Lost Telecom Treasure

How telcos can responsibly unlock millions of daily events—and finally lead the next era of AI, intelligence, and trusted digital services.

1. Introduction: A Billion-Dollar Treasure Hidden in Plain Sight

For two decades, the digital economy has rewarded one fundamental principle:
Companies that understand behaviour build better services.

Google grew by analysing what people search¹.
Meta mastered engagement through interaction signals².
TikTok predicts what users will watch through micro-behavioural modelling³.

Yet long before these companies existed, telecom operators were already the first organisations in history to observe digital behavioural patterns at scale:

• mobility

• session activity

• device context

• traffic categories

• consistent identity

But instead of turning these insights into value, most operators isolated them inside network systems—held back by regulatory uncertainty, fragmented OSS/BSS stacks, and organisational silos.

Today, operators generate tens of billions of multi-environment events per day describing behaviour, movement, interests, device context, and real-time activity.

This is not “legacy signalling.”

It is behavioural telemetry—rich, structured, and highly valuable if managed responsibly.

And the surprising truth?
The treasure was never lost or stolen. It simply remained unused.

This article explains why that happened, why the opportunity is now returning—and how real operator systems processing billions of daily events show that the next generation of AI-driven telco services is not just possible, but already underway.

2. The Lost Map: Why Telcos Slowed Down the Data Race

2.1 Telcos had the original behavioural dataset

Years before tracking pixels or programmatic advertising, telcos already held:

• precise mobility patterns

• device class and TAC

• app/domain category metadata

• traffic intensity

• roaming and travel indicators

• session continuity

• deterministic identity through SIM

• location

The same types of signals Big Tech monetises today.

2.2 Why telcos stepped back

This happened for understandable reasons:

Regulatory caution
GDPR, ePrivacy, and telecom secrecy obligations created a perception that “everything is forbidden,” even when properly anonymised intelligence remained fully legal⁴.

OSS/BSS fragmentation
Data lived across dozens of systems, each with different owners, formats, and latency constraints.

Vendor dependency
Operators expected large vendors to “solve” data monetisation, slowing internal innovation.

Organisational silos
Network, IT, marketing, analytics, fraud, and enterprise teams often lacked a unified intelligence strategy.

Reliability-first culture
Telcos were engineered for availability and regulatory compliance—not behavioural insight or AI use cases.

Meanwhile, Big Tech built AI engines, behavioural platforms, and monetisation ecosystems—powered by signals far less structured than what telcos already had.

Telcos didn’t “lose” the race.
They paused before even entering it.

3. The Treasure: Billions of High-Quality Daily Events

A typical mid-size operator produces:

• ~3 billion behavioural events/day

• \>1 trillion event records/year

• ~250 events per subscriber/day

These are real numbers taken from modern mobile networks, Network DATA (XDRs, HTTP metadata, RAN events, location updates, QoS/QoE classifiers, probe telemetry).

Each event may contain:

• HTTP category (e.g., sports, finance, travel, retail)

• app usage clusters

• mobility context

• device class and OS

• timestamps

• roaming indicators

• session patterns

• anonymised subscriber token

This is the same class of behavioural telemetry used by:

• YouTube recommendation engines

• TikTok For-You AI

• Amazon’s product prediction models

Except telcos have:

• more consistent timestamps

• deterministic identity

• complete app and domain coverage

• stable mobility graphs

The difference is simply this:
Operators never transformed these signals into market-ready intelligence.

4. Market Value: Based on NeoTela, Not Theory

All numbers below are grounded in NeoTela—a realistic simulation environment representing a mid-size European operator with actual event intensity, anonymisation models, category distribution, HTTP classification, subscriber density, and real CPM benchmarks.

CPM benchmark citations:

• EU retail media/mobility analytics CPM: €1.5–€3.5 (Vodafone Analytics, Orange Flux Vision reports)⁵⁶

• EU telco audience CPM for consented segments: €2–€8⁷

• Programmatic geo-behaviour CPM: €1.2–€4 (IAB Europe AdEx benchmarks)⁸

NeoTela conservative model

From 90M usable daily events (3% of total operator events):

If only 25% of 90M are monetizable:

• 22.5M events/day

• 8.2B events/year

Revenue ranges:

• €2.7–2.8M/year (low CPM, limited adoption)

• €5–8M/year (moderate conditions)

• €10–15M/year (premium use cases + DSP integration)

This assumes standard EU CPM/CPD rates, not inflated projections.

Investment reality:
Typical deployment costs (from real operator projects):

• ~€2M CAPEX (from scratch)

• ~€300k OPEX/year

The business case is healthy even in the most conservative scenario.

5. We Know This Opportunity Because We Built It

This narrative is not theoretical—it comes from real operator deployments.

