Language
Data Voices Manifesto
Data Voices Manifesto

The future of data
and AI 2030

#6 strategic predictions for an AI‑first organization
What Data & AI leaders really think
Preparing for 2030 starts today.

Data and artificial intelligence are already profoundly transforming organizations, how they work and how they make decisions. But what will the data and AI-driven company of 2030 look like? And what foundations must we build now to ensure its successful creation?

This Manifesto gives a voice to those who are at the heart of this transformation: the Data Voices, our community of leaders and experts who are changing how data and AI are used every day in complex, international and rapidly changing organizations.

6strategic predictions for 2030
~100Data & AI leaders surveyed, across all industries
5in-depth qualitative interviews
6.8/10average confidence level that predictions will be achieved
A word from our Data Voices Ambassador

We are at a turning point. As with the internet or social networks, AI can become either the best or the worst amplifier of what we do: using it poorly will harm our collective intelligence, while informed use will strengthen it. Beyond leading day-to-day projects, the role of the Chief Data Officer (CDO) and Chief Technology Officer (CTO) is evolving, becoming a guardian of common sense and realism amid the current flood of technology and AI marketing. This digital transformation has strong similarities with the energy transition: it is a collective responsibility for us all that must help shape society. Just as we choose our energy future, we must choose a digital path that serves humanity.

Michel Lutz Chief Data Officer & Digital Factory Head of Data & AI · TotalEnergies Data Voices Community Ambassador
Methodology

The Data Voices Manifesto is based on a mixed, quantitative and qualitative survey process conducted by Huwise among a hundred Data & AI leaders from major international organizations, all sectors combined. A bilingual questionnaire (French and English) was used to collect their perception of six structuring predictions on the future of data and AI by 2030. These findings were complemented by five in-depth qualitative interviews with community members chosen to cover a wide range of organizational contexts and industry sectors. This made it possible to consolidate, nuance and illustrate the quantitative results to highlight six strategic predictions around organizational evolution towards an AI-first model.

The results of the study

6 predictions for an AI-first organization

This initiative was designed as a collective effort, bringing together leaders who believe that data and AI must become strategic drivers that support the business and drive performance and innovation.

Click on a prediction to find out more: 2030 vision, confidence level, identified benefits, challenges, actions to be taken, and the thoughts of a Data Voice.Tap a prediction to learn more: 2030 vision, confidence level, benefits, challenges, actions and the thoughts of a Data Voice.

Confidence level
in achieving
this prediction:
7.2/10

By 2030, the data and AI leader will no longer just be a technical expert; they will be the CEO’s partner in setting and implementing company strategy, because data lies at the heart of digital transformation.

Expected benefits
Align every data & AI initiative with strategic goals81%
Ensure AI transparency, security, and accountability65%
Orchestrate collaboration between human intelligence and AI agents62%
See all (6) See less
Use cases that generate concrete ROI54%
Use predictive models to anticipate risks and identify areas for growth38%
Increase the CDO’s credibility with the board31%
Anticipated challenges
Lack of data literacy within boards73%
Unresolved issues around data quality, reliability, and access65%
Difficulty in prioritizing strategy over short-term technical needs54%
See all (6) See less
Lack of CDO communication and change management skills46%
CDOs associated with governance and infrastructure, not perceived as a business partner35%
Difficulty proving the business impact of data architecture31%
Priority actions
Align data & AI strategy with business objectives73%
Create cross-functional business/data teams62%
Ensure CDO presence in relevant senior management discussions54%
See all (8) See less
Implement KPIs and business value indicators46%
Adopt an approach based on rapid POCs, iterations, and scaling of use cases that create value27%
Translate technical metrics to align with business challenges27%
Adopt a test & learn approach23%
Strengthen ROI measurement tools23%
Data Voices' views on this prediction

Reconciling governance rigor with operational agility and effectiveness

The data leader is no longer simply a technical expert who has been promoted. They are the backbone of a multidimensional transformation. They must reconcile governance rigor with the agility of operational effectiveness.

Samia BoujatiouiGroup Head of Data Management · Groupe Coface
1 / 6
Data Voices' views on this prediction

A guardian of common sense amid the flood of technology marketing

Beyond their traditional expertise and responsibilities, the role of the CDO or CTO today is also to act as a guardian of common sense and realism amid the current flood of technology and AI marketing. We must invest in cultural change efforts now, because the digital divide will not only be about access to tools, but about the ability to understand and control them.

