Why Do Companies Collect Customer Insight and Still Fail to Act on It?

The company has 14,000 interview notes, 31 dashboards, a weekly voice-of-customer digest, and a searchable repository that can find the phrase “manual workaround” in under a second.

It cannot name the last product decision any of them changed.

This is not unusual because collecting customer insight and acting on it are different organizational capabilities. Collection creates an input. Action requires a relay: evidence must be translated into a causal claim, attached to a live decision, received by an owner with authority, coordinated across functions, delivered as a bounded response, and tested against the customer outcome.

Most insight programs measure the input and assume the relay exists.

It often does not.

Follow one signal until it disappears

Consider a recurring observation from customers of a business software product:

Administrators export approval requests to a spreadsheet before they send them to managers.

The research team codes the behavior as “workflow friction.” Product management adds it to an insights board. Sales recognizes the complaint but says the current quarter is committed. Engineering asks whether the workaround affects retention. Finance asks how much revenue is exposed. Nobody owns the answer to those questions, so the finding stays visible and inert.

Six weeks later, another study rediscovers it.

The failure did not occur in the interview. It occurred in the handoffs after it.

Ajay Kohli and Bernard Jaworski’s foundational formulation describes market orientation through three organization-wide activities: intelligence generation, dissemination, and responsiveness. They drew on 62 field interviews across functions and organizations. Listening was one part of the construct, not its finish line.

That distinction is easy to blur because collection leaves artifacts. Interviews produce recordings. Surveys produce percentages. Dashboards move. Response work is less photogenic: a decision right changes, two functions negotiate a constraint, a team cancels a favored feature, or an experiment reveals that the original explanation was wrong.

The spreadsheet workaround therefore needs a route, not another tag.

Action relay

Six handoffs carry a signal into a verified response

Each stage must produce a usable object and name the owner responsible for the next handoff.

Handoff 01

Signal

Required output

Observed customer behavior

Accountable owner

Research

Open-circuit failure

A label without context

Handoff 02

Translation

Required output

Causal claim plus rival

Accountable owner

Research + domain

Open-circuit failure

A quote treated as a cause

Handoff 03

Decision

Required output

Explicit disposition

Accountable owner

Decision owner

Open-circuit failure

Interest without authority

Handoff 04

Coordination

Required output

Bounded operating change

Accountable owner

Affected functions

Open-circuit failure

A recommendation outside workflow

Handoff 05

Delivery

Required output

Shipped response or test

Accountable owner

Delivery team

Open-circuit failure

A commitment without capacity

Handoff 06

Feedback

Required output

Measured customer outcome

Accountable owner

Decision owner

Open-circuit failure

Activity mistaken for learning

Closure rule. Every stage must produce an object another named owner can use.

The six handoffs are not interchangeable

A complete relay has six stages. Each produces a different object:

1. Signal: What did customers demonstrably do, experience, or avoid?
2. Translation: What causal explanation could account for that behavior, and what rival explanation could also fit?
3. Decision: Which open commitment could the evidence change?
4. Coordination: Which functions must alter work, policy, data, or risk controls?
5. Delivery: What bounded response will the organization ship or test?
6. Feedback: What customer outcome would show that the response worked?

A slide deck cannot substitute for a missing decision. An executive sponsor cannot substitute for delivery capacity. A shipped feature cannot substitute for feedback. Calling all six “insight activation” hides the location of the fault.

The research on cross-functional dissemination makes this especially clear. Elliot Maltz and Ajay Kohli studied 788 nonmarketing managers in high-technology equipment companies. More frequent or more formal dissemination did not produce a simple linear rise in perceived intelligence quality. Both relationships were nonlinear. Joint customer visits, distance between functions, trust in the sender, positional power, organizational commitment, and structural change affected how intelligence traveled.

In plain terms: forwarding the report harder is not a strategy.

Formal channels can help. The study also found a “mere formality” effect in which intelligence received through formal channels appeared to be used more. Yet excess frequency and formality can reduce quality or become ritual. A repository may make a signal easy to retrieve while leaving it difficult to believe, interpret, or apply.

Search breadth does not create integration

Many teams respond to weak action by broadening collection. They add social listening, sales calls, support tickets, product analytics, win-loss interviews, and another quarterly survey.

The result can be a larger pile of incompatible evidence.

Silja Korhonen-Sande tested customer-information processing among 228 manufacturing and R&D managers in large industrial firms. Broader search was positively associated with information integration. It had no direct relationship with information use. Integration—not exposure alone—predicted use in decisions. Resource inadequacy and physical distance from sales and marketing reduced integration.

