Telcos have the data. They cannot use it.
Telecommunications companies sit on some of the richest behavioural data in any industry. They know when customers use their phones. They know which apps are active. They know how much data is consumed and when. They see patterns of communication, movement, and digital activity that no other company has access to.
And they cannot use any of it. Not in the way that matters. The data sits in silos, locked behind consent walls, governed by regulations that make it toxic to touch. Telcos have spent billions on customer data platforms, analytics stacks, and personalisation engines that run on a fraction of the data available. The result is personalisation that is neither personal nor effective.
The consent wall problem
GDPR and ePrivacy rules require explicit consent for processing personal data for marketing purposes. Telcos have learned the hard way that consent rates are low and declining. In many European markets, fewer than 30 percent of customers opt in to marketing data processing. Some telcos report opt-in rates below 15 percent.
This means the personalisation engine is running on a minority of the customer base. The majority of customers receive generic communications. The telco is blind to their behaviour, their preferences, and their intent. The customers who opt out are not a random sample. They tend to be more privacy-conscious, more digitally sophisticated, and often higher-value. The personalisation system is optimising for the wrong population.
The silo problem
Even among opted-in customers, the data is fragmented. Network data lives in one system. CRM data lives in another. Billing data, app usage data, and customer service interactions are stored separately, often in legacy systems that predate the current analytics stack. Joining these datasets requires complex ETL pipelines that introduce latency, data quality issues, and additional compliance obligations.
By the time the data is unified and ready for use, it is stale. The customer’s behaviour has moved on. The intent signal that was live yesterday is history today. The personalisation engine is always looking at the past, never at the present.
The CDP disappointment
Customer Data Platforms were supposed to solve the silo problem. Unify the data. Build a single customer view. Enable real-time personalisation. The promise was compelling. The reality has been different.
CDPs work by centralising data. They pull information from multiple sources into a unified profile. But in a telco context, much of the most valuable data cannot be centralised without consent. Network behaviour data, app usage data, and location data are particularly sensitive. The CDP ends up with a partial view, built from the data that is easy to move rather than the data that is most useful.
The result is personalisation that feels mechanical rather than intelligent. Segment-based offers that are loosely relevant at best. Campaigns triggered by lifecycle events that happened weeks ago. Churn models that detect defection after the customer has already decided to leave.
Privacy as the architecture, not the obstacle
Intent inverts the telco personalisation model. Instead of moving data from the device to a central platform, the intelligence is produced on the device itself. On-device AI processes the full range of behavioural signals locally. App usage. Content engagement. Timing patterns. Cross-category behaviour. All of it processed on the handset, in real time, without any data leaving the device.
The output is a privacy twin: a mathematical representation of behavioural intent that contains no personally identifiable information. The telco receives the intent signal, not the underlying data. It knows what a customer is likely to want. It does not know or need to know the specific behaviour that produced the signal.
This changes the consent equation entirely. Because no personal data is transmitted, the regulatory requirements that block traditional personalisation do not apply. The telco can generate intent signals for 100 percent of its base, not the 15 to 30 percent who opted in. The personalisation engine runs on the full population, with full signal richness, at zero regulatory risk.
What this looks like in practice
A customer’s on-device behaviour indicates they are researching international travel. The privacy twin transmits an intent signal: travel readiness. The telco surfaces a roaming package offer at the right moment, before the customer calls to ask about rates. No PII was transmitted. No consent dialog was required. The customer received a relevant offer. The telco captured revenue it would have missed.
Another customer’s behaviour suggests declining engagement with their current plan. The intent signal indicates potential churn. The telco can intervene with a retention offer before the customer starts comparing competitors. Again, no personal data moved. The intelligence was produced on-device. The telco acted on a signal, not on a surveillance profile.
The competitive reset
Telco personalisation has been broken for years. The industry knows it. The investment in CDPs, data lakes, and analytics platforms has produced disappointing returns because the architecture was wrong. The model of centralising sensitive data, obtaining consent to use it, and then processing it centrally is structurally flawed in a high-regulation environment.
The fix is not better data pipelines or more sophisticated consent management. The fix is a different architecture. Process on-device. Never transmit PII. Produce intelligence, not data. Privacy is not the obstacle to telco personalisation. It is the mechanism that makes it work. The telcos that understand this first will set the standard. The rest will spend years wondering why their CDPs never delivered what was promised.