Priorities, not tables
Rank drivers by importance and actionability so the conversation moves quickly from evidence to decisions.
Parallax Product 01
Market research intelligence for teams that need to know what truly moves perception, preference, and choice.
Parallax Lens turns a curated research base into a navigable diagnostic system. Instead of reading another static report, teams can move through audiences, brands, time periods, segments, outcomes, and predictor sets to see which drivers matter, how each entity performs, where gaps open, and which actions deserve attention.
Commercial Lens
Parallax Lens gives leadership teams a structured way to answer the commercial questions that usually get buried inside decks: what is driving the metric, where the gaps are, which segments behave differently, and what deserves action now.
Rank drivers by importance and actionability so the conversation moves quickly from evidence to decisions.
Compare performance by entity, time, or segment while keeping the importance structure fixed.
Save screens into booklets and export analysis-ready Excel or presentation slides for the next meeting.
For Marketing Analysts
Lens is built for insights teams, brand strategists, and research consultants who need to explore quickly without losing methodological discipline.
Rank the variables that explain the selected outcome, with separate readings for total importance and actionable importance.
Keep importance fixed while opening performance by entity, time, or segment.
Map importance against performance to reveal weaknesses to prioritize and strengths to defend.
Explore how changes in predictor levels could move the outcome under the fitted model.
Track driver structures across time or compare lenses by audience, outcome, entity, segment, or predictor set.
Turn a business question into a structured investigation that scans stored cells and assembles reusable screens.
Methodology
Lens starts by making the research structure explicit: project grain, stacked entity, filters, time rules, entity scopes, weights, dependents, independents, image nets, eligibility rules, and seed predictor sets.
That contract becomes the fuel for the analysis tree. The interface does not guess what a project means. It reads declared roles, validates selections, and composes views from reproducible cells.
Normalize the survey base, declare the stacked grain, define weights, classify variables, document bands, and audit coverage.
Precompute standard cells across the declared dimensions and keep on-demand slices available when the question changes.
Use typed screens with deterministic composition ids so analyses, favorites, booklets, and exports remain reproducible.
Separate commercial action from statistical artifacts, preserving model notes, signed coefficients, correlations, and fit statistics.
Continuous outcomes use weighted linear regression. Binary outcomes use Firth logistic regression, with accuracy and Tjur R2 carried where appropriate.
Importance is ranked by each predictor's RWA share of the model base measure: R2 for linear models and Tjur R2 for logistic models.
The engine stores signed coefficients. The interface can flag negative or non-actionable effects without rewriting the underlying model output.
A driver must clear the importance threshold, have a positive coefficient, and show positive zero-order correlation with the dependent outcome to be treated as actionable.
Lens preserves correlation structure and model details so users can interpret correlated predictors with the right caution.
Binary funnel outcomes are protected from near-tautological predictors that would dominate a logistic decomposition without adding useful strategic signal.
Every cell is addressed by declared dimensions. Every screen can be saved, recovered, exported, and pinned to analysis through deterministic composition.
Parallax Lens is for teams that want to interrogate their market, preserve methodological discipline, and leave with decisions they can defend.