What are customers worth?
WorthSignal
Customer value, retention and marketing ROI—turned into guided decisions instead of another spreadsheet archaeology project.
Marketing analytics product builder · Norway
I'm Ulrik Erlingsen, creator of Signal—a growing suite of local-first analytics products built for clear decisions and honest scrutiny.
Explore the workWhat are customers worth?
02Who is meaningfully different?
03What do customers value?
04Where does the brand stand?
05When will the market follow?
06Where should the next budget go?
07What moves with satisfaction?
08Go, hold, rework, or kill?
09Did the treatment cause a change worth acting on?
10Is this score measuring what you think it is?
11What are people actually saying—and does the pattern hold?
12What should it cost?
13Which recommender earns the slot?
14Where do journeys flow, stall, and end?
15Is the brand moving, or is the tracker just noisy?
Selected work · The Signal suite
Make the hard part visible. Name the method. Keep the caveat beside the answer. Give a marketer a useful first read—and an analyst enough evidence to challenge it.
Three products launched under new names after their working titles failed a publication-name screen. Method quality does not override a failed name screen—products get renamed instead.
What are customers worth?
Customer value, retention and marketing ROI—turned into guided decisions instead of another spreadsheet archaeology project.
Who is meaningfully different?
Customer segmentation that compares methods, checks whether memberships survive resampling, and is allowed to say no reliable solution exists.
What do customers value?
Ratings-based conjoint analysis with respondent-level part-worths, design checks, attribute importance and preference scenarios.
Where does the brand stand?
Auditable positioning across brands, waves and segments—with map fidelity, association ownership and POP/POD evidence kept beside the picture.
When will the market follow?
Bass diffusion planning from published analogies, history fitting and explicit warnings when a pre-peak forecast cannot identify the ceiling.
Where should the next budget go?
Diminishing-return planning, real constraints, panel evidence and a digital-economics audit for marketers who want to challenge an optimized media plan.
What moves with satisfaction?
Survey driver analysis that separates relative importance from conditional direction—and turns the ranking into a testable hypothesis.
Go, hold, rework, or kill?
New-product gate support that separates criteria, evidence, constraints and economics—plus extension, alliance and reputation exposure.
Did the treatment cause a change worth acting on?
Decision support for randomized experiments with continuous or binary outcomes, design checks, robust intervals, multiplicity control and practical thresholds.
Is this score measuring what you think it is?
Exploratory measurement diagnostics with factor structure, reliability, a frozen holdout plan and an invariance gate before cross-group score comparisons.
What are people actually saying—and does the pattern hold?
Open-text evidence with corpus and duplication audits, structured comparisons, stable NMF patterns, validated sentiment checks and a human codebook handoff.
What should it cost?
Pricing evidence from three honest routes—price experiments, sales history or willingness to pay—kept separate and fed into margin economics with the uncertainty intact.
Which recommender earns the slot?
Offline evaluation of recommendation policies with temporal splits—accuracy, coverage, novelty, concentration and cold-start evidence instead of one magical score.
Where do journeys flow, stall, and end?
Descriptive journey evidence from real event logs—transitions, path support, drop-off and positional roles, with a transparent Markov sensitivity check and no causal credit claims.
Is the brand moving, or is the tracker just noisy?
Brand-tracking waves compared with intervals, multiple-comparison control and declared practical thresholds—so real movement is separated from sampling noise before anyone reacts.
How I build
The interface starts with the decision, not the algorithm. The method earns its place by making that decision clearer.
Named models, visible assumptions, cited sources, synthetic demos and portable evidence—not a mysterious score.
Diagnostics, limits and failure states sit beside the result. A tool should be able to say “the evidence is weak.”
Creator of the Signal tools
About Ulrik
“I combine marketing strategy, applied analytics and product design to make classic methods more usable, while keeping their assumptions visible.”
The Signal suite is an independent, AI-assisted body of work. Each product is documented, tested against analytical fixtures or synthetic data, and built to run locally without accounts, telemetry or hidden data storage.
I care about the space between a strategy lecture and a real decision: where a sound method becomes a tool someone can actually operate, understand and question.
Follow the work
Follow public releases or start a conversation about marketing analytics and product design.