Orange seeks hyper-personalization in adtech through data

Table of Contents

Orange’s strategy for 2028

Orange’s new roadmap toward 2028, presented under the slogan “Trust the future”, does not put the word “AI” as the main tagline, but the technology appears on almost every front. According to what the company communicated as part of its new strategy, the operator is proposing everything from “trusted” cloud services and AI security to a target of €600 million in opex savings supported by automation and advanced analytics.

In parallel, the strategy seeks to directly impact the commercial business: less churn, more cross-selling, and a more “contextual” relationship with the customer, based on usage data and decision models.

Pillar in “Trust the future” What it means in practice Goal/result associated with 2028 (as communicated)
Cross-cutting AI (not as a slogan, but as a layer) Models embedded in commercial and operational processes (decision-making, automation, prioritization) “100% of transactions” with AI assistance/augmentation (stated goal)
“Trusted” cloud and AI services Cloud/AI offering focused on trust and control Foundation to scale use cases and services (including the ambition to replicate Bleu)
Operational efficiency Automation + advanced analytics applied to operations €600M in opex savings (stated target)
Data monetization (personalization and adtech) Activation of signals for upsell and audience segmentation New revenue lines in mature markets and reduced churn

The importance of hyper-personalization in adtech

Orange is transferring the logic of hyper-personalization—traditionally associated with offers and customer service—to the advertising arena. In adtech, a telco’s differentiating value is its aggregated knowledge of behavior (network usage, service consumption habits, area-level location, among others), which can be turned into more precise segmentation for campaigns.

In the approach described by Orange, the focus is on activating signals to build audiences and moments of contact, rather than directly exposing personal data in campaigns.

The bet responds to a

clear trend: in a market where attention is scarce and competition for advertising performance is intense, segmentation based on first-party data is gaining weight over less deterministic approaches.

From signals to effective measurement
Signals → Audiences → Activation → Measurement (how hyper-personalization is “translated” into adtech)
1) Signals (inputs): usage patterns and context (e.g., data consumption, type of product contracted, area-level zone), plus commercial signals (e.g., device financing).
2) Audiences (building): rules/models that group users into segments useful for a campaign objective (reach, affinity, intent).
3) Activation (execution): delivery of campaigns to defined “subsets,” with frequency control and cross-channel consistency.
4) Measurement (learning): performance evaluation and adjustment of segments/creatives; if there is no consistent measurement, segmentation remains limited to one-off tests.
Key point: the value is not only in “having data,” but in turning it into actionable segments with governance and consent control.

Use of AI to improve the customer experience

The hyper-personalization Orange describes rests on a premise: the data have existed for a long time, but without AI it is not feasible to exploit them at scale given the volume and speed needed to turn them into useful commercial actions.

In a roundtable with the press and analysts, CFO Laurent Martinez explained that the goal is to use usage patterns to propose offers at the right moment. The approach is aimed at increasing revenue through upsell, rather than recommending the cheapest plan.

AI in CX: From Signals to Action
How AI is usually implemented in CX (with operational checkpoints)
1) Signal capture: usage events, browsing/app, purchases, incidents and context (channel, time, product).
– Checkpoint: quality and latency (if they arrive late, “personalization” becomes irrelevant).
2) Model/segmentation: propensity, churn, affinity, or hybrid rules.
– Checkpoint: bias and coverage (does it work the same in prepaid/postpaid, regions, segments?).
3) Decision: “next best action/offer” and prioritization (what to offer, when, and through which channel).
– Checkpoint: limits on commercial pressure (frequency, exclusions, cross-channel consistency).
4) Action: execution across channels (app, SMS, call center, store, web) with traceability.
– Checkpoint: omnichannel consistency (avoid the customer receiving contradictory messages).
5) Feedback: measurement and learning (acceptance, conversion, complaints, churn, NPS/CSAT if applicable).
– Checkpoint: closed loop (if it isn’t fed back, the model doesn’t improve and costs rise).

AI-augmented transactions

CEO Christel Heydemann set an ambitious goal: that 100% of transactions be “augmented” by AI. In practice, this means incorporating models that assist decisions and workflows at key points in the customer lifecycle: purchase, renewal, device financing, support, and retention.

