Do accountability partners work? The evidence, honestly
Accountability partners can work, but not the way the famous 95% statistic claims, because no one has found the study behind it. Real trials show that a partner helps most when progress is recorded and reported, and when someone is actually watching the result. Simply announcing a goal can backfire: in one set of experiments, people whose intentions were noticed by others then worked less hard at them.
Almost every article about accountability partners opens with the same number: a 95% chance of reaching your goal if you have a specific accountability appointment with someone. It is a great number. It is also, as far as anyone can tell, not from a study.
The real evidence is more interesting, more useful, and includes one finding that should change how you talk about your goals.
Where does the 95% statistic come from?
The figure is usually credited to the American Society for Training and Development (ASTD), as the last rung of a ladder: 10% if you hear an idea, 25% if you decide to do it, 40% if you decide when, 50% if you plan how, 65% if you commit to someone, and 95% if you book a specific accountability appointment with them.
It appears on coaching sites, in business columns and on university alumni pages. None of them cites a paper, a sample size or a method. Some date it to 2010, others to no year at all. We looked and could not find the study, and we are not aware of anyone who has.
That does not prove the idea is wrong. It means the number is decoration. Here is what has actually been measured.
What does the research actually say?
Friends plus support kept people going. In a 1999 trial, 166 people joined a weight-loss programme either alone or with three friends or family members, and were then randomly given either the standard programme or the standard programme plus social support strategies [1]. Of those who came alone and got the standard programme, 76% finished it and 24% kept all the weight off over the following six months. Of those who came with friends and got the support, 95% finished and 66% kept it off.
Two honest caveats. People chose whether to come with friends, so the people who did may simply have been different. And it is a weight-loss study from 1999, not an exercise study from this decade. But the size of the gap is hard to ignore, and the support strategies were randomly assigned.
Reporting your progress beats keeping it to yourself. A 2016 meta-analysis pooled 138 experiments with almost 20,000 people [2]. Interventions that got people to monitor their progress more often helped them reach their goals. And the effect was larger when the results were reported to someone or made public, and when they were physically written down.
Note what that is and is not. It is not “tell people your goal”. It is “show people your results, regularly”.
A watcher helps while the game is running. Competition helps after it stops. A 2019 trial gave 602 adults a step tracker and a step goal, then randomly assigned them to a control group or one of three social setups for 24 weeks [3]:
- Support: each person chose a sponsor, who got a weekly note saying whether they had hit their goals.
- Collaboration: teams of three, where one randomly chosen member’s result each day decided whether the whole team kept its points.
- Competition: groups of three, with a weekly leaderboard.
All three beat the control group during the game. Competition did best, at 920 more steps a day; support added 689 and collaboration 637. Then the game ended. Over the next twelve weeks, every group slipped, and only the competition group stayed clearly ahead of the controls. The support and collaboration groups no longer were.
More gamification is not the same as more accountability. In a related trial, 98 two-person teams of a person and a friend or family member tracked their weight and steps together [4]. Teams that also got points and levels on top of that did no better than teams that simply had the partner and the tracking.
When does telling people backfire?
This is the finding that should change your behaviour.
In four experiments published in 2009 [5], researchers asked people to state an intention tied to who they wanted to be, for example law students planning to read law journals to become good lawyers. When someone else took notice of that intention, people then acted on it less: less effort over the following week, and less readiness to act when the chance came up.
The explanation the authors tested is uncomfortable. Being acknowledged gave people a premature sense of already being the person they were aiming to become. The nod from someone else felt like progress, so the real progress mattered a little less.
The effect only appeared among people strongly committed to the goal, which is exactly the person who tends to announce it.
So “tell everyone you’re training for a marathon” is not neutral advice. Announcing a plan can substitute for doing it. Reporting a result cannot.
Why do most gym buddies quietly stop?
Put the findings together and the usual partnership looks badly designed.
It runs on self-report. “Did you go?” “Yeah, mostly.” Nobody checks, and nobody wants to be the friend who asks for proof.
It runs on politeness. The partner’s job, in practice, is encouragement. Encouragement is pleasant, and in the step trial the sponsored group’s lead faded once the weekly reports stopped.
It front-loads the announcing. The energy goes into the plan, the group chat, the shared start date. That is the part the 2009 experiments suggest can feel like progress without being any.
And nothing happens when you miss. A missed session costs an awkward message at most. Usually it costs nothing, and a miss that costs nothing becomes a pattern.
What makes accountability actually work?
The research points at four properties, and none of them is about how supportive your friend is.
- The result is recorded, not remembered. A step count, a logged session, a timestamp. Something that exists whether or not you feel like reporting it.
- Someone sees it on a schedule. Weekly at least. The sponsors in the step trial got a report every week without the participant having to send one.
- Missing is visible. The partner has to be able to say “you missed” without it being a confrontation, which means they need the data, not your version of it.