Michel LutzChief Data Officer & Digital Factory Head of Data & AI · TotalEnergies
2 / 6
Data Voices' views on this prediction

Transcending technical complexity to speak in business terms

What I see with some CDOs is a lack of strategic vision combined with limited access to executive leadership. The data leader’s role is to transcend technical or administrative complexity and speak in business terms. If the executive understands the impact, they will naturally ask: “What do you need to succeed?”

Aldrick ZappelliniChief Data & AI Officer · Crédit Agricole Group
3 / 6
Data Voices' views on this prediction

Translating complexity into a coherent vision

The data leader must adapt technical concepts so that they engage diverse audiences, speaking equally to FinOps experts, while talking to frontline staff and senior leaders. It is this ability to translate complexity into a coherent vision that underpins the legitimacy of their role.

Camille MaireHead of Data Governance / MDM / Data Quality · Groupe ETAM
4 / 6
Data Voices' views on this prediction

The role of the data leader goes beyond simple delivery

The role of the data leader goes beyond simple delivery and includes three other dimensions: strategic vision aligned with business priorities, organizational transformation (operating models and ways of working), and cultural change to support adoption while ensuring strong foundations.

Chafika ChettaouiChief Data & AI Transformation Officer · AXA France
5 / 6
Data Voices' views on this prediction

Transforming data from a technical asset into a strategic business enabler

The role of the CDO is to transform data from a technical asset into a strategic business enabler: industrialized, adopted by business teams, and directly contributing to business performance, at a time when the rapid acceleration of AI is increasing reliance on high-quality data.

Lynda DehmaneChief Data Analytics & Cognitive Solutions Officer · Generali France
6 / 6
Confidence level
in achieving
this prediction:
7.3/10

By 2030, data-driven culture will be a reality for all. Every employee, regardless of their role, will be able to use data fully independently, supported when required by dedicated intelligent agents. Data will no longer be the domain of a few—it will become the very DNA of organizations.

Expected benefits
Faster, more informed decision-making69%
Increased productivity through AI agent automation69%
Enhanced collaboration between business teams and data experts58%
See all (7) See less
Generate more innovation and business opportunities50%
Better adoption of data and AI tools38%
Better risk management through the systematic use of data35%
Reduce IT workload around providing simple reports31%
Anticipated challenges
Poor availability and lack of trust in data69%
Difficulty in engaging all profiles, especially those least familiar with AI69%
Limited support from senior management around scaling data culture58%
See all (4) See less
Lack of investment in continuing training42%
Priority actions
Deploy data marketplaces inspired by e-commerce principles65%
Launch continuing training programs65%
Raise awareness of the proper understanding and use of data58%
See all (5) See less
Appoint data champions in each department58%
Drive usage by prioritizing the right use cases, closely connected to day-to-day work58%

The goal is to leave no one behind

When it comes to culture, the objective is not to impose rigid standards that would slow down our most advanced teams, many of which have invested heavily in data for years. The real challenge is to define a “minimum viable” cultural foundation. We need to raise the average maturity level so the Group can operate more homogeneously, while still allowing the most advanced parts to continue accelerating and sharing their experience to act as engines for those with fewer resources. The goal is to leave no one behind.

Read the rest of the quote See less

This positioning radically transforms change management. Data ceases to be a local constraint to become a lever for global coherence, facilitating synergies without stifling business specificities.

Crédit Agricole has chosen an approach based on practical implementation via a Group data marketplace to industrialize data sharing: “think data first” rather than theory.

Aldrick ZappelliniChief Data & AI Officer · Crédit Agricole Group
Confidence level
in achieving
this prediction:
6.7/10

By 2030, the organization’s entire portfolio of strategic data assets will be managed, approved, and shared as autonomous, reusable data products, available through self-service access for both employees and AI agents.

