---|
| D1 | CAC creep | Each additional unit of spend reaches a less qualified pocket of the audience. | Budget increase | Cost per outbound click rising 3 days running | Creative engine · Site & offer | Pause at ceiling +10% |
| D2 | Channel fatigue | The same audience sees the same offer until it stops registering as new. | Frequency above threshold | Frequency > 2.5 with flat CTR | Funnel variety · Creative engine | Frequency cap at 2.5 |
| D3 | Funnel leak | Traffic arrives and the page, offer or checkout loses it before conversion. | Colder traffic mix | LP view → ATC rate falling | Site & offer · Experiment loop | QA checklist before every ship |
| D4 | Creative decay | A winning ad has a half-life; performance erodes while spend stays flat. | Cumulative impressions | 3-sec hold rate down 20% from peak | Creative engine | Retire below 80% of peak hold |
| D5 | Meta / data | Signal loss, attribution drift and learning resets distort what we think is working. | Platform or pixel change | Platform vs. warehouse variance > 15% | Data spine · Compliance guardrails | No decision on >15% variance |
| D6 | Compliance | Claims fail policy review; delivery stops without warning and spend reallocates badly. | Sharper claim in copy | Any rejection in the last 7 days | Compliance guardrails | Policy checklist pre-submission |
| D7 | AOV ceiling | CAC is only affordable up to what a customer is worth; a flat offer caps the ceiling. | Static offer under load | AOV flat while CAC rises | Site & offer · Creative engine | Contribution margin after spend > 0 |
The leading indicator is what I watch; the gate is what stops it without a meeting. By the time CAC itself moves, the cycle is already lost.
Where a column has no filled cell, that disturbance is unmanaged. The matrix makes the gap visible before the spend does.
| Capability | CAC creep | Channel fatigue | Funnel leak | Creative decay | Meta/data | Compliance | AOV ceiling |
|---|---|---|---|---|---|---|---|
| Creative engine | ■ | ■ | · | ■ | · | · | · |
| Funnel variety | · | ■ | ■ | · | · | · | ■ |
| Site & offer | ■ | · | ■ | · | · | · | ■ |
| Data spine | ■ | · | ■ | · | ■ | · | · |
| Experiment loop | ■ | · | ■ | ■ | · | · | · |
| Compliance guardrails | · | · | · | · | ■ | ■ | · |
Every capability is built the same way: a written standard, an engine that produces against it at volume, and a correction that fires when the data disagrees.
Row three of the matrix, opened up. Every other capability has the same three parts and the same correction.
I run an independent workstream. These five are the only places it touches the rest of the company.
| Who | Owns | Decides alone | What I need from them | Cadence |
|---|---|---|---|---|
| Me — Growth Manager | Acquisition end to end — creative, funnels, site and offer, experiments, reporting. | What gets tested, what gets killed, what ships. | Nothing day to day. That is the point. | Daily update · Friday readout |
| CMO — Sponsor | Budget and the CAC band. | Spend level. Sign-off on claims. | The band agreed in writing, and decisions inside three days. | Monday numbers · Friday Loom |
| PM — Coordination | Jira, the setup projects, access, sequencing. | Who does what, and in what order. | Jira piped into Claude, then unblocking and routing. | Board current, daily |
| Media buyer — Delivery | Account structure, pacing, delivery. | Bid and budget mechanics inside the band. | Kill and scale calls actioned within 24 hours. | Wed launches · Fri decisions |
| Brand — Creative | Assets and brand consistency. | What is on-brand and what is not. | Assets against each brief, one round per week. | Per creative round |
| Dev — Engineering | Warehouse, tracking, site deploys. | Implementation approach. | Ship within the agreed SLA, or tell me it will not. | As tickets land |
Nobody has to ask where anything is. The week has the same shape whether it went well or badly.
