A habit fails in one of three places: the cue is missed, the routine is too large, or the reward is too distant. But before you fix any of that, you need a number that tells you which one broke. Most tracking gives you a total—steps, minutes, checkmarks—and a total is a eulogy, not a diagnosis. The Steps to Behavior Change model splits the process into five stages: knowledge, approval, intention, practice, and advocacy. Each stage demands its own metric. If you measure practice when the real stall is intention, your data will lie to you. Self-efficacy—the personal belief that you can execute the action—is the silent variable. Research links it to a wide range of health behaviors, from contraceptive use to safe listening. Measure self-efficacy weekly, and you’ll see trouble before the habit collapses.

Friction audit

Start with a one-question audit: what is the smallest point of resistance that stops you? Not the big, dramatic barrier—the tiny one. For an evening walk, the friction might be changing into outdoor clothes. For a two-minute meditation, it’s opening the app and staring at a menu. List every micro-step between intention and action. Then delete one. If you sleep in workout clothes, the morning exercise friction drops. If you bookmark a direct meditation link, the decision friction evaporates. Measure the result as a binary: did I do the thing, yes or no. Track that for five days. If the yes count is below three, the friction you removed wasn’t the right one. Audit again. The goal isn’t a streak; it’s a signal.

Habit-stack design

A habit-stack anchors a new routine to an existing one. The formula is “After I [current action], I will [new tiny action].” The current action must be a fixed event you never skip—brushing teeth, pouring coffee, locking the front door. The new action must be laughably small. After I pour my morning coffee, I will do one wall push-up. After I lock the front door, I will take three deep breaths. The stack works because the cue is already wired into your brain. You’re not relying on memory or motivation; you’re piggybacking on a circuit that fires every day. Test one stack for a week. Measure only whether the new action happened immediately after the cue. If it didn’t, the cue wasn’t automatic enough, or the action was too large. Shrink it further. One push-up. One breath. The stack is a design, not a hope.

Anchor cue

The cue is the trigger that launches the routine. In behavior-design terms, it’s the event that says “start now.” A weak cue is a time on a clock—8:00 p.m. walk—because clocks require you to notice them. A strong cue is an action you already do, like setting down your dinner plate. Pair the new habit to that action: after I set down my dinner plate, I will put on my walking shoes. The cue must be specific, immediate, and unavoidable. Test it by asking: can I do the cue without thinking? If you have to remember, it’s not a cue; it’s a reminder. Reminders fail on tired Tuesdays. Measure cue reliability with a simple count: how many times this week did the cue actually occur? If you set the cue as “after I finish lunch” but you skip lunch twice, the cue failed. Redesign it to an event that happens every day, like standing up from your desk. The anchor cue is the load-bearing wall of your habit. If it cracks, everything collapses.

Measurement

Measurement that matters answers one question: what broke? Choose a metric for each stage. For knowledge, track whether you can state the why of the habit in one sentence. For intention, track a weekly self-efficacy score: “On a scale of 1-10, how confident am I that I can do this tomorrow?” For practice, track a binary yes/no. For advocacy, track whether you recommended the habit to someone else. Most people track only practice and wonder why the habit feels fragile. But if your self-efficacy score is a 4, the practice metric is a lagging indicator. The real work is upstream. Set a tiny measurement cadence. Every Sunday, rate your self-efficacy for the week ahead. Every evening, log the binary. Review the numbers not to judge yourself, but to find the failure point. If the binary is low but self-efficacy is high, the problem is environmental—friction or cue. If self-efficacy is low, the problem is belief. Fix belief by shrinking the action until it’s impossible to fail. One push-up. One sentence of gratitude. One block walked. When the action is that small, self-efficacy rises because success is guaranteed. Then you scale.

Data without a decision is noise. A step count of 10,000 doesn’t tell you why you skipped Tuesday. A self-efficacy rating of 5 does. Build your measurement system to surface the weakest link, then reinforce it with a smaller anchor, a tighter cue, or a friction cut. The habit becomes durable not because you tracked it, but because you knew what to fix.