Changing several things at once: why you won't know what worked
AI-generated content. How we make it
In brief
- If you start several changes on the same day, their effects get mixed up: an improvement could be due to any one of them, to the combination, or to something else that changed in those same weeks. In research, this is called confounding.12
- Research has the same problem: a trial that evaluates a whole program tells you whether the package works, not which part does. To find out, each component is switched on and off separately.4
- When it comes to actually changing, going in sequence hasn't been shown to beat doing everything at once: the trials that compare the two give mixed results, and one review concluded that both approaches should be considered equally effective.6810
- If you want to know what works for you, stagger your changes: one change, one measure to write down, and the next change once the first has settled in. This is a deduction from the above, not a rule tested in trials.
When someone decides to take better care of themselves, they often start several things on the same day: going to bed earlier, walking more, eating differently at dinner. If a few weeks later they feel better, the natural question is what worked. And it almost never has an answer.
This article explains why, what research says about whether it's easier to change one thing at a time or everything at once — a question less settled than it's often made out to be — and how to order your changes so you can draw conclusions. It builds on an idea from the habits guide: starting with a few changes, chosen by you.
Why you won't know what worked
In research, this problem is called confounding: the effect of what you want to evaluate gets mixed up with that of another factor, and the relationship you see is distorted. It can make an effect appear that doesn't exist, or hide one that does.1 If you start three changes on the same day, each one is a confounder for the other two: they all start together, so any difference you notice coincides with all three.
On top of that come two traps that appear even if you change just one thing. Comparing a before and an after assumes that, without the change, everything would have stayed the same; but other things change in those weeks too, and they get mixed up with the effect.2 And if you started right in a bad patch, what comes next will usually move back toward what is usual even if you do nothing: this is regression to the mean.3 We explain both in detail in how to turn a data point into a question and a test.
This doesn't make changing several things at once a mistake. If what you want is to feel better and you don't mind which part did it, the problem is minor. It matters when you have to decide what to keep: in our view, a change that costs you time or effort is more worthwhile if you know it adds something.
Research has the same problem
Behavior change programs usually combine several components: content, messages, follow-up, who delivers them. A randomized trial that compares the whole program with a control group tells you whether the package works, but doesn't let you isolate the effect of each part; some components may be having an effect, others none, and others may even be subtracting from it.4
To separate them, researchers use factorial designs, in which each component is switched on or off separately. With two components there are four combinations — both, only one, only the other, and neither — and participants are randomly assigned among them.4
The same logic can be applied within a single person. In a series of “n-of-1” trials with 10 adults, over 60 days it was randomly decided each day, and separately, whether that day's message gave the participant a step goal or another kind of goal, and whether they could see what their pedometer showed. According to the published abstract, two were helped by the days with the pedometer in view, another two by the days with a goal, and six by neither technique.5 This could only be known because each technique was switched on and off independently.