| Day | What I do | What it produces | What I need that day | From |
|---|---|---|---|---|
| Mon | Read last week — win rate, CAC, cells resolved. Pick this week's hypotheses. | Monday numbers · this week's tests | Yesterday's spend actioned and reconciled | Media buyer |
| Tue | Write the brief for the chosen cells and run the generation prompt. | The creative brief · first concepts | Assets against last week's brief | Brand |
| Wed | Build or correct one page. Review concepts through the four lenses. | A shipped page · reviewed ads | Deploy inside the SLA | Dev |
| Thu | Call results. Kill, hold or scale — decisions written down. | Results called · a scale recommendation | Kill and scale actioned within 24h | Media buyer |
| Fri | Correct whatever the week disproved. Send the readout. | The written readout · what changed | Nothing — this one is mine | — |
| Daily | CAC pull, guardrail check, live scan, four-line update. | A four-line update · degradation alerts | Blockers raised same day, not Friday | PM |
Every package carries an owner, a predecessor and an exit criterion someone else can verify. Durations need your calendar.
| WBS | Element | Pkgs | Owner | Pred. | Exit criterion |
|---|---|---|---|---|---|
| P0 | The plan | 11 | Me + CMO | — | Objective, ceiling and scope agreed in writing |
| P1 | Workspace setup | 11 | PM | P0 | Nothing blocked on a missing permission or tool |
| P2 | Stakeholder clarity | 10 | Me | P0 | Everyone knows what they own and what I need |
| P3 | Research | 3 | Me | — | Four swipe shelves populated, market read |
| P4 | First launch | 5 | Me | P1, P2 | One site and one ad set live on thin tracking |
| P5 | Customer data build | 7 | Dev | P4 | One dashboard all four people read |
| C1 | Control loop | 5 | Me | P5 | Standing — reconcile weekly, rewrite on disagreement |
| C2 | Ads | 11 | Me + brand | P3 | Standing — one brief, one round per week |
| C3 | Site | 10 | Me + dev | P1 | Standing — one page a week, scan every day |
| C4 | Experiments | 5 | Me | C2, C3 | Standing — cells picked Monday, called Thursday |
| C5 | Keeping it legible | 5 | Me | — | Standing — someone else could run this |
| O1 | Comms | 8 | Me | P2 | Standing — daily update, Monday numbers, Friday readout |
The matrix is the roadmap. Widening scope means adding a column on purpose, with the capability to match it already in place.
Access granted Monday. Ads live Friday. Reading real numbers the Monday after.
The cube · $399
The bundle · $499
Ceilings assume zero LTV — games-ecosystem revenue is upside only. The ceiling is net of any live offer: no discount runs on these boards, so the full $120 stands.
A new page and new ads launched together tell you nothing — you can't tell which one moved. So we change one thing at a time, in both directions, in the same week.
Winners from A and B meet in week two — the better page carrying the better ads. Week one is not judged on sales. It is judged on maximum validated learning per dollar.
Before an ad runs, we write what we expect and what makes us cut it. Three days later the numbers decide — not a meeting.
| # | The ad | The bet | We cut it if… | Status |
|---|---|---|---|---|
| 1 | 01 · Demonstration | Showing the twist (the video) beats any still image. | people stop watching in the first 3 seconds | Live first |
| 2 | 02 · The price | Explaining the $399 price works better than hiding it. | clicks cost twice the average after 3 days | Queued |
| 3 | 09 · Parent | Parents care that nothing scrolls — not that it teaches. | fewer clicks than our average ad | Queued |
| 4 | 05 · The objection | Saying "fidget toy" ourselves works better than dodging it. | the comments turn against us | Queued |
| 5 | 06 · The gift | Gift buyers pick the bundle over the single cube. | a bundle sale costs more than $135 after its 50-purchase clock | Queued |
| 6 | 21 · Durability (new) | "Built to be dropped" beats every parent ad — durability was the real buyer's first question. | hook rate under 25% at 72 hours | Awaiting drop-test claim |
What a winner is worth: a bet wins at CAC ≤ $120 across 50+ purchases. Then its budget steps up 20–30% every 2–3 days under the monthly cap — never doubled, which resets the auction — and the winning angle gets written into the next round of boards. Two clocks on every kill: creative metrics (hook rate, CTR) may kill at 72 hours; CAC may kill only after 50 purchases or 7 days.
Five potential ICPs, built from real signals only — our reviews, the live ad comments, return reasons. Each board is scored by the four who could plausibly meet it in a feed. Where two agree it fails, the board gets edited before a dollar is spent. See the boards on ad concepts.
Ten ads are built and scored. The page is built. The bets are written, each with a kill rule that fires without a meeting. What is missing is access and a first-week budget.