Every figure the site quotes lives on this page, never rounded and never reworded, with the study it comes from and one line on what it does and does not let Impart say. Everything else on the site links here rather than printing a number of its own.
Peer-reviewed findings with effect sizes. On this page they are the rows: a figure, the plain-language claim, the technical claim, a linked source, and what the figure licenses.
Pattern recognition from having started over and helped others start. Real, and the actual differentiator, but it is not research and is never dressed as research. On this page it always carries this gold rule and this label.
That is how much mindset training on its own changes results, once the studies that were never published are counted. It is close to nothing. When the experiment was run again carefully, beliefs genuinely changed and results did not move.
Macnamara and Burgoyne (2022), Psychological Bulletin, meta-analysis of growth mindset interventions: d = 0.05, and non-significant after adjusting for publication bias. Sisk, Burgoyne, Sun, Butler and Macnamara (2018), Psychological Science: overall effect d = 0.08, mindset-achievement correlation r = .10. Li and Bates (2019), Journal of Experimental Psychology: General: the interventions changed what people believed but produced no effect on resilience, cognitive ability, or attainment.
Saying that mindset alone tests at nearly zero. It does not license any framing that leans on Dweck, growth mindset, or believe it and achieve it. Anyone informed discounts everything that follows that framing.
That is the same belief, written as a specific plan for what you will do and when. Take two people with the same goal. The one with an if-then plan does better than roughly 74% of the ones who set that goal without one.
Gollwitzer and Sheeran (2006), meta-analysis of implementation intentions: d = 0.65 across 94 independent tests and more than 8,000 participants. The average person using an if-then plan lands at roughly the 74th percentile of people who set the same goal without one. The mechanism: the plan initiates striving, shields it from disruption, enables disengagement from failing courses of action, and conserves capacity.
Saying that behavior-specified plans substantially outperform goals alone. It does not license a percentage outcome for any Impart user. The figure describes the studies, not this instrument.
These are the same finding. Belief change on its own moves almost nothing. Belief change specified as behavior moves a great deal. The gap between d = 0.05 and d = 0.65 is the entire space Impart operates in.
So the line is: I'm not selling mindset. Mindset alone tests at nearly zero. I'm selling the thing mindset is missing, which is the part that tests at 0.65.
The audit scores four elements. They do not have the same amount of evidence behind them, and saying which has the most is part of being honest about the instrument. Strongest first.
The figure is the 0.65 above. Nothing is added to it here.
The Task element measures exactly the gap the literature says is decisive: whether stated belief is matched by behavior. The prescribed first move, run the smallest possible action that only makes sense if the belief is true, this week, is an implementation intention in plain language.
People who spent time thinking about how they were working did 18% better on a second attempt than people who spent that same time on more practice. Reflection did not beat doing nothing. It beat more practice.
Di Stefano, Gino, Pisano and Staats, Learning by Thinking: How Reflection Aids Performance, Harvard Business School working paper 14-093, revised on SSRN as Learning by Thinking: How Reflection Can Spur Progress Along the Learning Curve and also circulated as Making Experience Count. Participants who reflected on their strategy performed 18% better on a second attempt than peers who spent that same time on additional practice. Time taken away from the work and given to thinking about the work produced better performance than the work did.
Saying that structured reflection beats spending the same time on more practice. It does not license a figure for what reflection does inside Impart's own instrument, which has not been measured.
Who you know changes where you end up, and this is the rare place where the evidence is an actual experiment rather than a correlation. The people who move you forward are usually not your closest ties.
Rajkumar, Saint-Jacques, Bojinov, Brynjolfsson and Aral (2022), Science, A causal test of the strength of weak ties: a randomized experiment on LinkedIn covering 20 million people over 5 years. Weak ties were confirmed as more useful for advancement. The relationship is non-linear: moderately weak ties beat both strong ties and very weak ties. Weak ties matter most in digital sectors; in less digital industries, strong ties do more. Granovetter (1973), The Strength of Weak Ties: in a survey of 282 people, weak ties outperformed strong ties for finding employment. Burt, Structural Holes: people bridging distinct clusters generate better strategic ideas.
Saying that network composition causally affects advancement. It does not license the audit's current question, which asks about people ahead of you, as the tested structure. That maps to tie quality; the evidence is about moderately weak ties that bridge clusters. Related, not identical, and a future question revision.
