# The 90-Day Sprint — Twelve Weekly Checklists **From Chapter 10 of *The 4-Hour Side Hustle*.** The book gives you five phases. This breaks them into the twelve weeks you actually live through, with the hours each one costs and the two weeks that are harder than the rest. **Total: 68 hours across twelve weeks.** Weeks 1 and 12 are deliberately light. Weeks 8 and 9 are the peak, and that's planned rather than accidental. --- ## PHASE 1 — Validate · Weeks 1–2 · 3.5 hrs/week *Chapters 1 and 2. Spend nothing this phase but $21.* ### Week 1 — 3.5 hrs · buy nothing - [ ] Start the seven-day time audit, four buckets: Create / Sell / Serve / Admin - [ ] Log as you go — reconstructing it on Sunday produces a flattering, useless picture - [ ] Telegram approval channel live (one channel, everything reports here) - [ ] Record 15–20 minutes describing your day job, out loud - [ ] Run the Skill Extraction prompt, both passes — the argue-against pass is not optional - [ ] Score your surviving idea on all five axes. **Must clear 18** > **Buy nothing this week.** The time audit needs Toggl's free tier or a spreadsheet. You don't yet know what you're building. ### Week 2 — 3.5 hrs · $21 total · the demand test - [ ] Build the Build/Run Ledger from last week's audit - [ ] Identify your top three automations by payback ratio — anything under 10× doesn't get built - [ ] Carrd page live in 45 minutes. Set a timer - [ ] Stripe Payment Link created — decide live or test mode **before** publishing - [ ] Plausible or Fathom installed (30-day trial covers this) - [ ] If live: refund reminder in the calendar before the page goes up - [ ] Drive exactly 100 visitors from one source - [ ] Click rate recorded in the ledger - [ ] Every comment and reply saved verbatim, un-tidied > **If the click rate comes back under 3%, that is the plan working.** Go back to Chapter 2.1 and re-score. A week spent here saves a quarter. --- ## PHASE 2 — Offer + Command Center · Weeks 3–4 · 6.5 hrs/week *Chapters 3 and 4. Your recurring bill starts: ~$32/month.* ### Week 3 — 6.5 hrs · the week nobody should skip - [ ] Airtable `Command Center` — four tables plus the Automation Log - [ ] Two-module test scenario proving the plumbing (new Contacts row → Telegram) - [ ] Naming convention decided and written down - [ ] Error handler on every scenario, from the first one - [ ] Claude or ChatGPT Project created, named after the business - [ ] Business Brief written — 500 words, by you, not by AI - [ ] Five writing samples loaded - [ ] Raw customer language from week 2 loaded, unedited - [ ] Tested with a 100-word customer email > The Command Center and the week 1 time audit are the two things that are impossible to retrofit comfortably. Everything after this assumes both exist. ### Week 4 — 6.5 hrs · ship at 80% - [ ] Three offers written on one page — rungs one, two, three - [ ] Rung two built and shipped **at 80%** - [ ] Priced on outcome, at the top of your comfortable range - [ ] Checkout live on Lemon Squeezy or Gumroad - [ ] Plain-language 30-day refund policy on the page and in the receipt - [ ] Six-module refund chain built and tested — budget 40 minutes - [ ] Version 1.1 scheduled in the calendar, three months out - [ ] **Re-run the four-bucket time audit.** Compare against week 1 — first of three --- ## PHASE 3 — The Content Machines · Weeks 5–7 · 6.0 hrs/week *Chapters 5 and 6. Add ~$35/month.* ### Week 5 — 6.0 hrs - [ ] Weekly 90-minute recording block in the calendar, permanently - [ ] 20 topics in the Airtable Content table - [ ] `/raw-video/` folder created - [ ] First recording session done - [ ] Clip chain modules 1–4, built and tested one at a time > Build one module at a time and test after each. A nine-module chain assembled in one sitting fails in a way that takes longer to diagnose than building it slowly would have taken. ### Week 6 — schedule this one lighter than the plan says - [ ] Clip chain modules 5–9, still one at a time - [ ] Four platform caption prompts saved in your Project - [ ] `WEAK CLIP` router in place before the scheduling module - [ ] Buffer posting schedule pre-set — two channels, not four - [ ] First clips going out > **Week 6 is the morale trough.** Moderate hours, nothing has paid off, you've built a product nobody has bought and a chain that hasn't published. Everyone hits this. It is not a signal about you, your niche, or your content — platforms don't distribute to accounts they haven't characterized yet, and there is no version of this where week 6 is exciting. ### Week 7 — 6.0 hrs · bank the recordings - [ ] Domain and site live - [ ] Search Console connected - [ ] Transcript-to-article chain running - [ ] First article published - [ ] `[NEEDS AUTHOR INPUT]` routing to Needs Edit, stopping there - [ ] **Three recording sessions banked** — you will need them in weeks 8 and 9 --- ## PHASE 4 — Capture and Conversion · Weeks 8–9 · 8.5 hrs/week *Chapters 7 and 8. The peak. Add ~$15/month.