# Habit Tracking That Sticks: Use Density Targets, Not Streaks

> Most streak counters reset after one missed day. Set density targets like 10 of 14, add a tiny floor version, and build a tracker that survives a rough week.

- Canonical page: https://learnhubplay.nl/blog/habit-tracking-that-sticks
- Published: 2026-07-29 (updated 2026-10-09)
- By: LearnHubPlay, an independent software studio in Amsterdam
- Topic: Habits & Routines
- Tags: Habit tracking, Habit formation, Behaviour change

Say you have walked for ten minutes on 19 days in a row. On day 20 a delayed train eats your evening, and the tracker reports a streak of zero. Nineteen walks happened, but the screen no longer shows them, and the easiest response is to stop opening the app.

A tracker decides what a bad day looks like, so its design affects how likely you are to come back after one. This guide covers what research does and does not say about missed days, how to set a density target (a "10 of the last 14 days" number) for each type of habit, and how to build a paper or app tracker whose numbers hold up after a rough week.

It ends with a two-minute weekly review. A notebook is enough to run it.

## Why a broken streak can sting, and what one missed day does to a habit

### How a broken chain can shape your next decision

In seven studies, Jackie Silverman and Alixandra Barasch (*Journal of Consumer Research*, 2023) found that highlighting an intact streak in a behaviour log raised later engagement more than highlighting a broken one. The abstract compares those two conditions, not a broken streak with no streak. The effect depended on how the log represented the behaviour, not on what people had actually done, and the authors argue that people treat keeping a logged streak going as a goal in itself.

Two moderators matter for design: the effect was stronger when people blamed themselves for the break and weaker when they could repair the streak. The abstract does not say how long after the break engagement was measured.

Two older ideas describe a similar reaction. Both come from outside habit research:

- **The what-the-hell effect.** Janet Polivy, a University of Toronto psychologist who studies dieters, uses this label for what happens once a diet feels broken. In a lab study (Polivy, Herman and Deo, *Appetite*, 2010; 106 young women, per University of Toronto Magazine), everyone ate the same slice of pizza, then rated cookies the researchers weighed. When women were led to believe their slice was larger than another participant's, restrained eaters (dieters) tended to eat more cookies and unrestrained eaters fewer. Polivy's reading, as the magazine reports it: the diet already felt ruined, so restraint seemed pointless.
- **The abstinence violation effect.** In Marlatt and Gordon's 1985 relapse prevention model, a lapse can trigger self-blame, guilt, and lost confidence, with stronger reactions when it is pinned on a stable flaw such as "no willpower."

Applying either to a missed walk is an analogy, since one comes from dieting and the other from addiction research, and the second is contested. In a study of 203 smokers who had quit (1,001 lapse episodes), Kirchner, Shiffman and Wileyto found that reactions to the first lapse did not predict relapse. After later lapses, a drop in confidence predicted a faster slide to the next lapse, while stronger internal blame went with a slower one, the reverse of what the model would predict. If anything transfers, it is that confidence is worth protecting.

### What the habit research says about one missed day

The best-known study of how long habits take is Lally, van Jaarsveld, Potts and Wardle (2010). Ninety-six volunteers each picked an eating, drinking, or activity behaviour and did it daily in the same context, such as after breakfast, for 12 weeks, rating each day how automatic it felt; 82 gave enough data to analyse. Their abstract states: "Missing one opportunity to perform the behaviour did not materially affect the habit formation process." More consistent performance went with better-fitting habit curves. Among people whose curves fitted well, reaching 95% of projected peak automaticity took 18 to 254 days, and anything past day 84, when the study ended, was extrapolated. The timeline is covered in [how long it takes to form a habit](https://learnhubplay.nl/blog/how-long-to-form-a-habit).

> **Note:** Read the finding at its real size. Participants were volunteers choosing their own once-daily behaviour, automaticity was self-reported, and the habit curve fitted well for 39 of the 82 people with usable data. A 2022 methods review by Gardner, Rebar and Lally found that the exemplary real-world studies all concerned behaviours done at least once daily. "One miss does not derail the habit" is what the best-known study found. "Many misses are fine" is not what the abstract claims.

The same review notes that "the key determinant of habit development is consistent repetition, not time," and that plotting habit against calendar days overlooks missed performances. The authors offer this as methodological advice, and add that for once-daily behaviours time works as a rough proxy for repetition. A tracker should therefore count repetitions, and one miss should not wipe out progress. A density number plus a running total does both.

We know of no study that compares density tracking with streaks head to head, so treat it as a design inference from the research above, not a proven intervention.

## Density: the number a missed day can only dent

Density is days done divided by days in a rolling window. Twelve completions in the last 14 days is "12 of 14," or 86%. Each day the oldest day drops off and the newest is added, so one miss fades out of the number after two weeks instead of resetting it. A paper calendar or a spreadsheet gives you the same "x of the last 14 days" count.