Real anonymised example #1 (Central and Eastern Europe)

A national operator deployed an online event-correlation platform processing 11–13B events/day to support:

• emergency communications

• roaming regulation compliance

• cell-level movement analytics

• real-time advertisement

• regulator obligations

All with full pseudonymisation and rotating-key governance.

Real anonymised example #2 (Central and Eastern Europe)

A mid-size operator used mobility events to support:

• emergency communication

• regulator obligations

No personal identifiers were ever exposed—only pseudonymised behavioural aggregates.

5.1 Platforms processing 15B events/day

We have designed and operated systems that:

• process up to 15B events/day

• run in online mode

• apply real-time policy

• enforce privacy by design

• correlate network and behavioural signals

These systems run today inside regulated national infrastructures.

5.2 Public-safety and regulatory functions

Including:

• national public warning

• roaming regulation

• emergency messaging

5.3 Strong privacy guarantees

Every deployment follows:

• pseudonymisation

• encryption (AES-GCM, TLS 1.3)

• tokenisation

• key rotation

• audit readiness

• NIS2-aligned security standards

Not a single PII attribute leaves the protected domain.

5.4 Conclusion from field projects

The treasure is not hypothetical.
It is live.
It is proven.
It just needs to be activated for new service categories.

6. Why Telco Data is Better (Technically) Than Big Tech Data

6.1 Coverage

Telcos see all apps, all devices, all domains—consented and anonymised—rather than only what happens inside a single platform.

6.2 Identity stability

Cookies disappear.
Advertising IDs reset.
App logins break.

SIM identity is deterministic, universal, and stable.

6.3 Well-structured data

Big Tech collects behavioural noise.
Telcos collect high-quality network metadata.

6.4 AI readiness

AI models perform best on:

• dense datasets

• consistent timestamps

• deterministic identifiers

This is exactly what telcos produce.

7. From Lost Treasure to AI Mine

This opportunity is far larger than advertising.

7.1 AI Copilots for network operations

Telemetry + AI can:

• detect anomalies

• predict congestion

• Identify device failures

• guide engineers

• optimise experience

7.2 Fraud & identity intelligence

Behavioural telemetry enables:

• impossible travel detection

• SIM/device anomalies

• suspicious movement patterns

• device-trust scoring

Banks dream about the data a telco sees in five minutes.

7.3 Mobility & Retail Intelligence

Cities and partners need:

• footfall

• commuter flows

• tourist behaviour

• POI visits

Telcos already have this—responsibly anonymised.

7.4 Experience & churn AI

Behavioural degradation is the earliest churn indicator.

7.5 Privacy-first personalisation

Categories → offers, boosters, packages, upgrades
Only with consent + anonymisation.

8. Responsible Intelligence: Privacy First

The only acceptable approach is:

• anonymised

• pseudonymised

• consent-driven

• encrypted

• minimal

• governed

With:

• pseudonymised tokens

• aggregated signals

• category-based intelligence

• rotating keys

This is not surveillance.
This is responsible, privacy-preserving network intelligence.

9. Rediscovering the Map: A Practical Telco Roadmap

PHASE 1: Build the foundation

1. unify events (network + HTTP + categories + mobility)

2. apply pseudonymisation & encryption

3. enforce GDPR/NIS2 policy engine

4. build a behavioural taxonomy

5. expose safe APIs to B2B/B2G partners

PHASE 2: Activate intelligence

6. train AI models on event horizons

7. partner with DSPs, banks, tourism boards, retailers

8. monetise responsibly and transparently

This is not reinvention.
This is activation.

10. Conclusion: Will Telcos Finally Claim the Treasure?

Every operator generates billions of events describing real behaviour.

This is not just network exhaust—it is a refined, structured behavioural dataset.

We know this because we built systems that:

• process 15B events/day

• support national safety

• meet strict regulations

• protect privacy

• deliver real intelligence

The treasure was never lost.
It was only undiscovered.

Today, the map is clear.
The mine is open.

The question now is: which operator will be bold enough to lead the next era of telco-powered AI?

References / Citations

¹ Google 10-K reports, Search & Ads revenue breakdown (2022–2024)
² Meta Q4 2023 Earnings – behavioural modelling & engagement metrics
³ ByteDance/TikTok AI papers (RecSys, SIGIR, WWW 2021–2023)
⁴ EDPB & GDPR Recital 26: anonymisation & pseudonymisation guidelines
⁵ Vodafone Analytics (Global Data Insights CPM benchmarks, 2021–2023)
⁶ Orange Flux Vision (Mobility data pricing reports, 2020–2023)
⁷ European Telco Audience CPM ranges – MobileSquared & GSMA Intelligence
⁸ IAB Europe AdEx Benchmark 2023 – programmatic & geo-audience CPM