* ### Week 8 — 8.5 hrs - [ ] Instagram switched to Professional, Facebook Page connected — **do this first, on your phone** - [ ] One keyword flow live and tested from someone else's account - [ ] DM sequence written to give before it asks - [ ] Public reply as well as DM, worded differently each time - [ ] Kit tag records the specific video, not just "instagram" - [ ] Monthly Kit export to cloud storage automated - [ ] Lead magnet built — interactive, not a PDF - [ ] **Re-run the four-bucket time audit.** Compare against weeks 1 and 4 — second of three ### Week 9 — 8.5 hrs · the heaviest week - [ ] 20 verbatim customer sentences collected, typos left in - [ ] Nine-block sales page live, under 700 words - [ ] One AI call per block — never the whole page in one pass - [ ] Five-email welcome sequence written in one sitting - [ ] Link triggers and branching configured in Kit - [ ] Order bump added at checkout - [ ] Post-purchase chain built and tested **with a real card** > **Weeks 8 and 9 are the workload peak** — 10.3 hours in the heaviest week, against a 68-hour total. Lean on the three recording sessions you banked in week 7. If something has to give, give up the Saturday build block, not the recording. --- ## PHASE 5 — Delivery, and the Switch to Run Mode · Weeks 10–12 *Chapter 9, then Chapter 11. 3.0 hrs/week in weeks 10 and 11. None in week 12.* ### Week 10 — 3.0 hrs - [ ] Single first action defined, under 15 minutes - [ ] Delivery email hand-written — not templated - [ ] Three-minute orientation video recorded - [ ] Day-2 nudge automated - [ ] `First Action Completed` tracked in the Command Center ### Week 11 — 3.0 hrs - [ ] 30 days of support questions logged and reviewed - [ ] FAQ built from the log, not from imagination - [ ] Answers placed **inside the product**, where the question arises - [ ] Support hours published - [ ] Two 20-minute support windows in the calendar ### Week 12 — no building at all - [ ] Full Chapter 11 weekly rhythm running for one complete week - [ ] Five blocks, three hours fifty-five minutes - [ ] Sunday review run from the script - [ ] **Re-run the four-bucket time audit** — third of three. Compare all four readings - [ ] Four alerts configured, and no fifth - [ ] Quarterly audit scheduled in the calendar > **The temptation in week 12 is to keep building.** Don't. The switch to run mode is a real transition, and Chapter 11 exists because people fail it. A machine you keep extending is a machine you never actually run. --- ## Starting mid-plan You don't always start at week 1. | Where you are | Start at | Note | |---|---|---| | Product exists, no content machine | **Week 5** | Do week 1's audit anyway — it costs no extra hours | | Content exists, no offer | **Week 3** | Read Chapter 10's "error one" twice | | Both exist, nothing automated | **Week 3** | Expect eight weeks rather than twelve | **What nobody should skip:** the time audit in week 1 and the Command Center in week 3. Both are cheap, both are impossible to retrofit comfortably, and both are what the remaining weeks assume you have. --- ## What each week costs | Week | Hours | Running spend | |---|---|---| | 1 | 3.5 | $0 | | 2 | 3.5 | $21 one-off | | 3 | 6.5 | ~$32/mo begins | | 4 | 6.5 | +$0 | | 5 | 6.0 | +~$35/mo | | 6 | 6.0 | — | | 7 | 6.0 | — | | 8 | 8.5 | +~$15/mo | | 9 | 8.5 | — | | 10 | 3.0 | — | | 11 | 3.0 | — | | 12 | ~4.0 (running, not building) | — | --- ## The three time audits **Weeks 4, 8 and 12.** Same four buckets as week 1, same seven days of logging, compared against your week-one numbers each time. This is the only way to see the thing you are actually building. Revenue lags the machines by months — hours don't. A week-8 audit showing Admin down from 6 hours to 2 is real progress in a month where nothing has sold yet, and without the audit you would have no evidence of it at all. Each re-run takes ten minutes rather than an hour, because the template already exists and you already know the four buckets. --- ## The two hard weeks, and they're not the same kind of hard **Week 6 is emotionally hard.** Moderate hours, no payoff yet. Schedule it lighter than the plan says. **Weeks 8–9 are physically hard.** The hours peak. That's why weeks 1 and 12 are deliberately light — the sprint is shaped around this, not surprised by it. Knowing which is coming, and when, is most of what gets people through both.