Customer insights · Evidence quality · Investment decisions
A clear customer story can still support the wrong investment.
Arcus helps leaders turn customer feedback, behaviour and commercial evidence into defensible decisions. We examine whose experience is represented, what explains the pattern and which action is worth testing before the organization commits resources.
The decision before the program
01 / Whose needs are missing from the evidence?
02 / Does the finding describe an opinion, a behaviour or a cause?
03 / What would change the investment decision?
When positive feedback conflicts with commercial results
Imagine a fictional service business whose customer survey shows strong interest in a premium support package. Sales remain weak after launch. Management must decide whether to invest in promotion, change the package or stop the rollout. Each response assumes a different explanation for the gap.
Arcus’s central thesis is that an insight earns its commercial value by surviving a challenge to its coverage, interpretation and implications for action. A persuasive quotation or a large survey can reveal something important without establishing that a proposed investment will work.
This service focuses on the judgement between collecting evidence and committing resources. The question is not simply what customers say they want. It is which explanation best fits the available evidence, what remains unknown and how the organization can resolve the uncertainty at a reasonable cost.
Four explanations for a misleading customer signal
Investigate competing explanations before deciding that the market, the product or the research has failed.
01 / The wrong audience answered
Existing enthusiasts, frequent users or people with strong complaints may be easier to reach. Non-buyers, departed customers and people facing access barriers may be missing. Compare the research audience with the population affected by the decision.
02 / The question measured something else
A positive reaction may reflect politeness, an attractive description or an imagined situation without a real budget constraint. Check wording, context, competing offers and what respondents understood they were evaluating.
03 / The behaviour has another cause
Purchases may reflect availability, sales effort, timing or contractual requirements rather than the feature customers mention. A correlation or regression coefficient can guide investigation without identifying the effect of changing that feature.
04 / The organization cannot deliver the response
Customers may value the proposition while the buying process, service capacity or channel incentives prevent adoption. Examine the complete experience and delivery requirements before commissioning another communication campaign.
Evidence that challenges comfortable assumptions
Dutz and colleagues’ research on survey nonresponse links a labour-market survey with administrative data for participants and nonparticipants. It finds substantial differences that persist after adjustment for observed characteristics, and uses randomized participation incentives to examine nonresponse. The setting is not a customer study; the relevant lesson is that observable demographic balance alone need not establish representativeness.
Schmidt and Bijmolt’s 2020 meta-analysis of willingness-to-pay measurement compares hypothetical and real valuations across 77 studies. It finds systematic hypothetical bias and variation by method and context. It also challenges the assumption that an indirect method is necessarily more accurate. These findings do not justify applying a universal correction factor to a client survey.
Arcus’s interpretation is to assess coverage and measurement separately. Better recruitment cannot repair a question that measures the wrong construct, and a well-designed question cannot reveal the views of people absent from the research. Neither improvement, by itself, proves the commercial effect of management’s proposed action.
The Arcus insight-to-decision diagnostic
Use this original framework for one finding and one pending decision. It is a management diagnostic, not a statistical certification.
01 / Coverage
Define the decision population and who is actually represented. Examine recruitment, response, missing groups and the buying unit. Document what can and cannot be inferred beyond the participants.
02 / Meaning
State precisely what was observed: a reported need, a stated preference, an actual purchase or a service outcome. Check alternative interpretations, contradictory evidence and the conditions under which the finding holds.
03 / Consequence
Connect the finding to a specific choice, expected behaviour and commercial or service outcome. Identify the assumption that would reverse the recommendation. Separate attractive segments from segments the organization can reach and serve.
04 / Commitment
Choose the next action in proportion to uncertainty and reversibility. Specify an owner, cost, evidence threshold and review point. Record whether the decision is to implement, test, investigate further or stop.
Worked example: the denominator changes the story
Suppose 120 of 200 survey respondents express interest in a new service: 60% of respondents. If 1,000 eligible customers were invited, those 120 people represent 12% of the invited population. Neither figure establishes the interest of the 800 nonrespondents, and neither is a conversion forecast.