Is it easier to change one thing at a time? The trials disagree
Knowing what worked is one question; managing to change is another. On the second, there are trials that compare two ways of setting up the same program: tackling several behaviors at once or one after another. A systematic review found only six up to 2015. Three saw some difference in at least one behavior: two in favor of going in sequence for smoking and one in favor of doing it all at once for dietary fat. There were no differences in how many people dropped out. The review concluded that the evidence is limited and that both approaches should be considered equally effective.6
| Trial | Participants | What they compared | What it found |
|---|---|---|---|
| Primary care, United States (2007)7 | 289 people aged 45 to 64 with hypertension, at a public clinic | Smoking, salt and steps: all three from the start or a new one every 6 months, over 18 months | At the end, only between 5% and 7% of each group, including the usual-care group, met at least two goals. For the goals taken one by one, results tended to favor the “all at once” group |
| CALM telephone program, United States (2013)8 | 200 adults aged 45 or older who said they often felt stressed, with little activity and little fruit and vegetable intake | Exercise and diet at the same time, or one first and the other at 4 months, over 12 months | At 4 months, only the group that started with exercise had increased their activity. At 12, only in the “all at once” group did the average meet all the recommendations |
| Online program, Netherlands (2014)9 | 5,055 adults from the general population | Personalized advice on 5 behaviors: all at once or one at a time | Small changes in both groups. Going in sequence stood out more at 12 months, and doing it all at once at 24. The authors couldn't conclude which is better |
| Make Better Choices 2, United States (2018)10 | 212 adults in Chicago with four diet and activity behaviors to improve | Physical activity from the start or added in week 7 (fruit and vegetables and screen time from the start in both groups), with an app, calls from a coach and small incentives, over 9 months | Both groups improved more than the control group. Going in sequence was somewhat better at 6 months, but not at 3 or 9 |
One more nuance: in Make Better Choices 2, doing everything at once didn't harm participants, but its authors warn that in other trials that approach was associated with more dropouts or relapses, and that it isn't known how many behaviors can be changed at once.10
Another way to look at the question is to count how many changes each program asked for. A meta-analysis of 150 reports on diet, exercise or smoking programs, with about 93,600 participants, found a curve: programs that recommended two or three behaviors were associated with more change than those that recommended just one, or four or more. Programs with just one were few and their result is very uncertain, and in the shorter programs the most favorable estimate was, on the contrary, for recommending just one, also with a lot of uncertainty.11 That meta-analysis counts how many behaviors a program asked for in total, not whether it introduced them at once or in sequence, and it compares different programs: it doesn't show that a person would do well to focus on two or three things.
Figure 1. How much behavior changed by how many changes the program asked for
What to do instead: stagger your changes
If you want to know what works for you, what follows from all of this is to stagger: one change first, and the next once the first is up and running. This isn't the result of a trial. The review concluded that going in sequence and doing it all at once should be considered equally effective,6 although in two of the trials in the table the results leaned toward doing it all at once.78 Given that, we think it's reasonable to see staggering as costing little in exchange for being able to tell what each change contributes; this is our judgment, not a conclusion from the sources. One possible order:
- Choose the first change: one that matters to you and that fits into your week as it is.
- Start writing down a single measure a few days beforehand, so you have a baseline. If you really want to check whether that change helps you, alternate periods with and without it, as we explain in the step-by-step personal test.
- Wait until the first change has settled in before adding the second. There's no validated time frame: the trials in the table added the next behavior in week 7, at 4 months or every 6 months, as a design choice.1087 A habit usually takes weeks or months to become automatic, as we explain in how long it takes to form a habit.
- Add the next change and keep writing down the same measure. If you notice another change, you'll have a clue as to what's behind it, though not proof, for the reasons given at the start.
On when to add the next one, the authors of Make Better Choices 2 suggest it might have worked better to do so once the first goal had been reached, rather than on a fixed date.10 On order, there's very little: in the CALM trial, those who started with exercise also improved their fruit and vegetable intake after a year, once diet had been added at 4 months, while those who started with diet never increased their activity relative to the control group.8 It's a single trial, in middle-aged, stressed adults; it isn't enough to set a rule.
If what you want to improve is your rest, our sleep guide suggests the same starting point: choosing a single habit and giving it a few weeks.
Staggering isn't always possible or necessary. If a health professional has suggested a plan with several changes, that plan comes first.
What isn't known
- How many behaviors can be changed at once without the result suffering: no one has established it.10
- How long to wait between one change and the next: the trials chose different intervals, and their authors acknowledge that other intervals, or waiting until the first goal was reached, might have given better results.810
- Whether order matters: there's a hint that starting with diet might later hold back activity, which the authors themselves say has yet to be confirmed.8
- What happens when someone changes on their own, without a program behind them: the trials didn't study it. And many of their results are based on what the participants themselves reported.8910
Keep reading
Sustainable healthy habits: how to start and stick with them
How to choose your first change: one that matters to you, fits into your week and lets you check whether you're actually doing it.