The distinction matters. Search finds more pieces. Integration explains how a piece changes what the manager already knows about the product, operation, customer, and constraint.

For the spreadsheet workaround, integration might connect four facts:

• managers need a printable audit trail;
• administrators cannot configure conditional approval rules;
• the product stores decisions but not the evidence behind them;
• regulated customers cannot approve a workflow whose exceptions are invisible.

Now “workflow friction” becomes a causal proposition: customers leave the product because the approval system cannot make exceptional decisions inspectable. That claim can be wrong. It can also guide a decision and a test. The original label could do neither.

Wesley Cohen and Daniel Levinthal call the broader capability absorptive capacity: the ability to recognize the value of external knowledge, assimilate it, and apply it. It depends on prior related knowledge and develops cumulatively. The same interview can therefore create a design change in one company and a forgotten highlight in another. The difference is not customer eloquence. It is what the organization can connect the evidence to.

The customer does not carry the insight across your org chart

The strongest evidence in this literature concerns the organization between customer contact and innovation.

Nicolai Foss, Keld Laursen, and Torben Pedersen studied 169 Danish firms drawn from a survey of the 1,000 largest firms. Their model did not find customer knowledge leaping directly into innovation. The relationship was completely mediated by new organizational practices: intensive vertical and lateral communication, rewards for sharing and acquiring knowledge, and delegated decision rights.

The reported total standardized effects summarized in the study make the asymmetry visible. Internal communication carried 0.49. Delegation carried 0.38. Knowledge incentives carried 0.18. Direct customer interaction carried 0.13.

These are model-specific coefficients from one national setting, not universal intervention prices. They should not be copied into a management scorecard. Their ordering supports a more durable conclusion: the internal organization can matter more than another unit of customer contact.

Organizational evidence

Customer contact moved through internal organization

In one model, communication and delegated decision rights carried larger total effects than customer interaction alone.

Total standardized effects

0.000.100.200.300.400.50

Internal communication0.49

Delegated decision rights0.38

Knowledge incentives0.18

Customer interaction0.13

Reported total standardized effects from one model of 169 Danish firms. They are not universal intervention estimates.

Delegation is particularly important because customer insight often arrives as a recommendation without a decision right. A research lead can say that the approval workflow fails. A product manager may own the roadmap but not compliance policy. Engineering can change the workflow but not the audit requirement. Sales can describe urgency but not allocate delivery capacity.

The insight has many interested parties and no complete owner.

An owner is complete only when the person can:

• select among the live alternatives;
• obtain the people needed to test one;
• accept a bounded cost or trade-off;
• explain a non-action;
• return after delivery with outcome evidence.

Anything less is sponsorship theater.

Buying the tool does not complete the system

This does not make research repositories or CRM systems useless. It makes them insufficient.

Hanna Salojärvi, Liisa-Maija Sainio, and Anssi Tarkiainen found that teams, top-management involvement, key-account formalization, and CRM technology enhanced customer-knowledge utilization in large industrial account relationships.

Simos Chari, Anssi Tarkiainen, and Salojärvi later reanalyzed the setting with fuzzy-set qualitative comparative analysis. Key-account teams and CRM systems appeared as core predictors of high utilization, but they were not sufficient by themselves. Different combinations of customer orientation, top-management involvement, formalization, teams, and technology produced viable paths.

That is a useful warning against the single-fix reflex.

• A repository without operating ownership becomes a museum.
• An insight team without authority becomes a newsletter.
• Executive attention without translation produces urgent ambiguity.
• Formal workflow without trust produces ceremonial compliance.
• Embedded researchers without protected synthesis can become request takers.

There is no universal organization chart for insight. There is a universal requirement for completeness: the chosen configuration must carry evidence all the way to a tested response.

More customer participation can expose a weaker relay

Customer participation is often treated as an uncomplicated good. Invite more customers into ideation, prototyping, testing, and development; improve the product.

The evidence is more conditional.

Todd Morgan, Michael Obal, and Sergey Anokhin examined 243 firms of different sizes across 14 industries. Customer participation related to new-product performance directly and through innovativeness, but the effects depended on absorptive capacity. Firms with stronger capacity gained more, especially later in development.

Customer participation therefore increases the value of an action system. It can also increase the load on a weak one. More participants create more needs, exceptions, proposed solutions, and conflicting time horizons. Without integration and decision rules, participation expands the queue.

Susan Albers Mohrman, David Finegold, and Allan Mohrman derived a knowledge-system model from 1,200 engineers in 10 technology firms and validated it on a hold-out sample. Core behaviors included linking knowledge sources with needs, accessing procedural knowledge systematically, examining the organization as a system, and experimenting with new approaches.