The logic is twofold: on the one hand, improve conversion and satisfaction with more relevant interactions; on the other, reduce operating costs through automation and intelligent prioritization.

Control Points by Channel
“100% of transactions augmented”: checklist of control points by channel/moment
– Purchase/sign-up (web/app/store):
– [ ] Context-based plan/device recommendation (without breaking commercial rules).
– [ ] Minimum internal explainability: reason for recommendation logged.
– Renewal/plan change:
– [ ] Intent detection (search signals, usage, complaints) and timely offer.
– [ ] Eligibility control (avoid offering something not available to that customer).
– Device financing:
– [ ] Assisted assessment (risk/affinity) + human review when applicable.
– [ ] Clear and consistent messaging across channels.
– Support (contact center/chat/app):
– [ ] Intelligent classification and routing (priority, urgency, customer value).
– [ ] Automatic case summary to reduce repetition.
– Retention/churn:
– [ ] Early alerts and graduated actions (not just discounts).
– [ ] Impact measurement (churn avoided vs. cost of incentives).
If any of these points isn’t instrumented, “AI-augmented” usually ends up as partial automations or pilots.

Ad segmentation in France

In France, Orange already has an ad segmentation program up and running based on first-party data, and the company is expanding its reach. The goal is to offer brands and advertisers better-defined audiences for specific campaigns, leveraging the operator’s knowledge of its customer base.

Although Orange does not publicly detail performance metrics on this point, the move fits with the search for new revenue lines in a mature telco market, where organic growth of the core (connectivity) is limited.

Element Status (as described) What it typically enables
First-party data segmentation in France Active (“up and running”) More clearly defined audiences for specific campaigns
Program expansion Expanding More use cases/segments and broader commercial coverage
Public performance metrics Not publicly detailed Forces impact to be assessed case by case (e.g., lift, conversion, reach)

Expansion of adtech services in Spain

The next market highlighted is Spain, where Orange says these services have been available for a few months and should accelerate in the short term. Martinez underscored the scale: 33 million customers in the country, a base that makes it possible to build segments for marketing campaigns targeted at specific subsets of the population.

The proposition is to “sell” segmented audiences for campaigns, leveraging combined internal signals to improve targeting accuracy.

From pilot to commercial scaling
Typical telco adtech rollout (from “a few months” to ramp-up)
1) Controlled pilot: few advertisers/use cases, simple segments, basic measurement.
– Checkpoint: validate that consent and exclusions are applied correctly.
2) Ramp-up: more segments, more inventory/channels, automation of audience creation.
– Checkpoint: avoid degradation due to “over-segmentation” (segments that are too small or unstable).
3) Commercial scaling: audience catalog, operational SLAs, recurring reporting.
– Checkpoint: governance (who approves segments, how usage is audited, how incidents are handled).

Patterns of customer data usage

Orange illustrated hyper-personalization with a case in Africa: many customers prepay for data once or twice a week. If the operator detects that certain users run out of data early on Fridays—a day associated with salary payments—it can trigger specific offers at that time.

In Spain, Orange cited another type of signal: when it sells financed devices, it obtains information such as the customer’s salary, which it can combine with other data, for example area of residence, to build segments useful for campaigns.

The technological foundation of this capability, according to the company, is MasStack, an IT stack developed in-house and based on open-source software.

Example Available signal Trigger (moment) Described action
Africa (prepaid data) Top-up frequency and data depletion pattern Users who run out of data early on Friday Activate specific offers at that moment
Spain (financed device) Commercial information (e.g., salary) + area of residence Building segments for campaigns Sell segmented audiences to advertisers

Compliance with data regulation

Asked about regulatory compliance, Martinez indicated that Orange ensures it has formal consent through customer contracts. The company presents this point as a pillar for scaling its hyper-personalization and adtech initiatives without eroding trust, a central asset in its strategic narrative.

Personalization and trust at scale
Personalization vs. trust: trade-offs that often appear when scaling telco adtech
– More personalization can improve relevance and conversion, but:
– It reduces the margin for error: if the customer perceives “too much knowledge,” reputational risk increases.
– Stricter consent and more visible controls tend to increase trust, but:
– They can reduce reach (fewer eligible users) and slow commercial ramp-up.
– More granular segments can boost performance, but:
– They increase operational complexity (auditing, exclusions, omnichannel consistency) and the risk of unstable segments.
The practical key is that consent and exclusions are not a “legal step,” but a technically verifiable condition in every activation.