- You cannot quietly change the deal. The most important property and the rarest. If you can move the goal or the deadline at 9pm without anyone knowing, the arrangement is a preference, not a commitment.
The best partner, by these rules, is not the most encouraging one. It is the one who holds a key you cannot get back without asking.
What if you don’t have a person?
You can get part of the effect on your own, and the rest from design.
Record your progress and make it visible. The meta-analysis found monitoring helps even before anyone else is involved; reporting it adds to that.
A partner does not have to be real to raise your effort. In lab studies, people held plank exercises for longer when they worked alongside a virtual partner who was a little better than them than when they exercised alone [6]. That is a lab effect, measured in seconds of persistence, not a training plan. But it suggests that part of what a partner does is simply raise the bar.
Take away your own ability to let yourself off. That is what a commitment device is, and it is the job a good partner was doing all along. Different accountability apps stake different things: money, a message to a friend (that is Forfeit’s approach), or access to something you want.
Dare stakes the last of those. It reads your steps, exercise minutes or active calories from Apple Health, so the result is recorded rather than reported, and there is no button to mark a day done. Miss the deadline and the apps you chose lock until you make it up. And there is an optional PIN, which Dare suggests you let a friend set, that guards giving up a live lock, editing or removing a dare and turning off strict mode. That is the “key you cannot get back without asking”, built in. The research above supports that shape; none of it tested our app.
The short version: a partner works when they see the truth on a schedule and you cannot quietly change the deal. Everything else is encouragement, and encouragement wears off.
Sources
- Wing, R. R. & Jeffery, R. W. (1999). Benefits of recruiting participants with friends and increasing social support for weight loss and maintenance. Journal of Consulting and Clinical Psychology, 67(1), 132–138. 166 participants; recruitment alone or with friends was chosen, the social support condition was randomly assigned.
- Harkin, B., Webb, T. L., Chang, B. P. I., Prestwich, A., Conner, M., Kellar, I., Benn, Y. & Sheeran, P. (2016). Does monitoring goal progress promote goal attainment? A meta-analysis of the experimental evidence. Psychological Bulletin, 142(2), 198–229. 138 studies, N = 19,951; effect on goal attainment d = 0.40, larger when outcomes were reported or made public and when physically recorded.
- Patel, M. S. et al. (2019). Effectiveness of behaviorally designed gamification interventions with social incentives for increasing physical activity among overweight and obese adults across the United States: the STEP UP randomized clinical trial. JAMA Internal Medicine, 179(12), 1624–1632. 602 adults, 24-week intervention, 12-week follow-up. Five of the twelve authors were from Deloitte Consulting.
- Kurtzman, G. W. et al. (2018). Social incentives and gamification to promote weight loss: the LOSE IT randomized, controlled trial. Journal of General Internal Medicine, 33(10), 1669–1675. 196 adults in 98 two-person teams; no arm had participants without a partner, so it does not measure the partner’s own effect.
- Gollwitzer, P. M., Sheeran, P., Michalski, V. & Seifert, A. E. (2009). When intentions go public: does social reality widen the intention-behavior gap? Psychological Science, 20(5), 612–618. Four experiments; the effect held for participants with strong, not weak, commitment to the identity goal.
- Feltz, D. L., Kerr, N. L. & Irwin, B. C. (2011). Buddy up: the Köhler effect applied to health games. Journal of Sport & Exercise Psychology, 33(4), 506–526. Plank persistence with a virtually presented partner whose performance was set to be always superior.
Common questions
Is it true that an accountability partner gives you a 95% chance of success?
Not as far as anyone can show. The figure is usually credited to the American Society for Training and Development, as the top rung of a ladder that runs from 10% for having an idea to 95% for a specific accountability appointment. It is repeated on hundreds of coaching sites, and none of them links to a study, a sample or a method. Treat it as a slogan, not a finding.
Do accountability partners actually work?
The evidence says they can. In a 1999 weight-loss trial, people who joined with friends and were given social support were far more likely to finish and keep the weight off than people who joined alone. A 2016 meta-analysis of 138 experiments found that tracking your progress helps, and helps more when the results are reported to someone. What does not reliably help is simply telling people your plans.
Can telling people your goals make you less likely to achieve them?
In some cases, yes. A 2009 set of experiments found that when other people noticed someone's identity-related intention, such as a law student's plan to read law journals, that person then worked at it less. The researchers' explanation is that being acknowledged gives a premature sense of already being the person you want to become. Reporting results is different from announcing plans.
What makes a good accountability partner?
Someone who sees the result, not your account of it, on a fixed schedule, and who is willing to say you missed. Most partnerships fail because they run on self-report and politeness: you say it went fine, they say well done. A partner works when checking in is unavoidable and lying would be awkward.
What if I don't have anyone to be my accountability partner?
You can get some of the effect without a person. Recording progress and making it visible helps on its own, and in lab studies even a virtual exercise partner made people hold a plank for longer. Beyond that, a commitment device does the partner's real job, which is removing your ability to quietly let yourself off.