That is the work between “customers said” and “we changed.”

Evidence landscape

The action problem appears at several organizational levels

Independent studies locate failure in dissemination, individual integration, organizational practice, absorptive capacity, and knowledge work.

Cross-functional relay

nonmarketing managers

frequency, formality, trust, and distance

Information integration

manufacturing and R&D managers

search, integration, resources, and use

Organizational mediation

Danish firms analyzed

communication, incentives, and delegation

Participation contingency

firms across 14 industries

customer participation and absorptive capacity

Knowledge system

engineers in 10 firms

linking, procedural access, and experimentation

Five independent study scopes. Counts remain separate because their populations, constructs, and methods differ.

Give every material signal a disposition

Insight backlogs become graveyards because they have statuses such as new, reviewed, and shared. These describe information handling, not organizational judgment.

A material signal should leave review with one of four dispositions:

• Act: Evidence and authority are sufficient for a bounded response.
• Test: A live uncertainty separates plausible actions; run the smallest diagnostic test.
• Park: The signal is credible but not actionable now; record the condition and date that will reopen it.
• Reject: The evidence is weak, outside strategy, contradicted, or too costly; record the reason.

The park and reject states matter because acting on every customer request is not customer orientation. It is abdication. Customers observe their context. The firm must integrate those observations across segments, risks, alternatives, and capabilities.

The failure is not a reasoned non-action. The failure is silence that looks like prioritization.

Chris Argyris’s account of actionable knowledge provides a demanding test: useful propositions expose the causal processes required to implement an intention and can be tested in practice. “Customers need flexibility” is descriptive. “If administrators can expose exception evidence inside the approval record, regulated teams will complete more approvals without spreadsheet export” is actionable. It names a mechanism and predicts an outcome.

Measure closure, not volume

Common insight-program metrics reward accumulation:

• interviews completed;
• reports published;
• repository searches;
• dashboard views;
• stakeholders attending readouts.

These can diagnose reach and activity. They cannot show that the relay closed.

Add measures closer to action:

• median time from material signal to explicit disposition;
• percentage of signals with a decision owner and deadline;
• percentage of tests that resolve the named uncertainty;
• percentage of shipped responses with a measured customer outcome;
• repeated rediscovery rate for unresolved signals;
• number of decisions changed, narrowed, delayed, or defended with recorded evidence.

Do not convert these into a vanity composite. A fast rejection is not always good. A high action rate may mean weak selection. A low decision-change rate may reflect excellent prior judgment—or a protected roadmap. Review the measures together and inspect the cases behind them.

Return to the spreadsheet workaround. The signal was not “activated” when it entered the repository. It became actionable when the organization separated observation from explanation, named the approval-policy decision, assigned an owner who could coordinate product and compliance, shipped a bounded audit-trail test, and checked whether spreadsheet exports fell.

The customer supplied evidence. The company had to build the route.

References

Argyris, C. (2005). Actionable knowledge. In The Oxford Handbook of Organization Theory.

Chari, S., Tarkiainen, A., & Salojärvi, H. (2016). Alternative pathways to utilizing customer knowledge. Journal of Business Research, 69(11), 5494-5499.

Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35(1), 128-152.

Foss, N. J., Laursen, K., & Pedersen, T. (2011). Linking customer interaction and innovation. Organization Science, 22(4), 980-999.

Jaworski, B. J., & Kohli, A. K. (1993). Market orientation: Antecedents and consequences. Journal of Marketing, 57(3), 53-70.

Kohli, A. K., & Jaworski, B. J. (1990). Market orientation: The construct, research propositions, and managerial implications. Journal of Marketing, 54(2), 1-18.

Korhonen-Sande, S. (2010). Micro-foundations of market orientation. Industrial Marketing Management, 39(4), 661-671.

Maltz, E., & Kohli, A. K. (1996). Market intelligence dissemination across functional boundaries. Journal of Marketing Research, 33(1), 47-61.

Mohrman, S. A., Finegold, D., & Mohrman, A. M. (2003). An empirical model of the organization knowledge system in new product development firms. Journal of Engineering and Technology Management, 20(1-2), 7-38.

Moorman, C. (1995). Organizational market information processes. Journal of Marketing Research, 32(3), 318-335.

Morgan, T., Obal, M., & Anokhin, S. (2018). Customer participation and new product performance. Research Policy, 47(2), 498-510.

Salojärvi, H., Sainio, L.-M., & Tarkiainen, A. (2010). Organizational factors enhancing customer knowledge utilization. Industrial Marketing Management, 39(8), 1395-1402.