Build Better Teams draws on published research into how teams perform, learn and change: what makes teams effective, why each unit matters, and why the journey runs in real work, over weeks, with specific commitments and follow-up.
Across 46 independent samples, structured debriefs improved performance compared with groups that did not debrief (d = 0.67); the authors put the average gain at about 20–25%. That is a research average, not a forecast for your team.
Across 112 studies of 7,763 teams, trust within the team was consistently linked to higher team performance (ρ = .30), even after accounting for the team's past performance.
Across 94 independent tests, people who linked a goal to a concrete "when X happens, I will do Y" plan were more likely to reach it, a medium-to-large effect (d = 0.65).
Across 51 controlled studies, teamwork training improved team performance by a medium-sized margin (d = 0.58). Interactive methods (workshops, practice and team reviews) had the clearest effect on how teams actually worked together.
How to read the numbers: d compares a group that took part in an intervention with one that did not, in standard-deviation units; around 0.5 is usually called medium and 0.8 large. ρ (rho) is, in these analyses, a correlation corrected for measurement error; correlations run from −1 to +1. Where the authors describe an effect in words, such as "medium-sized", we use their words.
Most important work in organisations depends on several people working together. The research is clear that a team's results depend not only on who is in the team but on how the team works (Mathieu et al., 2008; Kozlowski & Ilgen, 2006). Talented individuals do not automatically make an effective team. In one well-known study, star analysts who changed firms on their own saw their performance fall, often for years, unless they moved to a firm with stronger capabilities. Stars who moved together with their teammates showed no significant decline (Groysberg et al., 2008). In software teams, how often members had worked together predicted performance, while individual seniority did not consistently do so (Huckman et al., 2009).
The things that make a team effective (shared goals, trust, the ability to speak up and disagree well, clear roles, good coordination) are not fixed traits. They are patterns of working that a team can see, discuss and improve. Each unit of the journey focuses on one of them.
Teams perform better when they share a clear, specific goal. A meta-analysis of group goal-setting found that specific, challenging group goals produced considerably higher group performance than vague goals (d = 0.80). The same analysis found that individual goals aimed only at personal performance could hurt the group (Kleingeld et al., 2011). In a study of 71 bank teams, clarity about goals and about how work gets done went together with stronger team confidence and performance (Hu & Liden, 2011). More broadly, teams whose members share an understanding of the task coordinate and perform better (DeChurch & Mesmer-Magnus, 2010).
Psychological safety is a shared belief that it is safe to take interpersonal risks in the team: to ask, admit a mistake or disagree. In the study that defined the concept, teams with more of it showed more learning behaviour, and that learning was linked to better performance (Edmondson, 1999). A meta-analysis of 136 samples confirmed links between psychological safety and learning, engagement and performance (Frazier et al., 2017). Trust has an equally clear record: across 112 studies, trust within the team was related to team performance (ρ = .30) (de Jong et al., 2016). Cohesion is also related to performance, most strongly when the work requires members to depend on each other (Beal et al., 2003).
People differ in how they like to work, and team make-up matters. A meta-analysis found that a team's composition on specific personality traits, values and abilities is related to its performance in real workplaces (Bell, 2007). Teams also perform better when members share what they know and build a shared understanding of the work (Mesmer-Magnus & DeChurch, 2009; DeChurch & Mesmer-Magnus, 2010), and teams whose members have worked together more often perform better (Huckman et al., 2009). We are careful here: the research supports building mutual understanding, but we are not aware of controlled studies that test the specific "user manual" format used in this unit.
How a team plans, coordinates and monitors its work is consistently related to its performance and to members' satisfaction (LePine et al., 2008). Teams also do better when members actually share the information each of them holds (Mesmer-Magnus & DeChurch, 2009). Unclear roles are costly. Among seven common work stressors, role ambiguity was one of the two most strongly linked to lower individual job performance (Gilboa et al., 2008). Reviews of team-building research report that interventions focused on clarifying roles and setting goals were among the more effective forms (Shuffler et al., 2018; Klein et al., 2009).
Feedback helps on average, but not always. A landmark meta-analysis found that feedback improved performance overall (d = 0.41). It also found that in more than a third of the comparisons, feedback made performance worse, especially when it turned attention to the person rather than the task (Kluger & DeNisi, 1996). 360-degree feedback on its own produces only small average gains. Improvement is most likely when people accept the need to change, set goals and take specific actions (Smither et al., 2005). In one study of upward feedback, raters who stayed anonymous felt more comfortable and rated less generously than those who were identified (Antonioni, 1994). Peer Feedback is designed around these findings. Ratings are shown without the raters' names and each report is private to the person it is about. The conversation focuses on behaviour and on one thing each person will try, with teammates offering support.
