When Self-Tracking Backfires: Proxies, Obsession and Fixes

Personal metrics are supposed to serve your goals. Sometimes they quietly replace them. You walk laps around the kitchen at 11 p.m. to hit a step count, shuffle money between accounts so this month's savings rate looks better, or lie awake worrying about last night's sleep score.

None of this means tracking is a bad idea. It means every number you track is a stand-in for something you care about, and stand-ins can drift away from the real thing, especially once you start pushing on them.

This guide explains why that happens, how to spot the warning signs and how to design personal tracking that stays useful.

Goodhart's law, applied to you

The best-known name for this problem is Goodhart's law, after British economist Charles Goodhart. As quoted by researchers David Manheim and Scott Garrabrant, his original 1975 formulation was that "any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes."

The more familiar wording, "when a measure becomes a target, it ceases to be a good measure," is a later paraphrase. In a 2022 paper, researcher Oliver Braganza notes that this best-known phrasing was coined by anthropologist Marilyn Strathern, writing about university audits, rather than by Goodhart himself.

Goodhart was writing about monetary policy, but the idea scales down to one person. Manheim and Garrabrant describe a Goodhart effect as what happens when optimization causes a collapse of the statistical relationship between the goal you intend and the proxy you use for it.

Almost every personal metric is a proxy:

• Steps stand in for fitness and health.
• Hours studied stand in for learning.
• Savings rate stands in for financial security.
• A sleep score stands in for feeling rested.

When you just observe these numbers, they tend to track the goal reasonably well. When you start chasing the number itself, the link can weaken.

Four ways a proxy goes wrong

Manheim and Garrabrant sort Goodhart effects into four types. Each has a personal version; the examples below are hypothetical.

1. Regressional. When you select for a proxy, you also select for the gap between the proxy and the goal. For example, if you judge workouts only by step count, you'll favor the ones that produce steps over the ones that build strength.
2. Extremal. When a proxy is pushed to extreme values, the old relationship may no longer hold. Extra study hours may help up to a point; at some point the added hours are tired, distracted ones that look identical in the log.
3. Causal. When the link between proxy and goal is indirect, acting on the proxy can break it. Moving money between accounts to improve a savings-rate calculation changes the number without changing your finances.
4. Adversarial. When someone whose goals differ from yours influences the metric, it can be pulled away from your purpose. Imagine a workplace step challenge with prizes: the metric starts serving the competition rather than your health.

When the data becomes the worry

Sleep tracking offers a well-documented example. In a 2017 case series in the Journal of Clinical Sleep Medicine, Kelly Glazer Baron and colleagues described patients who sought treatment for self-diagnosed sleep problems based on their sleep tracker data. The authors coined the term "orthosomnia" for patients who were preoccupied with improving or perfecting their wearable sleep data, describing a "perfectionistic quest for the ideal sleep."

The authors noted that validation studies had shown consumer sleep trackers were unable to accurately discriminate stages of sleep and had poor accuracy in detecting wakefulness after falling asleep, yet they found patients' perceptions difficult to change. The broader lesson applies to any metric: if a number is raising your anxiety more than it's informing your choices, it's working against you. If sleep problems persist, that's a conversation for a health care provider, not a tracker.

Remember what your device is, and isn't

Many consumer apps and wearables are built for general wellness. The FDA's general wellness guidance explains that, under federal law, software intended "for maintaining or encouraging a healthy lifestyle" that is "unrelated to the diagnosis, cure, mitigation, prevention, or treatment of a disease or condition" is not a medical device. That's a good reason to treat a wellness app's readings as rough personal feedback, not as a diagnosis.

Popular targets deserve the same skepticism. The Physical Activity Guidelines for Americans say popular advice such as walking 10,000 steps a day "is not a guideline per se." A round number on a screen is not automatically the right goal for you. Our guide Activity Metrics: Tracking Minutes and Steps the Smart Way explains how to build a step goal from your own data instead.

Warning signs that tracking is backfiring

• You change what you do to move the number rather than to reach the goal behind it.
• You skip something worthwhile because it doesn't register on the tracker.
• You feel worse after checking, or you check far more often than you need to.
• You trust the number over clear signals from your own body or circumstances.
• Recording and reviewing takes more time and energy than the thing you're tracking.
• You can no longer say, in one sentence, why you track this metric.

Design tracking that stays honest

1. Write the goal next to every metric. One sentence, such as "steps, because I want to be able to hike with my kids," makes drift easier to notice.
2. Pair each number with a check on quality. Steps with how you feel, study hours with a delayed self-test, savings rate with whether your budget feels livable.
3. Review weekly, not constantly. Most personal goals play out over weeks, and frequent checking invites overreaction to noise.
4. Use ranges instead of single thresholds. "Most days between 6,000 and 8,000 steps" creates less pressure to game a cutoff than "10,000 or bust."
5. Give metrics an end date. Track something long enough to learn from it, then decide whether to keep it, change it or retire it.
6. Take breaks. Stop tracking for a few weeks and see whether the behavior holds. If it does, you may not need the number anymore.

If tracking anything related to your health, eating, sleep or mood is causing real distress, step back and consider talking with a health care provider. If you're in crisis or having thoughts of suicide, call or text 988.

Key takeaways

• Every personal metric is a proxy, and Goodhart's law warns that proxies can stop tracking the goal once you push on them.
• Manheim and Garrabrant's four types, regressional, extremal, causal and adversarial, each have everyday versions.
• Sleep researchers coined "orthosomnia" for patients preoccupied with perfecting their tracker data; anxiety is a sign a metric has turned against you.
• Wellness apps aren't medical devices, and round-number targets aren't guidelines.
• Write down the goal behind each metric, pair it with a quality check, review weekly and retire metrics that no longer help.

Sources

• David Manheim and Scott Garrabrant, Categorizing Variants of Goodhart's Law (arXiv, 2018; revised February 2019)
• Oliver Braganza, Proxyeconomics, a theory and model of proxy-based competition and cultural evolution, Royal Society Open Science (February 2022)
• Kelly Glazer Baron et al., Orthosomnia: Are Some Patients Taking the Quantified Self Too Far?, Journal of Clinical Sleep Medicine (February 2017)
• U.S. Food and Drug Administration, General Wellness: Policy for Low Risk Devices (January 2026)
• U.S. Department of Health and Human Services, Physical Activity Guidelines for Americans, 2nd edition (2018)