Skip to content
    Marres Insights
    Documentation contents

    DOCS / THE ANALYSIS WORKFLOW

    The analysis workflow

    The end-to-end evidence-to-decision process.

    Marres organises work around an Analysis File. The file keeps the research question, business context, evidence, models, scenarios, decisions, notes, history, and exports connected.

    1. Frame the decision

    Begin with the decision to be made, not the tool to be run. Record:

    • the focal question;
    • the decision owner;
    • the relevant time horizon;
    • the organisation and competitors in scope;
    • known constraints; and
    • the alternatives currently being considered.

    This frame can change as evidence arrives. Material changes should be recorded in the analysis history.

    2. Establish the baseline

    Operational datasets can be used to identify co-moving business dimensions and create conservative no-treatment baselines. Column roles are proposed, then reviewed by a person before analysis. Components describe covariance geometry; they are not automatically business concepts or causes.

    3. Build the evidence layer

    Public reviews, social material, uploaded records, and analyst-supplied context can be organised into explicit domains. Marres records coverage and usable evidence alongside scores so an apparently precise average is not mistaken for strong evidence.

    4. Model mechanisms

    Selected evidence profiles can be compiled into a comparable market dataset. The modelling layer examines opportunity, maturity, relationships between domains, competitive regimes, and diminishing returns. Validation blocks incomplete or duplicate subjects where comparison would be invalid.

    5. Explore uncertainty

    Deep-uncertainty analysis generates many plausible futures and groups similar outcomes. The aim is not to attach false precision to an unknowable future. It is to discover which actions remain acceptable across different conditions and which fail under identifiable scenarios.

    6. Simulate response

    The NS Loop stages repeated campaign-cycle decisions between a human-controlled organisation and model-controlled competitors. Each turn preserves context, records predictions, updates market state, and supports later calibration against observations.

    7. Synthesize the decision

    The Decision Console and Marres Brief bring the evidence and modelling outputs together. The final recommendation remains an analyst judgement. It should state the selected action, evidence supporting it, key sensitivities, failure conditions, and what should be monitored next.

    Traceability

    Runs, outputs, artefacts, notes, and history remain attached to their analysis. Users should prefer linked artefacts over copied values because linked artefacts retain provenance and can be revisited when assumptions change.

    Related product page: Learn about Marres

    Documentation draft · Source baseline 16 August 2026