The window length decides how hard a miss lands. Here is one miss after a perfect 28-day run:

| Measure | Before the miss | After one miss | Change |
|---|---|---|---|
| Streak | 28 | 0 | down 28, a total reset |
| 7-day density | 7 of 7 (100%) | 6 of 7 (86%) | down 14 points |
| 14-day density | 14 of 14 (100%) | 13 of 14 (93%) | down 7 points |
| 28-day density | 28 of 28 (100%) | 27 of 28 (96%) | down 4 points |

We suggest 14 days as a default. A 7-day window swings 14 points per miss, so a rough weekend looks like failure. A 28-day window is slow to show a real change. A 14-day window also always holds exactly two of each weekday.

**Worked example.** Back to the walker with the 19-day run: 19 walks, a miss on day 20, then walks on days 21 to 23. On day 23 the streak says 3. The 14-day density (days 10 to 23) says 13 of 14, or 93%, and the running total says 22 walks. Only the density and total describe what happened.

### How a rolling window forgives a bad week

A truly bad week still shows. Say a 7-of-7 week is followed by a 2-of-7 week, then two perfect weeks:

| Day | Last 7 days | Last 14 days |
|---|---|---|
| 14 (after the 2-of-7 week) | 2 of 7 | 9 of 14 (64%) |
| 21 (after one perfect week) | 7 of 7 | 9 of 14 (64%) |
| 28 (after two perfect weeks) | 7 of 7 | 14 of 14 (100%) |

On day 14 the 64% sits below a target of 10 of 14 (71%), which triggers a review, not a verdict. The 14-day number stays flat on day 21 because the perfect week that drops out is replaced by another perfect week, and it recovers fully only once the bad week leaves the window. Read the 7-day count as the early sign of recovery and the 14-day count as the verdict.

## Choosing a density target by habit type

The right target depends on how costly the habit is and how often it can happen. These numbers are our starting-point heuristics, not research thresholds. Count against what you planned, not the calendar: for habits that are not daily, use sessions in 14 days or weeks out of the last eight.

| Habit type | Count against | Starting target | First fix if under 50% for two reviews |
|---|---|---|---|
| Cheap daily (flossing, one journal sentence) | Last 14 days | 12 or 13 of 14 | Move the anchor; the cue is failing |
| Effortful daily (30 minutes of study, 300 words) | Last 14 days | 9 or 10 of 14 | Shrink the floor to 10 minutes |
| Planned sessions (strength training, runs) | Sessions you planned in 14 days | For example 5 of 6 | Plan fewer sessions, not more |
| Workday-only (shutdown checklist, daily plan) | Working days in last 14 | 8 or 9 of 10 | Attach it to closing the laptop |
| Weekly or monthly (weekly review, backup check) | Last 8 weeks | 6 or 7 of 8 | Book it as a calendar appointment |

Under 50% means fewer than 7 of 14 days, 3 of 6 sessions, 5 of 10 working days, or 4 of 8 weeks. A target also fixes how many blanks the window can absorb: 10 of 14 allows 4, 12 of 14 allows 2, and 5 of 6 allows 1. Lally's abstract links more consistent performance with better-fitting habit curves, so a lower target trades speed for sustainability. We have no data on where that trade-off sits.

Setting the number is a four-step process:

1. **Define the floor.** This is the smallest version that counts as done, such as one sentence or two minutes.
2. **Baseline for two weeks.** Log the floor with no target and see how many of 14 days you hit.
3. **Aim a notch above baseline, below your best fortnight.** If you hit 8 of 14, try 10 of 14, not 14 of 14.
4. **Revisit every two weeks.** Raise either the target or the size of the habit, never both at once.

Two edge cases, both our heuristics: with a baseline of 12 or more of 14, raise the habit's size, not the target. Under 5 of 14, the floor is too big or the anchor is not firing, so shrink or move it before setting a target.

> **Warning:** Do not use a density target for medication or anything medically timed. A tracker's "good enough" number is not dosing advice. Ask your prescriber or pharmacist what to do after a missed dose, and use the tracker only to see your pattern.

### Excused days: illness, travel, planned breaks

Mark a day "e" when you could not do even the floor: flu, a travel day with no access, a planned week off. An excused day leaves both the numerator and the denominator, so 10 of 14 becomes 10 of 13. Cap it at three per 14 days, or the denominator shrinks until the number means nothing. "I was busy" is a floor day, not an excused day. Again, this is our rule of thumb.

## Tiny floors and fixed anchors

BJ Fogg's Tiny Habits method (also the title of his book) has three parts, as laid out on [tinyhabits.com](https://tinyhabits.com/rewire/): an anchor (the moment the new habit follows, such as pouring coffee), a tiny behaviour (flossing one tooth is his example), and an immediate celebration such as saying "Good for me!" A recipe might read: after I pour my morning coffee, I write one sentence.

