Part 10 of 10 · Emotional eating, measured

Does a food and mood journal work? What changed for 1,259 people in 14 days, and what didn't

Most apps in this space make the same promise: feel better. We can't make it. We checked, and our own data says it isn't true.

Here is what is true. Among 1,259 people who tracked their meals for three months, eating without physical hunger fell 26% and overeating fell 25% between their first two weeks and weeks nine to twelve. How often they felt bad before eating did not change at all.

That second sentence is the one we're proudest of, and this article explains why.

It's the last in a series built on 1.5 million meals logged in the I feel food app, where people record a hunger reading and a feeling before and after every meal. → What 1.5 million logged meals show about emotional eating

What changed

We took every user with at least eight valid meals in their first 14 days of logging and at least eight in days 56 to 83, weeks nine to twelve. That's 1,259 people. Then we compared each person against themselves.

Ate without physical hunger14.7% first 2 weeks10.9% weeks 9–12 · −26%Ended uncomfortably full8.5% first 2 weeks6.4% weeks 9–12 · −25%Could not name the feeling12.8% first 2 weeks9.7% weeks 9–12 · −24%Felt bad before eating38.4% first 2 weeks38.2% weeks 9–12 · no change
I feel food tracker · 1,259 users, first 14 days vs weeks 9–12
Show the numbers
BehaviourFirst 2 weeksWeeks 9–12Change
Ate without physical hunger14.7%10.9%−26%
Ended uncomfortably full8.5%6.4%−25%
Couldn't name the feeling12.8%9.7%−24%
Felt bad before eating38.4%38.2%no change

About seven in ten of the people who changed at all changed in the right direction: 71% for eating without hunger, 67% for overeating. The first two rows are not noise. On 1,259 paired observations the statistics are overwhelming.

Why you can believe it

The obvious objection to any result like this is survivorship: the people who stuck around were going to improve anyway. We can't remove that entirely, and we say so below. But the design handles the biggest version of it.

We didn't compare people who stayed with people who left. We compared the same people with themselves, earlier and later. Whatever brought them to the app, they brought it to both measurements.

And when we loosened the rules, four meals per window instead of eight, a wider later window of weeks nine to sixteen, the sample doubled to 2,155 people and the result got stronger, not weaker. Eating without hunger fell from 15.3% to 11.3%. Overeating fell from 9.6% to 7.3%.

How long food journaling takes to work

The change is fast, and then it holds.

Week 1Week 2Week 3Week 5Week 8Week 12Felt bad before eating: 38.4% → 38.2%, no change15.7%8.9%10.6%6.0%Ate without hungerEnded uncomfortably full
I feel food tracker · 1,259 users, week by week since their first entry
Show the numbers
Week since first entryAte without hungerEnded uncomfortably full
Week 115.7%8.9%
Week 212.8%6.8%
Week 311.5%6.1%
Week 510.9%5.9%
Week 810.4%6.0%
Week 1210.6%6.0%

Almost the whole shift lands in the first two to three weeks. After that it plateaus, and nothing decays back. Most of what changes, changes in the first fortnight. If you're going to try this, 14 days is an honest trial.

What didn't change

Three things stayed flat, and we'd rather say so than leave them out.

How often people felt bad before eating did not move. 38.4% in the first two weeks, 38.2% two months later. Life did not get calmer. The anxiety, the tiredness, the long days all kept arriving at the same rate.

Guilt after eating did not move either. It held at around 15% across the whole window. → The guilt loop

People named slightly fewer feelings per meal over time, not more: 2.25 down to 2.08. That could be precision, picking the one right word instead of scattering five. It could be logging less carefully as the novelty wears off. Our data can't tell those apart, so we won't claim it as a benefit.

Why that's the better story

If we'd found that distress fell, we'd have a mood app and a claim we couldn't defend. Instead we found something narrower and, we think, more useful.

Tracking didn't make our users' lives less stressful. It changed what they did about the stress. They stopped reaching for food when their body wasn't asking, and they stopped eating past comfortable. The bad days kept coming. The meals stopped absorbing them.

That's the whole series in one line. Eating rarely fixes the feeling. → Does emotional eating help? The decision that determines a meal is made before the first bite. → Physical hunger vs emotional hunger And the thing you can actually change isn't the trigger. It's the response, in the five seconds before you eat.

The limits

A sceptical reader will raise four things, and they're right to.

  1. No control group. This is observational. Some of the improvement would likely happen through attention alone, and we can't say how much.
  2. Survivorship. These are people who kept logging for two months, drawn from 50,692 who logged at least once. The findings describe engaged users, not everyone who installs. The looser 2,155-person check helps but doesn't remove this.
  3. Measurement drift. People may learn to read the hunger slider differently over time, which would look like behaviour change. The after-meal fullness reading moving in step with the before-meal hunger reading is partial reassurance, since they're separate judgements, but not proof.
  4. It's correlation. "People who tracked reduced X" is what the data supports. "Tracking reduced X" is a stronger claim than we can make.

None of that changes the numbers. It changes how confidently to read them, and we'd rather you read them accurately.

Back to the start of the series → What 1.5 million logged meals show about emotional eating

Cite this data

I feel food tracker · 1,534,951 meals, 2021 to September 2026

  • Before eating: a hunger reading on a 0–9 slider and a feeling picked from a list of 45 emotions.
  • After eating: the same two readings again, minutes later, on the same meal.

This article's before-and-after comparison follows 1,259 of those users over 83 days.

I feel food (2026). Does a food and mood journal work? What changed for 1,259 people in 14 days, and what didn't. I feel food tracker data, 1,534,951 meals, 2021 to September 2026. https://www.ifeelfoodapp.com/blog/does-food-journaling-work
BibTeX
@misc{ifeelfood2026doesfoodjournalingwork,
  author = {{I feel food}},
  title  = {Does a food and mood journal work? What changed for 1,259 people in 14 days, and what didn't},
  year   = {2026},
  howpublished = {\url{https://www.ifeelfoodapp.com/blog/does-food-journaling-work}},
  note   = {I feel food tracker data, 1,534,951 meals logged 2021 to September 2026}
}

How this was measured. 1,259 I feel food users with at least eight valid meals in both their first 14 days and days 56–83, compared against themselves. Paired t-statistics: −8.87 for eating without hunger, −7.03 for overeating, −1.29 (not significant) for distress before eating. Replication on 2,155 users with four meals per window and a later window of weeks 9–16: t = −10.73 and −7.88. Data spans 2021 to September 2026 and is observational, with no control group. Full methodology →