Expected benefits
Facilitate access to data by business teams to create value81%
Make data usable by AI62%
Reduce time spent searching for data through self-service62%
See all (6) See less
Empower everyone to make more informed decisions58%
Reduce dependence on IT to prepare and leverage data42%
Shorten time-to-market for data and AI use cases42%
Anticipated challenges
Data is still trapped in silos73%
Lack of clear human ownership in the data product value chain69%
Business teams that struggle to adopt a true product mindset54%
See all (8) See less
Overly-fragmented data architectures46%
Lack of resources and skills to transform data into products35%
Value of data products perceived as abstract, with difficult to demonstrate ROI31%
Internal tools that can only be used by experts, slowing agility23%
Overly rigid data access and security rules19%
Priority actions
Adopt agile data product governance85%
Deploy a data marketplace for self-service access69%
Designate business employees as owners for data products69%
See all (5) See less
Create stronger employee awareness of the value of data50%
Use automated workflows to keep data products up to date38%

A shift from quantity to relevance

The data marketplace operates like a supermarket for data assets: reports, Denodo views, and metadata. Today, our ambition is to provide users with smart data: users know who the Data Owner is, they understand the validated definition of each indicator, and they benefit from built-in, region-based access security. This represents a shift from quantity to relevance.

Read the rest of the quote See less

We never start from scratch. The goal is not to impose a rigid theory, but to adapt the maturity of the data product to the functional reality of each business domain, taking a “reasonable” approach to data mesh.

A data product is defined at Coface by having an identified owner, a service-level agreement (SLA), and full traceability. The semantic layer and our Virtual Data Lake are the absolute prerequisites for agentic AI: without them, it is nothing more than a promise that risks multiplying unexpected issues or hallucinations.

Samia BoujatiouiGroup Head of Data Management · Groupe Coface
Confidence level
in achieving
this prediction:
6.3/10

By 2030, AI will power the entire data value chain within organizations—from collection to value creation—including quality, observability, governance, traceability, security, and cataloguing.

Expected benefits
Massive improvement in data quality through anomaly detection77%
AI in engineering seen as a driver of innovation73%
Accelerate time-to-market for data products54%
See all (7) See less
Enhance data observability with pre-incident alerts50%
Greater scalability of data usage38%
Drastic operational cost reductions through automating repetitive tasks23%
Improve data lineage23%
Anticipated challenges
Lack of trust in autonomous systems65%
Difficulty measuring the ROI of AI-driven data engineering58%
Difficulty in establishing automation governance42%
See all (6) See less
High initial investment to automate engineering38%
Lack of resources to transition engineers to AI monitoring roles38%
Incompatibility of existing systems with autonomous AI agents27%
Priority actions
Upskill and change the culture of data teams65%
Invest in AI agent observability and monitoring62%
Implement governance dedicated to automation58%
See all (7) See less
Adopt flexible, API-driven infrastructure technologies35%
Identify the tasks and processes to automate31%
Enable systems to autonomously adjust resource consumption19%
Acquire or develop in-house automation tools19%

Human expertise is not disappearing

We do not simply install a tool; we design complex systems containing skills, harnesses, and integrations to supervise AI. Human expertise is not disappearing — quite the opposite: it is shifting and delivering differently. Only if the system is properly supervised by knowledgeable humans will it behave as expected.

Read the rest of the quote See less

The major risk is of cognitive dependency. By delegating too much thinking to models controlled by a limited number of actors, we risk weakening our intellectual sovereignty.

Competitive advantage does not lie in the tool itself, but in the ability to orchestrate four pillars of knowledge: scientific knowledge, internal data, compressed LLM knowledge, and tacit knowledge from humans. Sustainable performance depends on human-machine collaboration, never on technology in isolation.

Michel LutzChief Data Officer & Digital Factory Head of Data & AI · TotalEnergies
Confidence level
in achieving
this prediction:
6.9/10

By 2030, data products will be reliable and high-quality, enabling all forms of data use, including the widespread adoption of agentic AI: integrated context, embedded semantic layers, and full interoperability.

Expected benefits
Greater AI reliability and reduced bias through certified data85%
More accurate and higher-performing AI models54%
Time savings via end-to-end autonomous agents50%
See all (8) See less
Improve decision-making agility by moving from raw data to strategic action42%
Access to information via natural language interactions31%
Reduce development time for AI use cases27%
Cost reduction and ROI improvement23%
Increase interoperability through machine-readable data products23%
Anticipated challenges
Data missing semantic and context layers65%
Insufficient data quality58%
Lack of shared standards and norms such as data contracts54%
See all (7) See less
Poorly defined or poorly structured metadata46%
Organizational and technological silos42%
Absence of dedicated budgets31%
Proliferation of Shadow IT (including Shadow Data & Shadow AI)19%
Priority actions
Systematically add contextual metadata to every data asset69%
Solve quality issues before developing AI models58%
Prioritize the data that feeds high-value AI uses50%
See all (9) See less
Audit data that can be used autonomously35%
Develop a universal human/AI glossary31%
Train teams on the challenges of generative AI and data structuring31%
Raise senior management awareness and increase budgets31%
Reorganize teams and roles around data product management27%
Create an internal automation roadmap12%

No outputs can be used without a control indicator

Companies may be tempted to prioritize rapid delivery and shiny innovations over accuracy and security. My team’s role is to remind everyone that if we feed garbage into a model without proper control indicators, we will not get anything truly usable as output.