The anchor-plus-repetition structure matches what Lally's participants did. We treat the celebration step as Fogg's theory: cheap to try, but not something we can back with evidence here.

For tracking, turn the tiny version into a **floor**: the smallest action that counts as done. Log it differently from the full version, and count both toward density.

> **Tip:** Mark full sessions "X," floor sessions "o," and misses with a dot or a blank. Density counts X and o together. The ratio shows whether the full version is realistic, and the floor means a terrible day still produces a completion.

Our rule of thumb: if floor days outnumber full days two to one for two reviews in a row, call the floor the habit for now, or define a middle version between "o" and "X."

Why a floor helps is partly inference: if confidence after a lapse matters, as the smoking study's later-lapse results suggest, a minimum you can always reach protects it. The same all-or-nothing trap breaks hard website blocks, covered in [why website blockers fail](https://learnhubplay.nl/blog/blocking-distractions-with-friction).

### Does the anchor need to be an event?

Not necessarily. Keller and colleagues (2021) randomly assigned 192 adults to link a nutrition behaviour to either a daily routine or a clock time and tracked them for 84 days. Both groups gained automaticity, with no difference between them. Only 27 of the 117 participants who followed the protocol met the authors' bar for successful habit formation; their median time to peak automaticity was 59 days. If your days have no stable routine, a clock time can work.

## Make progress cumulative so a bad week cannot erase it

Since repetition drives habits, show repetitions. Put a lifetime count next to the density number and never reset it, so a missed day pauses the count instead of erasing it.

<figure><img src="/images/blog/habit-garden.svg" alt="Three plants at day 7, day 30, and after a missed day, still growing" width="960" height="600" /><figcaption>Progress you can see, which a missed day pauses but does not erase.</figcaption></figure>

Two small rules keep the picture honest:

- **Milestones count repetitions,** such as the 50th and 100th walk, not consecutive days.
- **Allow repair, within limits.** Silverman and Barasch found that the streak effect weakened when people could repair a broken streak. Our own rule of thumb, which the studies do not test: log yesterday with what you actually did, but not last month.

## A bad-week checklist, a recovery rule, and a two-minute review

### The tracker checklist

These are our design rules, not research findings, apart from where a source is named. Check any tracker, paper or app, against them:

- The main number is a rolling window ("x of the last 14"), not a chain.
- A floor version counts as done, and full versions are tallied separately.
- Misses are blank cells, not red crosses or zeros, and you can log yesterday.
- Logging takes less effort than the floor itself (James Clear notes that tracking forces you into two habits).
- Excused days have a written cap, and the review is in your calendar.

### The recovery rule

The best-known recovery rule is "never miss twice." James Clear popularised it in *Atomic Habits*: "Missing once is an accident. Missing twice is the start of a new habit." It is a rule of thumb, not a research finding. It fits Lally's single-miss result, but we are not aware of a study testing two misses as a threshold.

Write your own as an if-then: *If I miss a day, tomorrow I do the floor version straight after the anchor, and I do not make up the missed one.*

### The two-minute weekly review

1. Read each habit's 14-day count and compare it with the target.
2. At or above target for two reviews: raise the target or the size, not both.
3. Under target but at least 50%: change nothing about the habit, and check whether the blanks cluster (see the table below).
4. Under 50% for two reviews in a row: shrink the floor or move the anchor.
5. Write one sentence naming the situation behind each miss ("meeting overran"), not a trait ("lazy"). Naming a situation rather than a trait may sidestep the stable-flaw reading that the abstinence violation model describes, but the smoking study above did not find that self-blame made lapses worse, so treat this as a cheap experiment, not an evidence-backed fix.
6. Choose one change for next week.

If you already do a weekly review, this slots into the [20-minute weekly review template](https://learnhubplay.nl/blog/weekly-review-template).

### Reading your blanks

These readings are our heuristics, not research results:

| Pattern in the blanks | Likely cause | First fix |
|---|---|---|
| Same weekday every week | The anchor does not exist that day | Add a second anchor for that day |
| One block of days | A life event | Mark excused days, then restart with the floor |
| Scattered, density 50 to 70% | The floor is too big, or the habit ranks low | Shrink the floor |
| Everywhere, under 50% | The anchor or the habit choice is wrong | Move the anchor, or swap the habit |

### A paper template

```text
Habit: Write one sentence in the journal
Anchor: After I pour my morning coffee
Floor (counts): 1 sentence      Full: 1 paragraph
Target: 10 of 14
Key: X = full, o = floor, . = miss (blank), e = excused

Week 1   M  T  W  T  F  S  S
         X  o  X  .  X  X  .
Week 2   M  T  W  T  F  S  S
         X  .  .  o  X  o  X

Last 14 days: 10 of 14  (7 full, 3 floor)
Total repetitions since start: 63
If I miss: tomorrow, floor version right after the anchor.
```