Assume a proposed launch has $50,000 in fixed costs and would contribute $100 per additional customer before those fixed costs. It needs 500 additional customers to cover that cost under the stated assumptions. Extrapolating 60% interest to all 1,000 customers would suggest 600 interested people, but it combines an unsupported nonresponse assumption with an unsupported interest-to-purchase assumption.
The practical next step could be targeted research among missing groups followed by a limited, deliverable offer with observable purchase behaviour. The aim is to test the assumption that matters for the launch decision. These fictional figures illustrate reasoning, not expected response rates or a recommendation to contact customers without appropriate permission.
Choose the method that resolves the uncertainty
A useful research program can combine methods. Each has strengths, costs and limits.
Interviews and observation
Useful for understanding circumstances, language, unmet needs and friction. Require careful recruitment and interpretation. They reveal explanations to examine, rather than population prevalence or a guaranteed demand estimate.
Surveys and structured choice research
Useful for comparing needs and trade-offs across a defined audience. Require an appropriate sample, tested questions and transparent assumptions. Larger samples reduce some random uncertainty but do not automatically remove selection or measurement bias.
Behavioural and service analysis
Useful for examining what happened, where journeys break down and which customers encounter problems. Require reliable definitions and appropriate data access. Existing behaviour reflects the options and constraints customers faced; it does not reveal every unmet need.
Pilot or controlled comparison
Useful for testing whether a feasible change affects behaviour under specified conditions. Requires delivery capacity, a suitable comparison and enough observations. Results may not transfer to another audience, season or scale; an inconclusive test should remain inconclusive.
Make the evidence usable in management decisions
Begin with the decision date, alternatives, cost of being wrong and reversibility. A modest operational change may justify a small exploratory study, while an expensive product launch may require stronger behavioural evidence and commercial scenarios. More research is valuable only if it can improve the decision enough to justify its cost and delay.
Set the recruitment and analysis plan before interpreting results. Include people who did not buy or could not complete the journey where relevant. In business-to-business settings, distinguish users, purchasing staff, budget owners and approvers: one person’s enthusiasm may not represent the buying organization.
Review question wording, missing data, small subgroups and conflicting findings. Avoid naming an observed association a causal driver without a defensible identification strategy. Weighting and modelling can be useful, but their assumptions and limitations should be visible to decision-makers.
Budget for recruitment, accessibility, research design, fieldwork, analysis and operational testing. Establish appropriate permission, privacy, access and retention arrangements before linking customer records. Give the responsible business owner a clear route to act on findings and report back on what changed.
How Arcus helps turn customer evidence into action
The engagement can begin with existing research, a disputed interpretation or a new commercial question.
01 / Frame the decision
A decision brief identifying alternatives, stakeholders, critical assumptions and evidence gaps. Leaders agree what information could materially change the choice.
02 / Gather and challenge evidence
A proportionate research plan, customer or stakeholder work and analysis of available behavioural evidence. Deliverables distinguish observations, interpretations and unresolved competing explanations.
03 / Compare the options
An insight synthesis, segment or journey implications and practical alternatives with contribution scenarios, costs and implementation requirements. Recommendations state where confidence is strong and where testing remains necessary.
04 / Support implementation
A pilot or action roadmap, responsibilities and review criteria. Arcus helps teams translate customer findings into changes that can be delivered and evaluated, rather than leave the work as a presentation.
Discuss the decision behind your customer research
Tell Arcus what decision is pending, which evidence you already have and why the answer remains uncertain. We can scope a review of existing research, targeted customer investigation or a broader program connecting insight with implementation. Begin with the business question rather than identifiable customer records.
Related services include market research, marketing effectiveness and brand strategy. Advisory experience is available through the central consultant directory.
Historical perspectives
The original page featured From Pop Culture to Peep Culture as a perspective on digital culture and social media. It is retained as a historical reading reference, not a current forecast.
The original Brand Strategy and Innovation interview with Geoff Craig discusses a vision and framework for innovation. The Unilever role cited on the original page reflects its historical attribution.