How a team handles disagreement matters at least as much as how much it disagrees. Conflict that becomes personal is consistently linked to lower team performance and satisfaction (De Dreu & Weingart, 2003; de Wit et al., 2012). Disagreement about the work itself is not reliably harmful, and it can help, especially when it stays separate from personal conflict (de Wit et al., 2012). A meta-analysis of 45 studies found that how teams handle their differences explained an additional 13% of the variance in team performance, beyond how much conflict they had (DeChurch et al., 2013). This unit therefore looks at the team's approaches to disagreement, not at who disagrees.
Checking progress is part of what makes change happen. A meta-analysis of 138 experiments found that interventions which prompted people to monitor their progress improved goal attainment (d = 0.40). The effect was larger when progress was recorded, or reported to others or made public (Harkin et al., 2016). The follow-up survey asks the same questions again for the units the team chooses and shows how members now describe the team, so the team can decide what to keep.
(a) In the team's real work, not at an offsite. A meta-analysis of 89 studies found that a supportive work environment is linked to whether training carries over into the job. The link was stronger for interpersonal skills such as leadership than for technical ones (Blume et al., 2010). In leadership training, programmes held on-site showed stronger effects on organisational results than off-site ones (Lacerenza et al., 2017). Debriefs worked in real work settings as well as in simulations (Tannenbaum & Cerasoli, 2013). BBT's surveys and meetings use the team's own data and real situations.
(b) Spread over weeks, not crammed into one day. Across 63 studies, spaced practice produced better performance than massed practice (d = 0.46) (Donovan & Radosevich, 1999). In leadership training, programmes spread over several sessions showed larger effects on on-the-job behaviour (δ = .92 vs .45) and on results (δ = .48 vs .18) than single-session programmes (Lacerenza et al., 2017). These compare different programmes, so they point the way rather than prove one ideal schedule. The journey runs roughly one unit a week for about six weeks.
(c) The team discusses its own results in a structured meeting. Debriefs, in which a team reflects on its own recent performance, improved performance compared with control groups (d = 0.67) (Tannenbaum & Cerasoli, 2013). Randomised studies of guided team reflection found a medium-sized effect on performance (g = 0.55) (Lines et al., 2021). One experiment with student teams found that team feedback improved performance only when it was paired with a guided team discussion of that feedback (Gabelica et al., 2014). Six of the seven units end in a facilitated meeting on the app's meeting page, with a set structure: look back, explore the results, agree commitments; the follow-up has its own meeting, run from its facilitation guide.
(d) Specific commitments: who does what, by when. Across 94 independent tests, plans that link a specific situation to a specific action ("when X happens, I will do Y") had a medium-to-large effect on reaching goals (d = 0.65) (Gollwitzer & Sheeran, 2006). BBT asks for a concrete action, an owner and a review date. Naming the moment that will prompt the action ("when we close a project meeting, we confirm who owns each next step") brings a commitment closest to what was tested. Group goals focused on the team's shared result help performance; goals focused only on individual performance can hurt it (Kleingeld et al., 2011). Making commitments also tends to change behaviour, though results vary across studies (Lokhorst et al., 2013). A caution we take seriously: in some studies, announcing an aim about the kind of person you want to be made people less likely to follow through (Gollwitzer et al., 2009). That is one more reason BBT asks for concrete actions rather than statements of intent.
(e) Weekly follow-up and a closing survey. Monitoring progress improves goal attainment, especially when the progress is written down or shared (Harkin et al., 2016). Improvement after feedback is most likely when people set goals and take action on them (Smither et al., 2005). Each meeting opens by reviewing last time's commitments, and the follow-up survey closes the loop.
(f) Structures that make honesty easier. People speak up more, and teams learn more, when interpersonal risk feels safe (Edmondson, 1999; Edmondson & Lei, 2014). Anonymous raters feel more comfortable and rate less generously (Antonioni, 1994). Feedback works worse when it feels like a judgement of the person (Kluger & DeNisi, 1996). BBT reports team surveys only as combined results, keeps Peer Feedback reports private to each person, and builds meetings around patterns rather than individuals.
How to read the scores. The team surveys are short by design, and some topics rest on a single question. They are conversation starters, not precise measurements, and their reliability and validity have not yet been independently tested. The three-answer minimum protects anonymity; it does not make a small team's average more precise. A change on the follow-up survey shows how members now describe the team. On its own, it does not prove that the journey caused the change or that performance improved. The original IPIP-NEO-120 has published validation evidence; we do not assume it carries over unchanged to the adapted version.
The research gives a credible basis for the design: clear goals, structured reflection, specific actions and follow-up. The studies above describe interventions similar to parts of the journey, tested with other teams, often in health care, the military, aviation, laboratories or classrooms. They show what tends to work on average; they do not guarantee a result for any particular team, and they are not studies of Build Better Teams. Some findings come from individuals rather than teams, and we say so where it matters. Several effects were smaller in the most rigorous studies, and some, like feedback, can backfire when done badly. We have designed BBT around these cautions as well as the positive findings. As our founding teams complete the follow-up survey, we intend to share what we learn, keeping what teams report separate from claims about the journey's impact.
Seven units over about six weeks: six end in a team meeting and something to put into practice, and the seventh, the Follow-up Survey, shows what changed.