Read the rest of the quote See less

The solution requires three essential pillars: human accountability from business teams, rigorous operational processes and a high-performance MDM infrastructure.

Today, employees only perceive quality at the end of the chain, through the distorting prism of reporting. Continuous education is required to explain this. It is more a matter of culture change and documentation than pure IT systems.

Camille MaireHead of Data Governance / MDM / Data Quality · Groupe ETAM
Confidence level
in achieving
this prediction:
6.3/10

By 2030, data and AI governance will no longer be a rigid framework imposed on the organization, but a living, adaptive system—more federated than centralized—embedded at the heart of each data product, and naturally integrating requirements for trust, sustainability, and efficiency.

Expected benefits
Precisely track the usage and impact of each data asset73%
Guarantee real-time data quality and security65%
Detect and correct bias and drift at source58%
See all (6) See less
Automatic compliance with evolving regulations42%
Ensure compliance and security through certification38%
Eliminate bottlenecks related to manual approvals38%
Anticipated challenges
Cultural resistance: moving from centralized control to distributed autonomy65%
Difficulty translating unclear ethical/legal principles into automatable rules65%
Inadequate tools and systems for agile governance58%
See all (5) See less
High levels of risk46%
Aligning standards between cloud, on-premise and SaaS35%
Priority actions
Integrate safety, compliance and sustainability by design62%
Adopt a data mesh approach by empowering business teams54%
Replace periodic audits with continuous monitoring50%
See all (8) See less
Embed governance into technical processes and pipelines42%
Adopt Policy-as-Code tools38%
Train business teams in data ethics38%
Invest in decentralized governance tools such as data marketplaces27%
Automate data lineage for full traceability15%

Putting responsibility back where value is created

If the term “governance” still generates resistance, it is because it remains associated with a bureaucratic mindset that discourages operational teams. Instead, focusing on Business Data Ownership changes the perspective by putting responsibility back where value is actually created.

Read the rest of the quote See less

At AXA France, the “governed decentralization” model is built on three complementary pillars: a Data & AI Office responsible for AI strategy and governance frameworks, a technological pillar based on DataOps, and a culture where responsibility for data is shared by all.

There is no long-term AI without responsible AI. Ethics can no longer be treated as a checkbox exercise. It is central to our business, and it requires us to ensure that AI is designed and used neither incorrectly nor too late.

Chafika ChettaouiChief Data & AI Transformation Officer · AXA France
Roadmap

Data Voices' priorities in a nutshell

When invited to rank the 6 predictions in order of importance, the Data Voices outline a two-step process: first to build the human and data foundations, then to push on with the most ambitious transformations.

Phase 1Laying the foundations
01
Spreading a data culture among all employeesThe top priority: transforming mindsets before tools, processes, or architectures.
02
Make data AI-ready, reliable and high-qualityWithout quality, there is no reliable AI. Without data prepared like products, AI agents cannot act effectively.
03
Evolving the role of the CDO into a business strategistOnce culture has been created and data quality established, the next step is to evolve the CDO role.
Phase 2Deliver ambitious transformations
04
Install agile and ethical governance by designGovernance does not precede transformation, it happens side by side with it.
05
Generalize the data product modelRequires an already mature organization to see the process through.
06
Automate data engineering with AIThe most difficult prediction to achieve, as it relies on the other five having already been completed.
What the Data Voices tell us
Confidence levels

Where are data leaders most confident?

Average confidence level in achieving each prediction by 2030, as reported by the Data Voices surveyed. Ranking from highest to lowest confidence (score out of 10).

