The Listening Gap in the Age of Smart Products

Smart products are buzzing like never before and they are transmitting signals, with the global datasphere estimated to be at 175 zettabytes (per IDC), a quarter being real time, with 51% being IoT data. IDC also estimates that by end of 2025, an average connected person could have approximately 4900 digital interactions per day – roughly one every 18 seconds.  In my previous blog post, I had talked at length about how AI readiness can become business readiness when we know how to listen when smart products start speaking.

However, proper “listening” of such signals, still lag.

In Wavestone’s (formerly NewVantage Partners) 2024 survey, only 48.1% say they have created an industry level leadership through a data driven organization that is capable of doing such listening, with the top barrier being culture/people/process at 78%.

In a McKinsey’s 2025 global survey, although 88% of organizations claim to use AI in at least one function, only one-third of them have begun scaling and only 39% of them report enterprise level EBIT impact because of this. This again points to insufficient “listening” readiness impacting “thinking” based actions downstream.

Such “listening gap” results in value leaks leading to higher cost to serve, slower product iteration and missed business expansion signals.

So, the question is, what does “listening” to smart product signals mean and where is this gap that many organizations still struggle with?

The Listening Gap

Most “smart product” programs, as we see, plateau at telemetry led dashboards. Engineering can see usage spikes, support can see tickets, sales can see churn risks and finance can see cost to serve. However, most organizations struggle to connect these fragments into one unified decision grade truth- then embed “thinking” based actions into daily workflows with clear owners.

The result is predictable. Reliability issue surface late, adoption friction gets discovered after missing renewals, Services never reach the desired state of being predictive and product roadmaps are still shaped by internal “engineering wishes” rather than getting guided by missed customer expectations.

True “listening” is about closing this loop. It is about building a repeatable discipline that converts multiple product signals into a comprehensive unified understanding which can then help influence higher enterprise functioning and hence, value.

The Listening Framework

We believe true listening must follow four key steps in the right sequence.

1. Intent driven listening – Telemetry becomes valuable only when you decide, upfront, what you’re listening for. We believe one can start by focusing on four target areas – Reliability (when will a product fail), Friction (where are users struggling), Adoption (who is drifting, usage wise ) and Outcomes (is usage meeting desired business outcomes). This should then be followed by defining 10-20 golden signals that can best help each target area with one explicit owner to monitor this.

Example: Rolls-Royce’s latest Engine Health Monitoring system measures thousands of operating parameters and can “talk back” – responding to focused requests from an operational center and sending specific information based on analysis of hundreds of hours of information. This is intent driven listening and not passive ingestion

2. Making signals trustworthy – If teams do not trust product signals, they would not act. To make them trusted, it is important we enable two things – Integrity (clean, consistent, contextual) and Trust Rails (security, privacy, access control, retention and user consent). Without this one can get into lot of resistances and debates and can even miss the precision element of actions based on these signals.

Example: Caterpillar’s secured remote asset monitoring system highlights encrypted and authenticated remote connections, and an architecture where only product initiated outbound remote connections are allowed – reducing exposure to general internet traffic. That is “trust by design” built into the listening layer.

3. Translating signals into decisions – Insights alone do not create value – workflow changes because of insights – do. The practical pattern is “if X, then Y” leading to decision playbooks across product, support, customer success and sales leading to precise interventions, fixes and offers (and not just reports). If the insights from last week does not change how we approach work next week – the value of signals are only academic.

Example: John Deer’s 2024 Business Impact Report models how it’s technology stack (including “Operations Center”) translates data into operational impact, illustrating a “model farm” scenario that includes significant total cost savings. It is an example of signals driven decision leading to tangible business outcomes.

4. Closing the loop – Listening results in value only when actions because of this listening creates better products and operations. This calls for a value system – assigning a decision owner, running an action playbook based on signals analysis, track leading indicators (adoption, ticket counts, sales opportunities) through to lagging outcomes (downtime, CSAT, new wins and engagement growth) and iterate. This unlocks a growth flywheel: better listening > focused actions > better outcomes > more interest in listening leading to stronger differentiation and competitive advantage getting created.

Example: PTC’s CIMC smart connected factory case study reports 16% WIP reduction, 30% reduction in planned downtime, 13.2% reduction in energy consumption and scaling towards 35 plants with estimated energy savings of $15M. That’s the loop made measurable.

Closing Thought

Smart products are generating signals with volume, velocity and variety that would have been unthinkable a few years ago. The paradox is many enterprises are still listening in fragments, collecting telemetry data, building dashboards, running pilots, yet struggling to convert product “speech” into repeatable, measurable value.

This frontier hence is about the shift from “hearing” to “listening”. That requires not just collecting more signals but by running a weekly listening discipline targeted at the signals already collected. This requires a listening framework which captures four aspects of listening:

• Listen with intent
• Make signals trustworthy
• Translate signals into precise workflow changes
• Close the loop with actions that show up in outcomes and the P&L.

But here’s a hard truth – listening at scale is impossible without trust at scale. As signals become richer; and as AI begins to interpret, recommend, and increasingly act; enterprises need governance foundations for privacy, security, lineage, access controls, model oversight, explainability standards, human-in-the-loop policies, and auditability.

Without these, the very systems meant to create competitive advantage and enterprise value can become a source of risk. That would be our next frontier topic.