Research Lifecycle

Every engagement starts with a Decision Brief — a short document that names the specific business decision this research must inform, why it matters now, what constraints exist (timeline, budget, organizational politics), and what “success” looks like. No research begins until this document is signed off.

From the Decision Brief, we build a Hypotheses Map: a set of testable beliefs tied to the decision, ranked by risk. These are not vague themes — they're specific, falsifiable claims that the research will prove or refute. The hypotheses determine what we study, what we ignore, and how deep we go.

Research questions are then crafted to test each hypothesis without leading or biasing the findings. This is where most internal research goes wrong: questions are written to confirm what the team already believes. Rubikn's questions are designed to challenge assumptions, not validate them.

The Research Plan specifies methods (interviews, CRM analysis, competitive audits, ad platform review, content analysis, customer surveys), sources, analysis approach, and timeline. Method selection is driven by the hypotheses — not by what's easiest or most familiar.

The Sampling Plan defines which segments, roles, or data populations to study, with quotas and inclusion/exclusion criteria. In B2B, sampling is rarely random — it's purposive. We sample for maximum variation within the relevant population (e.g., churned customers AND power users, not just whoever responds first).

If interviews or surveys are involved, we build a Recruiting Plan with a screener to ensure participants actually match the criteria. If the research is secondary (competitive analysis, CRM pattern mining, ad intelligence), we build a Source Plan documenting where evidence will come from and how recency/reliability will be evaluated.

Instruments are the actual tools used to collect evidence: interview guides, survey questionnaires, competitive claim extraction sheets, CRM analysis templates, evidence tagging schemas. Every instrument is designed to test specific hypotheses without introducing bias.

Before full fieldwork, we run a pilot — a small-scale test of the instrument with 1–2 participants or data sources. The pilot catches ambiguous questions, broken logic, missing response options, and workflow issues before they contaminate the full dataset.

The Pilot Summary documents what broke, what changed, and what's ready for production fieldwork. The Fieldwork Protocol establishes consent procedures, recording standards, versioning rules, and naming conventions so the evidence library is organized from day one — not cleaned up retroactively.

Fieldwork is where research meets reality. Whether we're conducting customer interviews, mining CRM data, auditing competitor messaging, or analyzing ad performance — every data point is captured with consistent tagging, timestamped, and linked to its source.

We track fieldwork status and sample health throughout — monitoring which segments are covered, which are under-represented, and where emerging patterns might warrant adjusting the approach (theoretical sampling). Patterns are flagged as they surface, but conclusions wait until analysis.

Every piece of evidence is stored in an organized evidence library with source attribution, timestamps, and context notes. This traceability is what separates rigorous research from “we looked at some stuff” — and it's what allows the client to audit any finding back to its source.

Before analysis begins, every dataset goes through a QA pass. The QA Log records checks run, issues found, and fixes applied. Common checks include: duplicate detection, source verification, recency validation, and cross-referencing claims against multiple sources (triangulation).

Cleaned data is delivered with a codebook explaining fields, tags, and any transformations applied. Nothing is a black box.

Analysis follows a structured synthesis process. The Insight Map lays out themes, the evidence supporting each theme, and a confidence rating tied back to the original hypotheses. Confidence is graded on a 3-level scale: High (multiple independent sources confirm), Medium (supported but not fully triangulated), Low (single source or indirect evidence).

The Analysis Memo translates findings into implications and clear, prioritized recommendations for the decision at hand. Every recommendation is tied to a specific finding with an explicit confidence level — so the decision-maker knows what's certain, what's probable, and what's a bet.

Research that ends with “interesting findings” is research that failed. Every Rubikn engagement ends with three things: an Executive Readout, an Asset Package, and a Decision Workshop.

The Executive Readout is a tight deck or memo that states the decision, the evidence, and the recommended move — written for the person who needs to say yes, not the person who did the research.

The Asset Package contains tangible tools the team uses immediately: Reality vs. Rhetoric matrices, battlecards, positioning briefs, ICP profiles, growth scorecards, messaging architectures, landing page blueprints — whatever the engagement calls for. These aren't recommendations about what to build. They're the finished assets.

The Source of Truth is a single repository with datasets, transcripts, screenshots, and links so the record is fully traceable. Any finding can be audited back to its source.

The Decision Workshop is a focused session (typically 60–90 minutes) with a structured agenda, facilitation, and an action plan that names owners and next steps. The workshop isn't a presentation — it's a working session where the team makes the decision the research was designed to inform.