Belief and identity work on its own has thin support; see the 0.05 above. What does hold is career adaptability, which a meta-analysis of 90 studies ties to career identity, employability, promotability, income, engagement, satisfaction and performance.
Rudolph, Lavigne and Zacher (2017), Journal of Vocational Behavior, volume 98: meta-analysis of 90 studies. Career adaptability is significantly associated with career identity, employability, promotability, income, engagement, satisfaction and performance.
Saying that adaptability, not belief, is the part of Character with evidence behind it. It does not license Character as the whole answer.
How adaptable someone is and how confident they are in their own career decisions move together, moderately and consistently.
Stead, LaVeck and Hurtado Rúa (2021), Journal of Career Development, volume 49: multivariate meta-analysis of 18 studies and 6,339 people. Estimated correlations between career adaptability subscales and career decision self-efficacy: r = .36 to .44.
Saying that the two are related. It does not license saying that one causes the other. A correlation is not a mechanism.
Character is not load-bearing by itself. It earns its place as the thing that makes the Task move specific and personally correct. Positioned as the whole answer, it is the weakest claim Impart makes. Positioned as the input to a behavioral move, it is well supported.
The single most useful finding on this page, because it says most self-assessment tools are a coin flip and explains what separates the ones that work.
That is how often feedback made people perform worse, across 607 measured results. What decided it was where the feedback pointed. Aimed at the task, it helps. Aimed at the person, it hurts.
Kluger and DeNisi (1996), Psychological Bulletin, meta-analysis of feedback interventions: 607 effect sizes, 23,663 observations. Feedback improves performance on average, d = 0.41. Over one third of feedback interventions made performance worse. Effectiveness decreases as attention moves up toward the self and away from the task. The site prints over one third as 1 in 3.
Saying that most feedback tools are a coin flip, that a third make things worse, and that task-level feedback is what works. It does not license saying that Impart's own result improves performance. That has not been measured.
Most assessments fail because they tell you about yourself. This one tells you about your next action.
That is a design principle and a marketing line at the same time. It is why the audit returns one focus area rather than four scores to brood on, a first move rather than a profile, implications for the lowest element only, and language about behavior rather than worth.
The strengths literature runs against the lowest score rule on the surface. Anyone who has read StrengthsFinder will raise it, so it is answered here.
Strengths use correlates with performance. Deficit correction largely does not.
van Woerkom, Mostert, Els, Bakker, de Beer and Rothmann (2016), strengths use and deficit correction · Bratty and Dennis (2024), strengths interventions and workplace outcomes
Gallup's engagement, productivity and performance figures are proprietary in-house research, not independent peer review. They are not printed anywhere on this site.
Every system has at least one constraint that sets the ceiling. Improving anything other than the constraint does not improve the system.
Goldratt (1984). The five focusing steps: identify the constraint, exploit it, subordinate everything else to it, elevate it, then repeat, because a new constraint appears.
Saying that constraint logic is standard practice in operations and applies to individuals under the same conditions. It does not license applying it above threshold, where the strengths literature is the better guide and Impart should say so rather than oversell.
The same literature concedes the case that matters here: critical gaps that prevent basic function should be addressed directly. Strengths logic and constraint logic answer different questions. Strengths research studies optimization: someone already functioning, trying to perform better, usually in a team where a weakness can be delegated or routed around. Impart's frameworks address constraint: an individual who is stuck, where there is nobody to delegate to and the weak element is not a preference but a gate. A zero is not a weakness. It is a gate.
The retake cadence is the fifth focusing step, arrived at independently. And the audit already makes the switch. When all four elements score 75% or more, the result changes from Where the chain breaks to Your weakest link, and it holds, and reframes the lowest element from repair to maintenance. That is the strengths-versus-constraints switch, already implemented.
Impart is a set of coaching frameworks. Nothing on this site or in the instruments claims a clinical, diagnostic, or therapeutic benefit, and none has been tested for one. It is not a substitute for care.
Every figure on this page describes a component: a plan, a reflection, a network structure, where feedback points. Impart's own instrument, which assembles them, has zero outcome data today. The parts are established. The instrument has not been validated as a system.
I did not invent these. I found them, tested them on myself, and built the system that tells you which one you need right now. The parts are established. The routing is mine.
Figures on this page are transcribed from the Impart Evidence Base, dated 2026-08-17, and were checked against the sources linked above on 2026-09-06. If a figure on any other page of this site disagrees with this one, this page is right and the other page is the bug.