Prediction #2We will enable every employee to effortlessly leverage data to inform their decision-making and meet all their business needs
7.3
Prediction #1Data leaders will be recognized as experts and strategists creating value and building trust in AI
7.2
Prediction #5We will make data quality excellent as standard, and a driver for AI
6.9
Prediction #3We will turn every piece of data into a product accessible to every employee
6.7
Prediction #4We will entrust as much of data engineering as possible to AI
6.3
Prediction #6We will adopt agile and responsible data governance that becomes a competitive advantage
6.3
Overall confidence levelAverage over all 6 predictions
6.8/10
Your assessment

Plan ahead to 2030

Like the Data Voices, rank the 6 predictions in order of importance for your organization, then assess your confidence in achieving them by 2030. You’ll then receive your priority, overall confidence level and the quick win to kickstart progress.

For each prediction: set your priority from 1 to 6 (a rank already chosen is no longer offered for the other predictions), then drag the slider from 0 to 10. The “DV” marks show the ranking and average score of the Data Voices.

Prediction #1DV No. 3
5.0/10vs DV
Prediction #2DV No. 1
5.0/10vs DV
Prediction #3DV No. 5
5.0/10vs DV
Prediction #4DV No. 6
5.0/10vs DV
Prediction #5DV No. 2
5.0/10vs DV
Prediction #6DV No. 4
5.0/10vs DV
Take action now

The 6 essential quick wins

Take these actions now to start your transformation into an AI-first company and deliver on the Data Voices’ predictions by 2030.

Quick win#1
Data quality targeted at 2-3 priority business use casesThe fastest and most cost-effective action: qualify the actual state of the data needed for real use cases, without waiting for a global data quality program.
Quick win#2
Formalize a Shadow AI policy and provide associated solutionsNeither total blocking nor laissez-faire: secure experimentation perimeters and controlled data access tools.
Quick win#3
Launch your first data product on a limited basisWith an identified business owner, a defined SLA, and accessible documentation – rather than a large program.
Quick win#4
Deploy a consumption platform to democratize dataA data product marketplace to expose assets in the form of documented and self-service products.
Quick win#5
Implement data and AI cost monitoring in productionA dashboard for monitoring the real TCO to manage trade-offs and avoid investments that are not widely adopted.
Quick win#6
Develop the CDO's strategic credibility on LLMs and agentic AIMaster the fundamentals as a driver for cultural leadership – peer-to-peer benchmarking is the fastest approach.
Quotes from the Data Voices

Insights from the Data Voices in just a few words

Excerpts from the qualitative interviews conducted for the Manifesto. Read the full interviews by downloading the Data Voices Manifesto.

Aldrick ZappelliniChief Data & AI Officer · Crédit Agricole Group

We want the data marketplace to become the foundational layer of our shared culture. The goal is no longer theoretical training, but transforming behaviors by thinking data-first from the design stage and by highlighting the collaborative and business aspect of data sharing.

Samia BoujatiouiGroup Head of Data Management · Groupe Coface

The data product, understood as a set of processed, qualified, and packaged data, is no longer optional. It is the foundation that enables us to move from raw data to a strategic asset. The only way to truly empower business teams is to give them ownership through the internal data marketplace which we are deploying, and which marks the end of IT’s monopoly over data.

Michel LutzChief Data Officer & Digital Factory Head of Data & AI · TotalEnergies

The ambition for 2030 is to move from localized task augmentation to a genuine redesign of processes, conceived from the beginning with AI as a partner. This also requires designing much more open and interconnected IT systems and technical architectures.

Camille MaireHead of Data Governance / MDM / Data Quality · Groupe ETAM

Believing that innovation can happen through a new system without addressing the fundamentals, upstream processes, and legacy systems is an illusion. If the entire process, from data entry to data consumption, is not considered, new systems will merely reproduce or amplify existing errors.

Chafika ChettaouiChief Data & AI Transformation Officer · AXA France

Scaling AI first requires aligning priorities with the company’s strategic vision and carefully selecting a limited number of use cases that genuinely impact profit & loss, customer satisfaction, and operational efficiency.

Want to know more?

The full Manifesto brings together all the quantitative results and their analysis, as well as the five Data Voices interviews in full.

The Data Voices who contributed to this Manifesto

Thank you to the leaders who are driving the world of data and AI forward.

Adrienne RostaingGuerlainAldrick ZappelliniCrédit Agricole GroupAmar AmarnathKAPSARCAngélique Bidault-VerliacGroupe SNCFArielle RollandAccentureArthur SarazinVeltysBertrand BilloudSNCFBruno Espírito SantoE-REDES (Groupe EDP)Camille MaireGroupe ETAM
See all (58) See less