Focus Journal and Attention Experiments

By Better Wise

You already have the pieces: a protected block of time, a phone that stays quiet, and breaks that actually restore you. This lesson adds the feedback loop. A focus journal tells you whether those pieces are working, and a small experiment tells you which change deserves to stay in your routine.

What You Will Practice

• Keep a focus journal entry that takes under 60 seconds to write.
• Pick one output metric that reflects real work, not just time spent.
• Run a before-and-after experiment with a prediction and a decision rule written in advance.
• Compare five sessions per condition and decide what to keep.

A Worked Example: One Writer, Two Weeks

Maya writes documentation. She suspects her phone is the problem, but she has no evidence. Here is the loop she runs.

flowchart TD
  A["📋<br/>Baseline week"] --> B["✍️<br/>Predict + rule"]
  B --> C["🔧<br/>Change one thing"]
  C --> D["📓<br/>Log 5 sessions"]
  D --> E["📊<br/>Compare medians"]
  E --> F{"Meets rule?"}
  F -->|Yes| G["✅<br/>Keep it"]
  F -->|No| H["🔁<br/>Drop, test next"]

Step 1, baseline. Maya runs five 50-minute sessions exactly as she normally works. After each one she writes: the one deliverable she intended, how many times she got pulled away, how long it took to get back on task, and the word count of finished sections.

Step 2, predict. She writes: "If the phone is in another room with notifications off, median words per session rises by at least 15 percent, and my end-of-session energy stays at 3 or higher." She writes this before week two starts, so she cannot move the goalposts later.

Step 3, change one thing. Only the phone location and notification state change. Same block length, same time of day, same task type.

Step 4, log five sessions.

Step 5, compare and decide. 760 is up 23 percent, above her 15 percent threshold, and energy held. She keeps the change. Next experiment: one variable only, maybe starting blocks with a written first sentence.

_____ should change between the baseline and the test week, otherwise you cannot tell what moved the number.

Your Turn: Build Your Own Experiment

Exercise 1, pick your output metric. Choose a count that ends the session with something finished: words drafted, problems solved, tickets closed, tests written, slides built. Write your metric and how you will count it in one line.

Exercise 2, write your prediction and decision rule. Format: "If I [single change], then [metric] rises by at least [number], and [guardrail] holds." The guardrail protects against winning the metric while losing your energy or quality.

Exercise 3, log two real sessions. Use these fields: date and block length, intended deliverable, distractions, resumption time, metric value, energy 1-5, one-line note on what pulled you away. Tag any session that was atypical (meeting ran long, poor sleep).

Coaching Notes on the Three Exercises

• Exercise 1: "Hours of deep work" is not an output metric, it is an input. If your work is genuinely hard to count, use effort-weighted output, for example "sections completed" or "review comments resolved." A slightly rough metric beats no metric.
• Exercise 2: Keep the number modest. A 15 to 25 percent target is realistic within two weeks. If your rule says "keep it if it feels better," you have no rule.
• Exercise 3: Write the entry before you open anything else, while the session is still fresh. The note line is the part that explains the numbers. After five entries you will see a pattern, usually one recurring distraction type.

_____ is the minutes lost after an interruption before you return to the original task, and it is often the most honest cost in the journal.

Where Journals Quietly Fail

• Logging hours only. Time in the chair says nothing about whether attention converted into output.
• Changing two things. Phone off plus a new playlist plus a new block length gives you an answer to nothing.
• Deciding the rule afterwards. Choose the threshold first, then collect data. Otherwise every result looks like a win.
• Excluding bad days silently. If a session was atypical, log it and tag it. Better: write the exclusion rule before you start, for example "sessions interrupted by a call longer than ten minutes are tagged and left in."
• Chasing the metric. If your words per session rise while your review time doubles, the experiment found a trade-off, not a win. The guardrail exists for this.
• Comparing means on five sessions. One rough day drags the average around. Compare _____ so a single outlier does not decide your routine.

Rules of Thumb for Data You Can Trust

• Five sessions per condition is the minimum useful sample. Fewer, and you are reading noise.
• Change one variable per experiment. Run the next one after you have decided on this one.
• Keep everything else constant: same time of day, same block length, same kind of task.
• Log in under a minute. An entry you avoid writing is worse than a rough entry.
• Keep the experiment boring and repeatable. If you cannot run it again next month, it taught you nothing.
• Final line: if the change meets your rule, keep it and stop measuring it. Move the journal's attention to the next question.

What to Carry Forward

A focus journal converts a vague feeling ("today was scattered") into a signal you can compare, and a one-variable experiment tells you which adjustment earned its place. Keep the entries short, write the decision rule before the data, and retire each change once it is proven or rejected. In the next lesson you will string these blocks, breaks, and experiments into a realistic two-week deep work routine.

Front: Why log an output metric instead of just hours? Back: Hours measure time spent; an output metric shows whether attention became finished work.

Front: What belongs in a minimum focus journal entry? Back: Block length, intended deliverable, distraction count, resumption time, output metric, energy, one-line note.

Front: What is resumption time? Back: The minutes lost after an interruption before you return to the original task.

Front: How many variables change between baseline and test conditions? Back: One. Two changes make the result unreadable.

Front: When do you write your prediction and decision rule? Back: Before collecting any data, so results cannot be reinterpreted afterwards.

Front: Why compare medians instead of means on a small sample? Back: One atypical day skews an average; the median resists outliers.

Front: What is the purpose of the baseline week? Back: To capture current output without changing anything, so the comparison has a reference point.

Front: What makes a usable output metric? Back: Countable, tied to one deliverable, and recorded at the end of each session.

Front: How should an atypical session be handled? Back: Log it, tag it, and apply an exclusion rule that was written in advance.

Front: What is the minimum useful sample per condition? Back: About five sessions.

Front: Which two attention signals pair with the output metric? Back: Distraction count and resumption time.

Front: What are the two outcomes of an experiment? Back: Keep the change if it meets the rule, or drop it and test one new variable.

Front: What does rising output with a collapsing energy score signal? Back: A trade-off, not a win. The guardrail in the decision rule catches this.