The Weekly Rhythm: Four Real Weeks

Lesson 3 logged a real Week 1: 55 hours, 8 automated, 3 clients, revenue ratio 1.0, 10% recurring. Lesson 4 explained what the five numbers that came back actually mean. Neither lesson told you what it feels like to do this every week for a while, because there was only ever one week of data to look at. This lesson fixes that. Three more real weeks, logged against that same Week 1 baseline, run through the same script, on the same project.

The rhythm itself is almost boring, and that's the point. At the end of a work week, open Claude Code, log the six numbers, let the script run, glance at what came back. Repeat. The interesting part isn't the ritual. It's what four rows next to each other tell you that one row never could.

Week 2: the first real comparison point

Week 1 had nothing to compare against. Week 2 is the first week where every metric except Automation Index actually has to compute a real difference against something.

python3 .claude/skills/transformation-tracker/scripts/log_week.py \
  --week 2 --total-hours 52 --automated-hours 15 --active-clients 3 \
  --revenue-ratio 1.05 --recurring-pct 10.0 \
  --automated-this-week "Invoicing" --bottleneck "Onboarding still manual"

Total hours dropped from 55 to 52, automated hours nearly doubled from 8 to 15, and invoicing moved off the operator's plate. The bottleneck line still names onboarding, which was already the Week 1 bottleneck. That repetition is data too: automating invoicing did not touch the thing that was actually slow.

Automation Index nearly doubled, from 14.55% to 28.85%, because automated hours nearly doubled. Time Liberation Score shows 3.0 hours/week back, the plain difference between Week 1's 55 and Week 2's 52. Revenue Efficiency Multiple and Client Capacity Score both moved, but only barely: 1.11x and 1.06x. Recurring stayed flat at 10.0%, because nothing about this week changed how the operator was billing clients, only how many hours it took to serve them.

Week 3: the bottleneck line changes

python3 .claude/skills/transformation-tracker/scripts/log_week.py \
  --week 3 --total-hours 48 --automated-hours 22 --active-clients 4 \
  --revenue-ratio 1.15 --recurring-pct 15.0 \
  --automated-this-week "Onboarding checklist" --bottleneck "Reporting"

Onboarding, the thing that sat unresolved through Weeks 1 and 2, finally got automated in Week 3. Active clients moved from 3 to 4 in the same week that onboarding stopped being manual. That's not a coincidence worth ignoring: onboarding was the bottleneck standing between "clients I have" and "clients I could take on," and clearing it is exactly the week a new client showed up.

Client Capacity Score jumps to 1.53x, the largest single-week move of any metric across all four weeks logged so far. Revenue Efficiency Multiple climbs to 1.32x. The bottleneck line now names reporting instead of onboarding, which tells you where to look next before you've even opened a dashboard.

Week 4: automation nearly quadruples, and the other metrics don't keep pace

python3 .claude/skills/transformation-tracker/scripts/log_week.py \
  --week 4 --total-hours 43 --automated-hours 25 --active-clients 4 \
  --revenue-ratio 1.25 --recurring-pct 25.0 \
  --automated-this-week "Weekly report draft" --bottleneck "Sales follow-up"

Reporting, last week's named bottleneck, is automated this week. Recurring revenue jumps from 15.0% to 25.0%, the largest single-week move in that column across the whole run.

Reading four rows at once

Here's the full table, straight from report.py --table run against these four real weeks:

Read down the Automation Index column and it looks like the whole story: 14.6, 28.9, 45.8, 58.1, a number that climbs every single week without hesitation. From Week 1 to Week 4 that's a little under 4x the starting figure. If you stopped reading at that column, you'd conclude the other four metrics are riding along at roughly the same pace.

They aren't. Over the same four weeks, Revenue Efficiency Multiple went from 1.00x to 1.60x, and Client Capacity Score went from 1.00x to 1.71x. Both real, both climbing, neither anywhere close to automation's near-4x. Automating hours does not automatically turn into proportionally more revenue per hour or proportionally more client capacity. Client Capacity Score actually outpaced Revenue Efficiency Multiple across these four weeks, 1.71x against 1.60x, which is worth noticing on its own: this run absorbed more client-serving capacity per hour worked slightly faster than it converted that capacity into revenue per hour. Automating a task frees the hour. What that freed hour gets used for, and how well, is a separate question the Automation Index was never built to answer.

The bottleneck line is doing more work than it looks like

Look back at the one-sentence bottleneck answers across the four weeks: onboarding, onboarding again, reporting, sales follow-up. Every week names a different problem than the metrics alone would surface, and every week's named bottleneck gets solved by the following week's automated task, except the last one. Sales follow-up is still open at the end of Week 4. That's not a gap in the tracker. It's a real open problem the numbers haven't caught up to yet, and it's exactly the kind of thing a reader should expect to still be sitting there when a 12-week run is only a third done.

The rhythm, stated plainly

Log at the end of the week. Run the report. Read the whole table, not just the column that's climbing fastest. Reread the bottleneck sentence before you decide what to automate next, because it's naming the actual constraint, not a formula's best guess at one. That's the entire loop. Nine more weeks of this course are just this same loop, run nine more times, which Lesson 9 covers directly. What changes between now and then isn't the ritual. It's what you notice in the table once there's enough of it to disagree with itself.