Milestone map
Milestone map
3 milestones
Complete a Full Research Cycle on a Real Question
8–16 weeks (question + method + data collection + analysis + sharing)
Conduct a complete research cycle on a real question — one that is genuinely uncertain and that matters to a real audience (your employer, a publication, a community, or an academic field). The cycle must include: a written research question with scope boundaries (what the research will and will not answer), a methodology note (why you chose this method over alternatives), data collection (primary data: interviews, surveys, experiments; or secondary data: systematic literature review), analysis, and a written findings document with explicit limitations. The research must be shared with the audience — presented to a team, published, or submitted for review.
Proof required
Submit the research question and scope boundaries (1 paragraph), the methodology note (why this method — 150 words), the data collection artifact (interview transcripts, survey results, literature database, or experiment log), the findings document (500–2,000 words depending on scope), and evidence the research was shared (a meeting agenda showing the presentation, a published link, a submission confirmation, or a colleague email confirming receipt).
What gets checked
- Research question has explicit scope boundaries — 'what makes teams effective?' is not a scoped question; 'what management practices correlate with engineering team velocity in companies of 20–100 people?' is scoped
- Methodology note explains why this method and not an alternative — 'I used interviews because they are useful' is not a methodology note; 'I used semi-structured interviews over a survey because the question requires understanding the reasoning behind practices, which closed-response items cannot capture' is
- Findings document includes explicit limitations — 'I only spoke to 8 people so these findings may not generalise' is the minimum; 'this sample is biased toward companies that agreed to participate, which likely excludes teams with the lowest performance — the findings should not be extrapolated to poorly performing teams' is better
Resources
Foundationstart here
Depthgo deeper
What a verifier looks for
- Ask the submitter to state the alternative methodology they considered and rejected — a submitter who cannot name an alternative has not made a genuine methodological choice
- Ask what their findings would have looked like if their pre-existing hypothesis had been wrong — this tests whether the methodology genuinely allows for falsification or is designed to confirm a prior belief
- Ask what the most surprising finding was — genuine research always produces at least one unexpected result; a submitter who found exactly what they expected either conducted confirmatory rather than exploratory research, or did not analyse deeply enough
- Ask what the most important limitation is and how a future researcher should account for it — the quality of the limitations section is the single strongest indicator of research maturity
- A research question that is answered rather than investigated — 'how do I implement X?' is a question with a lookup answer, not a research question; research questions have genuinely uncertain answers before the research begins
- Data collection that confirms the pre-existing hypothesis without genuine openness to alternative findings — the most common research integrity failure is designing data collection to confirm what you already believe
- A findings document that reports what interviewees said without synthesising themes — transcription is not analysis; the findings document must state what the data means, not just what it contains
Publish or Present Research That Influences a Decision
6–12 weeks (commissioned research + output + decision documentation)
Conduct a second research project where the output — a report, a paper, a presentation, an analysis document — is explicitly commissioned by or presented to decision-makers who used the findings to make or inform a real decision. The research must use a different method from M1 (if M1 used interviews, M2 must use survey, experiment, or literature review — or vice versa). Document the decision that the research informed and the specific finding that most influenced it.
Proof required
Submit the research output (the report, paper, or analysis). Submit evidence it was commissioned or presented to decision-makers (an email commissioning the work, a meeting agenda, or a presentation recording). Submit a 150-word impact note: what decision it informed, which finding was most influential, and how you know (a follow-up email, a decision meeting note, or a stakeholder statement).
