We Analyzed Our First 161 Real Journal Entries: 87% Were Logged on 'Not Great' Days

Short answer: people don't journal when they feel great. They journal from the fuzzy middle - the days that aren't quite bad, aren't quite good, just heavy enough to need somewhere to put it. Across Nuju's first 161 entries, 87% were logged on Rough, Low, or Okay days. Only 3% landed on 'Great'. The median entry was 31 characters - shorter than this sentence.

Most journaling advice imagines the user as someone calmly recording the day's wins, writing in flowing prose, building a beautiful archive. The data from real users says something different. People open the app when something needs to go somewhere - and they write the smallest amount that gets it out. Here's what 161 real entries from the first 54 days of Nuju usage reveal about why people actually journal.

Methodology: All numbers in this post come from queries against Nuju's anonymized aggregate database, executed 2026-05-22. Sample: 161 entries from 24 unique users between 2026-03-28 and 2026-05-21. Where a bucket has fewer than 5 entries (e.g., Mood 5 with N=3), the claim is described as cohort-specific rather than generalizable. Re-querying as the dataset grows.

Finding 1: People journal from the middle, not the extremes

The single clearest pattern in the data is the mood distribution. If journaling were primarily a gratitude practice (as much of the wellness category suggests), the top of the scale should be heavily represented. It isn't. The opposite happens: the distribution skews hard toward the lower half of the mood scale, with the 'Okay' midpoint dominating.

Mood distribution across 161 entries

• Mood 1 (Rough): 14 entries - 8.7%
• Mood 2 (Low): 43 entries - 26.7%
• Mood 3 (Okay): 83 entries - 51.6%
• Mood 4 (Good): 16 entries - 9.9%
• Mood 5 (Great): 5 entries - 3.1%
• Combined Rough + Low + Okay (1-3): 140 entries - 87.0%
• Combined Good + Great (4-5): 21 entries - 13.0%

More than half of all entries (51.6%) were logged on 'Okay' days. The fuzzy middle - not bad enough to be in crisis, not good enough to feel grateful - is what actually drives journal opens.

Why this matters: it changes what good journaling looks like

If journaling apps are mostly opened from the middle and lower-mid range of the mood scale, the design implication is different from what the gratitude-journal genre assumes. The user is not in a calm reflective state. They are at low-to-medium emotional bandwidth, often opening the app because something needs externalizing. The best response is fast, low-friction, and built to make sense of mess - not to celebrate a peak.

Finding 2: The median entry is 31 characters

Across the 153 non-empty entries in the sample, the average length was 173 characters. The median was 31 characters. The 25th percentile was 11 characters. That means a quarter of all journal entries are 11 characters or shorter - roughly two words.

Entry length distribution (153 non-empty entries)

• Median: 31 characters (≈ 6 words)
• Average: 173 characters (≈ 33 words)
• 25th percentile: 11 characters (≈ 2 words)
• 75th percentile: 93 characters (≈ 18 words)
• 5% of entries were empty (mood-only logs)

The gap between the median (31 chars) and the mean (173 chars) is the signature of a heavily skewed distribution: most entries are very short, with a handful of long ones pulling the average up. The typical Nuju entry is closer to a tweet than a paragraph. This is consistent with James Pennebaker's 35+ years of expressive writing research at UT Austin, which has consistently shown that short, bounded entries produce most of the emotional-processing benefit - not long, open-ended writing.

Half of all journal entries in the first 161 fit in fewer characters than a single tweet. The product imperative is clear: anything longer than a 30-second flow is friction.

Finding 3: Energy tracks the same 'middle bandwidth' shape

Each Nuju entry includes an optional energy rating from 0 (drained) to 100 (energized). The energy distribution mirrors the mood pattern: dominantly in the middle.

Energy distribution across 161 entries

• Drained (0-24): 4.3%
• Low (25-49): 18.0%
• Steady (50-74): 70.8%
• Energized (75-100): 6.8%

Almost three-quarters of entries (70.8%) are logged at 'steady' energy - neither drained nor energized. Combined with the mood data, the consistent profile of a Nuju entry is: Okay mood, steady energy, short text. Not crisis, not joy. The middle state most days actually live in.

Finding 4: Entry length barely changes with mood

One natural hypothesis: maybe people write much more on terrible days, processing harder. The data does not support this. Looking at median entry length by mood:

Median entry length by mood

• Rough (mood 1): 44 characters
• Low (mood 2): 24 characters
• Okay (mood 3): 26 characters
• Good (mood 4): 40 characters
• Great (mood 5): too few entries (n=3) to compare reliably

On rough days the median entry edges up to 44 characters - still tweet-length. Even on the worst days in the dataset, the typical user is writing one short sentence, not a paragraph. The pattern of 'short, bounded entry' holds across the entire mood scale.

Finding 5: Habit formation has two camps, no middle

Across the 24 users in the sample, return behavior splits into two clear groups: most try once, a meaningful minority become committed users, and almost nobody sits in between.

User engagement distribution

• 1 entry only: 13 users (54%)
• 2-4 entries: 5 users (21%)
• 5-9 entries: 0 users (0%)
• 10-29 entries: 5 users (21%)
• 30+ entries: 1 user (4%)

The complete absence of the 5-9 entry bucket is interesting. It suggests there's no 'gradual try-out' phase - users either drop after a few entries or commit past 10. The implication: the first few entries do most of the work in deciding whether someone becomes a journaler. This is consistent with classic habit-formation research showing the first 10-20 repetitions of a behavior are the highest-friction window.

What the data adds up to

Across mood, energy, length, and engagement, the same picture keeps appearing: people open a journal app from a middle state, write the smallest amount that gets it out, and either decide it's useful in the first few entries or move on. The data argues against three popular pieces of journaling advice:

• 'Journal when you're calm' - most entries are not from calm states; they're from middle-bandwidth ones, often slightly low.
• 'Write at least a page' - the typical entry is one tweet long, and that's working fine for the people who stay.
• 'Build the habit slowly over months' - the data shows the first ~10 entries decide everything; there's no soft on-ramp.

Practical implications for your own journaling

Two evidence-backed shifts make daily journaling more likely to stick, based on both Nuju's data and the broader expressive-writing literature:

1. Lower the bar to one sentence. The median Nuju entry is 31 characters. If a single line gets the loop out of your head, the entry has done its job. Anything longer is bonus, not requirement.
2. Open the app on Okay days, not just bad ones. Most entries come from the fuzzy middle, where the feeling is real but not urgent. Logging on those days is what builds the pattern data the AI can read.

Honest limits of this dataset

This sample is small: 161 entries from 24 users over 54 days. It describes Nuju's early user cohort, not the global population of journalers. Some buckets (e.g., Mood 5 with N=3) are too small to claim more than anecdotal. We will re-run these queries and update the post as the dataset grows. The patterns above are descriptive of the people who showed up first, not a claim about journaling in general.

Bottom line

If you've felt like real-life journaling never matches the long, calm, page-a-day version that journaling content shows, the data is on your side. 87% of real Nuju entries are logged on 'not great' days. Half are 31 characters or fewer. The product implication: pick the journal that's built for the actual middle state - fast, mood-aware, AI that reads short entries back without demanding more. Nuju's free Ju Gets You reveal takes 60 seconds; you can decide after the read whether the fit is real.