### A spreadsheet version

Use one row per day, real dates in column A, and the marks (X, o, e, or blank) in column B. These formulas give the rolling count, the lifetime total, and the window size after excused days:

```text
Last 14 days:  =COUNTIFS(A:A,">"&(TODAY()-14),A:A,"<="&TODAY(),B:B,"X")+COUNTIFS(A:A,">"&(TODAY()-14),A:A,"<="&TODAY(),B:B,"o")
Lifetime:      =COUNTIF(B:B,"X")+COUNTIF(B:B,"o")
Window size:   =14-COUNTIFS(A:A,">"&(TODAY()-14),A:A,"<="&TODAY(),B:B,"e")
```

The syntax follows Microsoft's and Google's COUNTIFS documentation, but we have not run these in every spreadsheet app, so check them on a few dummy rows first. Dates must be date cells, not text.

## Where to start this week

Pick one habit, or at most three. Define its floor and anchor, log for two weeks with no target, then set one from the table above. Put the two-minute review in your calendar and keep a lifetime total from day one. For more routines worth building, see the [habits topic hub](https://learnhubplay.nl/topics/habits).

## Frequently asked questions

### Should I delete my streak counter?

Not if it motivates you. In Silverman and Barasch's studies, people shown an intact streak were more likely to repeat the behaviour than people shown a broken one; that does not show a streak beats no streak. Keep yours if a break makes you want to repair it, and put a density number beside it. Drop it if seeing zero makes you avoid the app or you tend to blame yourself for breaks, since the effect was larger with self-blame.

### How many habits should I track at once?

One to three is our rule of thumb, and nothing in the studies above tests it. Start with one, and add another only after the first has held its target for a month.

### What should I do on the first day back after being ill or away?

Do the floor version straight after the anchor, even if you feel you owe extra, and skip the catch-up: it recreates the all-or-nothing pressure the floor exists to remove. After a long gap, read the 7-day count first, because the 14-day count lags your recovery.

## Sources and further reading

- [How are habits formed: Modelling habit formation in the real world](https://repositorio.ispa.pt/handle/10400.12/3364) — Lally, van Jaarsveld, Potts and Wardle, European Journal of Social Psychology, 2010 (DOI 10.1002/ejsp.674). The abstract supports the design, the 96 and 82 participant counts, the missed-opportunity result, and the 18 to 254 day range.
- [On or Off Track: How (Broken) Streaks Affect Consumer Decisions](https://academic.oup.com/jcr/article/49/6/1095/6623414) — Silverman and Barasch, Journal of Consumer Research, 2023 (DOI 10.1093/jcr/ucac029). Seven studies on intact versus broken streaks, self-blame, and repair.
- [Getting a bigger slice of the pie. Effects on eating and emotion in restrained and unrestrained eaters](https://doi.org/10.1016/j.appet.2010.07.015) — Polivy, Herman and Deo, Appetite, 2010. The pizza-and-cookies design and result. The plain-language write-up, [Does Dieting Affect How You Think?](https://magazine.utoronto.ca/?p=16585) (Alison Motluk, University of Toronto Magazine, 2010), adds the 106-woman sample and Polivy's explanation of the "what the hell effect."
- [Relapse dynamics during smoking cessation: Recurrent abstinence violation effects and lapse-relapse progression](https://doi.org/10.1037/a0024451) — Kirchner, Shiffman and Wileyto, Journal of Abnormal Psychology, 2012 (online 2011). Tests abstinence-violation reactions in 203 smokers over 1,001 lapse episodes.
- [Habit formation following routine-based versus time-based cue planning: A randomized controlled trial](https://discovery.ucl.ac.uk/id/eprint/10118622/1/Jan%20paper.pdf) — Keller, Kwasnicka, Klaiber, Sichert, Lally and Fleig, British Journal of Health Psychology, 2021. The 192-adult trial, the 59-day median for the 27 of 117 who met the habit criterion, and the note that Lally's 18 to 254 day range covers only well-fitting curves.
- [How does habit form? Guidelines for tracking real-world habit formation](https://discovery.ucl.ac.uk/10144814/1/23311908.2022.pdf) — Gardner, Rebar and Lally, Cogent Psychology, 2022. Supports the repetition-not-time point and the once-daily limit of the evidence.
- [Rewire Your Brain](https://tinyhabits.com/rewire/) — BJ Fogg, Tiny Habits. The anchor, tiny behaviour, and celebration recipe.
- [The Ultimate Habit Tracker Guide: Why and How to Track Your Habits](https://jamesclear.com/habit-tracker) — James Clear, excerpt from *Atomic Habits* (undated web page). Source for "never miss twice" and tracking as a second habit.