How to turn a wellness data point into a question and a test
How to test whether a specific change helps you by alternating periods with and without it, and why comparing a before and an after misleads.
How long does it take to form a habit? Not 21 days, and not exactly 66
How long a habit takes to become automatic, so you don't add the next change too soon.
Sleep better without complicating your routine
If you want to start with sleep: why to choose a single habit and how to track it with a sleep diary.
Want to turn information into action?
naro helps you prioritize 2–3 high-potential habits and sustain them over time, with context and follow-up.
Sources
Each small number in the text leads to the source that statement rests on. Practical advice without a number is a deduction from these sources, and examples with a person's name are made up. It is general information: it does not replace an assessment by a health professional. How we make this content
Sources consulted up to 28 October 2026.
- Methodological article. Skelly AC, Dettori JR, Brodt ED (2012). Assessing bias: the importance of considering confounding. Evidence-Based Spine-Care Journal. Written for clinical research; only its definition of confounding is used here.
- Narrative review. Duan N, Norman D, Schmid C, Sim I, Kravitz RL (2022). Personalized Data Science and Personalized (N-of-1) Trials: Promising Paradigms for Individualized Health Care. Harvard Data Science Review. Intended for health care; only the logic of personal comparisons is carried over here.
- Methodological article. Barnett AG, van der Pols JC, Dobson AJ (2005). Regression to the mean: what it is and how to deal with it. International Journal of Epidemiology. We could only read the abstract.
- Methodological article. Collins LM, Murphy SA, Strecher V (2007). The multiphase optimization strategy (MOST) and the sequential multiple assignment randomized trial (SMART): new methods for more potent eHealth interventions. American Journal of Preventive Medicine. Proposes methods for designing programs; doesn't evaluate any.
- Series of n-of-1 trials. Sniehotta FF, Presseau J, Hobbs N, Araújo-Soares V (2012). Testing self-regulation interventions to increase walking using factorial randomized N-of-1 trials. Health Psychology. 10 adults over 60 days. We could only read the abstract.
- Systematic review. James E, Freund M, Booth A, et al. (2016). Comparative efficacy of simultaneous versus sequential multiple health behavior change interventions among adults: A systematic review of randomised trials. Preventive Medicine. Six trials of guided programs published up to October 2015. We could only read the abstract.
- Randomized trial. Hyman DJ, Pavlik VN, Taylor WC, Goodrick GK, Moye L (2007). Simultaneous vs sequential counseling for multiple behavior change. Archives of Internal Medicine. 289 African American patients with hypertension at a public primary care clinic. We could only read the abstract.
- Randomized trial. King AC, Castro CM, Buman MP, Hekler EB, Urizar GG, Ahn DK (2013). Behavioral impacts of sequentially versus simultaneously delivered dietary plus physical activity interventions: the CALM trial. Annals of Behavioral Medicine. 200 adults aged 45 or older with perceived stress; physical activity self-reported by participants.
- Randomized trial. Schulz DN, Kremers SPJ, Vandelanotte C, et al. (2014). Effects of a web-based tailored multiple-lifestyle intervention for adults: a two-year randomized controlled trial comparing sequential and simultaneous delivery modes. Journal of Medical Internet Research. Many dropouts, and not at random; small, self-reported effects.
- Randomized trial. Spring B, Pellegrini C, McFadden HG, et al. (2018). Multicomponent mHealth Intervention for Large, Sustained Change in Multiple Diet and Activity Risk Behaviors: The Make Better Choices 2 Randomized Controlled Trial. Journal of Medical Internet Research. 212 adults, mostly women with a college degree; three of the four behaviors self-reported by participants.
- Meta-analysis. Wilson K, Senay I, Durantini M, et al. (2015). When it comes to lifestyle recommendations, more is sometimes less: a meta-analysis of theoretical assumptions underlying the effectiveness of interventions promoting multiple behavior domain change. Psychological Bulletin. Compares programs with each other: shows associations, not causes.