The Daily Operating Rhythm — the whole business as three numbers

July 31, 2026 · Companion to the 100-day plan. One job = one number = one daily block = one screen.

How to read this page. This is how Akash’s working day and week are structured so that the 100-day plan actually happens. The idea in one line: the business has three jobs, each job is measured by one leading number, and each number gets one protected block of the day. Everything else is done by the AI agents — or doesn’t get done. Sections have “More context” expanders; any dotted underlined word opens an explanation in a side card.

The move

The day becomes three blocks — one per job — each existing only to move its number. Everything outside the blocks is delegated or parked.

The filter

One question kills all distraction: “Which of the three leading numbers does this move, by how much, this week?” No answer = parked.

The screens

Six screens the business already owns. Nothing new gets built — three things get retired (see the red box below).

The three jobs, and the number each one watches

Marketing

Content pieces shipped per day

more content published → more homeowners and contractors reached → more leads

Results it drives (measured later): leads generated per city · cost per qualified lead at or under $25

Sales

New conversations started per day

more conversations with contractors → more clients signed

Results it drives: free chatbot installs · new client partnerships signed per week

Delivery

Speed of closing open items

leads handed over faster, quotes chased harder → more cash collected

Results it drives: time from hand-off to quote to job · cash collected from clients’ won jobs

More context — why watch these three instead of watching cash

Cash arriving is a result — by the time it’s visible, the work that caused it happened 2–6 weeks ago. You can’t manage a result directly; you can only manage the activities that produce it. These three numbers (content shipped, conversations started, close speed) are the activities. If all three stay on target, the cash line on the 100-day plan follows with a delay. That’s also why the scariest instinct — “cash is low, watch cash harder” — is explicitly banned by rule 5 below.

The day — three blocks; everything between them belongs to the AI

When	Block	Who	The work (and only this)
5–8 AM	Marketing — maker block	Akash	Record, edit, publish — nothing else (standing rule: no admin, no messages). Ship at least one proof-based piece a day: real lead receipts, real job footage, the chatbot booking a real job. The content is the proof that sells clients, so publishing is itself the lead-generation work.
~8:15	Scoreboard read	agents	The automated morning report lands: the six numbers (three activity, three result), yesterday’s leads per city, and any hand-off stuck longer than an hour. Akash reads it — he doesn’t build it.
9–12	Sales — conversation block	Akash	Ten new conversations from the researched prospect shortlist (calls and messages, hand-run). Offer one free chatbot install a day. Move every open thread forward one step. Anything an AI can draft first, it drafts first.
1–2 PM	Delivery — speed block	Akash	Get the hand-off queue to zero (a lead older than 72 hours is dead inventory). Chase every open quote toward a close with the client. Log which closed jobs came from our leads. 30–60 minutes; the automated pipeline runs itself the rest of the day.
All day	Lead pipeline + ad operations + drafts	agents	The bot qualifies homeowners and hands them off instantly · ad rotations run from the approved library · reply drafts are staged for review · Akash is interrupted only when one of the three activity numbers is at risk.
~4 PM	Close of day (15 min max)	Akash + agents	Quick sweep: today’s three numbers vs. target, one voice note, tomorrow pre-staged. Standing rule: after 3 PM, no new multi-hour projects get started.

The week — four touches, about two hours total

Day	Ritual	What gets decided
Monday (30 min)	Scorecard + re-rank	Compare predicted vs. actual on the six numbers · name THE one bottleneck of the week in a single sentence · set the week’s three activity targets (for example: 7 pieces published, 50 conversations, every hand-off under an hour) · re-rank the project board — anything that doesn’t feed one of the three numbers is parked.
Wednesday (10 min)	Pace check	On pace or not, per number. Behind on one? Tomorrow’s block for that job gets bigger — and nothing else changes mid-week.
Friday (30 min)	Client review	Per client: leads sent → quotes → jobs won → cash. Coach or part ways with the weakest performer (their close rate is our revenue). Reconcile job-tracking: every closed job either traced to its lead or flagged.
Sunday (automatic)	Projections	The agents post next week’s predicted numbers; Akash corrects by voice note or accepts.

The six screens — all existing, nothing new

Screen	What it owns	Which number
Weekly scoreboard	THE view: the three activity numbers and three result numbers, predicted vs. actual — and nothing else on it	all six
Project board (Linear)	Projects and tasks — every card must name the number it moves, or it’s parked; the agents log decisions here	all (as a filter)
The CRM	Sales conversations and delivery follow-ups; the sales pipeline doubles as the delivery tracker; an AI drafts every reply	conversations · close speed
Lead ledger	The inventory book: leads generated → sold/delivered → closed → gone stale, per trade and city; feeds the proof pack automatically	leads · cash
Job-tracking wire (per client)	Each client’s closed jobs flowing back automatically, plus the monthly 10% statement; wired at onboarding — no wire, no client	cash collected
Agent board + memory	The AI team’s task dispatch and session logs — how work continues when Akash isn’t at the desk	(infrastructure)

Retired by this document: the sprawl of 27 reporting pages (being cut to 8 — already on the Execution Plan) · contractor ads built on claims instead of proof · and any new dashboard, pipeline, or tool that can’t name which of the three numbers it moves. During the cash sprint, the default answer to “should we build X?” is: “park it — and say which daily block it would have displaced.”

The five rules that keep this simple

One number per job. If a metric isn’t one of the six, the agents may track it, but it never reaches Akash’s eyes uninvited.
The day is only the three blocks. Content, conversations, close speed. A day that produces none of the three isn’t a plan — say so at the morning read.
Delegate numbers, not task lists. Agents (and later, hires) own a number and report predicted vs. actual. Nobody gets handed a to-do list.
The bottleneck earns the extra hours. Monday names one bottleneck; free time outside the blocks goes only there. Building tools counts only when tooling IS the bottleneck.
Activity beats results in every argument. When cash is scary, the instinct is to stare at cash harder — wrong. Ship the piece, make the calls, clear the queue; the cash follows 2–6 weeks later by design.

The one-question filter (print it): “Which of the three leading numbers does this move, by how much, this week?” Every new idea, tool, request, and opportunity answers it — or gets parked. Including ideas from the AI.

Technical notes — the doctrines this consolidates

Daily-three doctrine · maker-block rule · two-pipeline delegation model (people vs. project; delegate numbers only) · ruthless-filter lane standard · wide-matrix ad testing (Meta’s budget allocation is the signal) · 15-stage sales pipeline (migrated Jul-20) · scorecard predicted-vs-actual rail (weekly-scorecard.html). Built Jul-31.

Companions: First $100K in 100 days (what this rhythm is executing) · Execution Plan — July 31 (today’s list) · The Unblock List (today’s hands-on tasks).