Digital Transformation in Telecommunications: A Future Driven by AI

The Role of AI in Service Personalization

Orange’s strategy illustrates how AI is moving from being an isolated project to becoming an operational layer: it decides the “next best step,” adjusts offers, and prioritizes interactions. Personalization stops being static segmentation and becomes continuous orchestration.

Challenges and Opportunities in the Telecommunications Industry

The opportunity is clear: monetize data and improve efficiency. The challenge is also: balance commercial growth with expectations of

clients and regulators, and demonstrate that the added value outweighs the sensitivity of the data.

Strategies to Improve the Customer Experience

“Augmented transactions” suggests a model where AI accompanies every touchpoint: recommendations, support, retention, and financing. The key will be ensuring that automation does not degrade the experience, but rather makes it faster and more relevant.

The Importance of Real-Time Data Integration

Hyper-personalization depends less on having “more data” and more on integrating and activating it in time. Without robust pipelines and governance, the promise remains limited to one-off campaigns rather than an industrial capability.

Future Outlook: Toward an Omnichannel Ecosystem

Orange also hinted at ambitions to replicate services beyond France —such as its sovereign cloud offering Bleu—, although without fixed definitions. The strategic debate, according to the company itself, is how to scale at the European level: more markets, more use cases, and consistent execution across all channels.

Cross-Cutting AI in Telco: Priorities
In telco, “cross-cutting AI” usually focuses on three connected fronts:
– Growth: personalization (upsell/retention) and new lines such as adtech.
– Efficiency: automation of operations and support to capture opex savings.
– Trust: “trusted” cloud/AI, security, and data governance to sustain adoption.
The difficulty is usually not the model itself, but industrializing data, decisions, and measurement without breaking the omnichannel experience.

Transform the customer experience in telecommunications with Suricata Cx

The comprehensive solution for ISPs and telecommunications operators

Suricata Cx proposes a unified layer to manage service, sales, and operations with a centralized view of the customer, designed for telco and ISP environments.

Tangible benefits of automation in customer service

Automation applied to frequent inquiries, case classification, and routing can reduce response times and free up team capacity for complex incidents.

Sales strategies optimized through artificial intelligence

Propensity models and “next best offer” make it possible to prioritize opportunities and personalize proposals, especially when they are integrated

usage and behavior signals.

Payment recovery and efficient collections management

Analytics can improve collections management with risk segmentation, smart reminders, and flows tailored to the customer profile.

Omnichannel operations: the key to seamless customer service

Consistency across channels (store, call center, app, web, and messaging) is critical to sustaining personalization: the customer expects continuity, not to repeat their story in every interaction.

That same challenge of activating signals in real time without eroding trust requires an omnichannel operation where AI enhances each interaction with control and traceability. From Suricata Cx’s perspective, the value lies in turning data and consent into useful decisions—faster and more relevant—that reduce friction and churn without losing the customer context.

Expected outcomes of CX/AI in telco
What “expected” outcomes are sought when implementing a CX/AI layer in telco (and what conditions them)
– Less friction in support: higher first-contact resolution and shorter handling time when there is classification/routing and case summarization.
– Better commercial conversion: increased offer acceptance when the “next best offer” is triggered at the right time and in the right channel.
– Lower churn: fewer cancellations when early alerts are connected to graduated actions (not just discounts) and impact is measured.
– More consistent operations: fewer contradictory messages across channels when there is a single customer view and eligibility rules.
Conditions that usually determine the outcome: data quality/latency, omnichannel consistency, control of commercial frequency, and a measurement loop that feeds back into models and rules.

This analysis focuses on how a telco approaches monetizing and activating data in personalization and adtech; the reading is done from an operational omnichannel CX approach for ISPs and operators, where integration, traceability, and human control usually determine whether personalization scales without degrading the experience.

This text draws on publicly available information and on statements attributed to Orange executives current at the time of writing. Some initiatives, especially in adtech, may evolve quickly and vary in scope, metrics, or availability by country. Where no public metrics exist, the operation and typical commitments are described without assuming results; updates may occur if new data emerges.