What gets checked
- Research uses a different method from M1 — not a variation of the same method; semi-structured interviews and focus groups are the same method type; interviews and a quantitative survey are different
- Impact note names a specific decision — 'it helped the team think about X' is not a decision; 'it led the product team to deprioritise the mobile feature from Q3 because the survey showed mobile users were 12% of the target audience, not 40% as previously assumed' is
- The finding that influenced the decision is traceable to the data — 'the team agreed with our recommendation' is not traceable; 'the 67% preference for Option A over Option B in the survey (n=143) was the specific data point the head of product cited in the decision meeting' is
Resources
Foundationstart here
Depthgo deeper
What a verifier looks for
- Ask the submitter to name the specific moment the research finding entered the decision process — a specific moment (the meeting, the email, the slide) is evidence of genuine influence; 'it was part of the context' is not
- Ask what they would have done differently if the decision-maker had rejected the finding — this tests whether the submitter understands the difference between rigorous research and influential communication
- Ask what the research could not tell the decision-maker — the most impactful research outputs are clear about their limits; decision-makers who act on overconfident findings are more likely to be misled than ones who receive explicit limitations
- Confirm the method is genuinely different from M1 — not a superficially different variation of the same method type
- Research that was conducted and published but not used — M2 requires evidence that the research influenced a decision, not just that it was completed and shared; 'we presented it but the decision had already been made' is a valid learning but does not satisfy M2
- Impact note that describes stakeholder satisfaction rather than a decision — 'everyone found it very useful' is not evidence of influence on a decision; a specific decision with a specific finding that changed a prior assumption is the evidence standard
- A second project that is materially less rigorous than M1 — the minimum research standards from M1 (scoped question, methodology note, explicit limitations) apply to M2
Defend Research Portfolio to a Senior Researcher
1–2 weeks (60–90 min session)
Present your M1 and M2 research projects to a qualified reviewer: a senior researcher with 5+ years of applied or academic research experience, or a research director who commissions and critiques research for high-stakes decisions. The reviewer challenges the M1 methodology choice (what would a different method have found that yours missed?), the M2 impact claim (how do you know the research caused the decision rather than rationalised it?), and presents a novel brief: a research question, a 1-week timeline, and a constraint (limited budget, sensitive population, or contested topic) — and the submitter must design a research approach live.
Proof required
Submit your M1 and M2 outputs plus the reviewer's written assessment: the methodology challenge and your response, the impact challenge and your response, the novel brief and your live research design approach, and their sign-off confirming their role and a live session.
What gets checked
- Methodology challenge response names a specific finding that a different method would have surfaced and that your method systematically missed — not 'a survey would have given me more data' but 'a quantitative survey would have given me prevalence data; my interview approach cannot tell me whether the patterns I identified are common or exceptional'
- Impact challenge response names an alternative explanation for the decision and addresses it — 'the research caused the decision' is rarely falsifiable; the honest response names the strongest alternative explanation ('the head of product had already been leaning toward deprioritising mobile; the research provided evidence, not the impetus') and acknowledges the limit of impact claims
- Novel brief live research design names a method, sample, and timeline with explicit trade-offs — 'I'd run some interviews' is not a design; 'given the 1-week constraint, I'd run 5 expert interviews (3 hours data collection, 2 hours synthesis) rather than a survey (which would need 2 weeks for meaningful response rates), accepting that my findings will not be generalisable but will be directionally useful for the decision' is
Resources
Foundationstart here
What a verifier looks for
- The novel brief should include a constraint that requires a methodological trade-off — 'no access to the target population directly, only to intermediaries' or 'the topic is politically sensitive within the organisation' forces the submitter to design under real-world constraints, not ideal conditions
- The impact challenge should be calibrated: don't ask 'do you know if you caused the decision?' (which invites a yes/no); ask 'what is the strongest evidence that the research influenced rather than rationalised the decision?' — this forces the submitter to construct the case rather than defend a claim
- Reviewer minimum qualification: 5+ years of applied or academic research with evidence of influencing decisions based on their research outputs (published papers, internal reports that led to strategic changes, or equivalent)
- After the novel brief design, ask the submitter what they would do in week 2 if the data from week 1 was inconclusive — research under time pressure always hits this point; the response reveals whether the submitter's research design has a decision protocol when data is messy
- Methodology challenge response that defends the chosen method without acknowledging what it systematically cannot find — every research method has structural blind spots; a response without any acknowledged limitation has not engaged with the challenge
- Novel brief live design that starts with the analysis rather than the question — research design begins with clarifying the question, then choosing the method; a submitter who jumps to 'I'd use a survey' before restating the question as a more precise research question is designing a study, not doing research
- A reviewer who conducts but does not commission or critique research — the impact challenge requires someone who has personally evaluated whether research findings caused decisions; this requires experience on the receiving end of research reports