I’m Ulrik. I build tools that turn marketing theory into something you can explore, understand and use.
Follow the questions ↓︎Marketing is a sequence of questions. Listen to people. Understand the differences. Shape the offer. Make a decision. Learn what happened. I’ve built a tool for fifteen of those questions.
A comment. A survey response. A satisfaction score. Before making a decision, understand what people are saying—and whether the evidence holds up.
Turn open-text responses into patterns you can inspect. Explore themes and comparisons while keeping the original evidence within reach.
Checks duplication, compares language across groups, and tests whether patterns hold. Sentiment checks and a human codebook handoff keep interpretation connected to the text.
Examine whether survey items measure a coherent idea before comparing the resulting scores.
Explore factor structure and reliability with a frozen holdout plan. An invariance gate comes before cross-group score comparisons.
Investigate which factors relate to satisfaction and turn their relative importance into a testable hypothesis.
Separates relative importance from conditional direction. A ranking can guide a new hypothesis without claiming it proves cause and effect.
Once the evidence is clearer, the next question is who it describes.
An average can hide a lot. Look at customer groups, their value over time, and the journeys that bring them closer—or lead them away.
Find customer groups and check whether those groups survive a closer look. Useful segments need more than a convincing chart.
Compares methods and checks whether memberships survive resampling. It is allowed to conclude that no reliable segmentation exists.
Bring customer value, retention and marketing ROI into a clearer view of where to focus.
Connects customer value, retention and marketing ROI in guided decisions. The question is how customer relationships develop beyond an individual purchase.
Follow transitions, drop-off and recurring paths through event logs to understand how journeys unfold.
Keeps path support and positional roles visible. A transparent Markov sensitivity check explores the pattern without making causal credit claims.
Understanding people creates a better starting point for shaping the offer.
What matters to a customer? Where does a brand stand? What should it cost? These are connected questions, each deserving its own evidence.
Explore how product attributes shape preferences and what changes when an offer changes.
Ratings-based conjoint analysis provides respondent-level part-worths, design checks, attribute importance and preference scenarios.
Look at brand positioning across brands, waves and segments, with the evidence beside the map.
Keeps map fidelity, association ownership, and points-of-parity and points-of-difference evidence available alongside the positioning picture.
Explore pricing evidence and margin economics without losing sight of uncertainty.
Price experiments, sales history and willingness-to-pay evidence follow three separate routes. Each feeds into margin economics with uncertainty intact.
A promising offer still needs a practical plan to reach the market.
Markets develop over time. Budgets have limits. New ideas compete for resources. Explore the options before committing to one.
Explore diminishing returns, real constraints and alternative media plans. Keep the assumptions close to the recommendation.
Combines diminishing-return planning, panel evidence and a digital-economics audit so an optimized media plan can be challenged.
Explore adoption scenarios using published analogies or historical observations.
Supports history fitting and planning from published analogies, with explicit warnings when a pre-peak forecast cannot identify the market ceiling.
Bring criteria, evidence, constraints and economics into the same new-product decision.
Separates gate criteria from evidence and economics, while considering extension, alliance and reputation exposure.
Then comes the part that keeps the work honest: finding out what changed.
Did an experiment change something meaningful? Which recommendation policy earns its place? Is the brand moving—or is the tracker simply noisy?
Evaluate randomized experiments with practical thresholds, design checks and uncertainty kept in view.
Supports continuous and binary outcomes, multiplicity control and practical thresholds. The decision goes beyond whether a result passes a statistical cutoff.
Compare recommendation policies across accuracy, novelty, concentration and cold-start evidence.
Uses temporal splits for offline evaluation and brings different dimensions of recommendation quality into the comparison.
Compare tracking waves and distinguish meaningful movement from sampling noise.
Combines intervals, multiple-comparison control and practical thresholds before reacting to changes in brand-tracking waves.
Fifteen tools. One practice: make the question clear, show the workings, and keep learning.
I’m Ulrik Erlingsen. My work brings marketing strategy, applied analytics and product design into the same practice.
I independently design and build Signal: local-first products with documented methods and inspectable evidence. The interface is part of making a question easier to work with.
Start with the decision.The question gives the method a purpose.
Show the workings.Keep assumptions within reach.
Keep uncertainty visible.Leave room to challenge the answer.