--- title: "Data Strategy" date: 2026-07-12 description: Every AI system runs on data the balance sheet cannot see, so Boards should govern, structure, and monetise it as the asset it is. author: Mario Thomas canonical: https://mariothomas.com/briefings/data-strategy/ --- ## Start here Two short essays to orient you: what governing data as an asset actually means, and the order to take this briefing in. ### The Asset the Balance Sheet Cannot See Every AI conversation arrives at data eventually, and most arrive too late. Data strategy, as this briefing treats it, is the governance and commercial discipline applied to the asset every AI system runs on: the proprietary records, relationships, and knowledge that may now contribute more to enterprise value than the buildings that house them. Global intangible value reached an all-time high of **USD 97.6 trillion in 2025** (Brand Finance 2025), yet accounting standards prohibit capitalising internally generated data, so the balance sheet never sees the data, and the stewardship disciplines that follow the balance sheet never reach it. What goes wrong is rarely ambition. It is altitude. The Boards I meet seldom lack a data strategy on paper; what they lack is stewardship. Data initiatives are delegated to IT as technology projects, ownership fragments across functions, quality degrades undetected, and too often nobody assesses whether the data is fit for the purposes AI is now asking of it. A physical asset treated this way would be condemned. Data, in governance terms, already is. The position these articles take is that the disciplines already exist. Boards know how to steward assets: named ownership, regular condition assessment, maintenance, and impairment testing. Extending them to data is a governance choice, not an accounting one, and it does not need to wait for the standards to be reformed. The argument then goes a level deeper. Quality tells the organisation whether its information is reliable; structure tells the machine what it means. The definitions encoded in an organisation's [ontologies and knowledge graphs](/blog/ontology-knowledge-graph-explainer/) now carry the weight a chart of accounts has always carried, because they shape what the organisation's AI systems treat as true. And because proprietary data is the AI input competitors cannot purchase, it is also where the commercial opportunity sits. I have been monetising information assets since I began digitising 271 years of newspaper content in 1998, and the pattern has not changed: identify the assets, structure them, and build value exchange models that reward the organisations that created them. The AI era raises the stakes on both sides of that equation, multiplying what proprietary data is worth and testing whether its owners capture any of it. Read this briefing and you should be able to make one judgement about your own organisation: whether it is stewarding its data as the asset its AI strategy silently assumes it to be, and if not, which of ownership, quality, structure, or commercial model is the gap the Board should close first. ### From Stewardship to Structure to Revenue The core articles are numbered for a reason, and the sequence is the argument. Start with [the articles](#core-reading): The Invisible Asset makes the governance case, establishing why data deserves balance-sheet discipline even though the balance sheet cannot show it. Ontologies and Knowledge Graphs then takes the argument a level deeper, from whether data is reliable to whether a machine can reason over it, which is where durable AI advantage is now being decided. The sequence then turns to the commercial question. The Enterprise Data Advantage sets out how to identify, value, and monetise data assets, drawing on the newspaper digitisation work where I first learned the trade. Creating Sustainable Value in the AI Era closes the sequence with the question the earlier articles raise: once AI systems consume and synthesise proprietary data, how does its owner make sure the value flows back? The supporting pieces in [the further reading](#further-reading) show the same thinking delivered to live audiences, at a private equity evening and on a conference keynote stage. Alongside them sit the pieces that take the argument to its foundations: the data loop from The Great Remaking, which shows why proprietary operational data compounds and cannot be bought, and the four indicator types that turn data into decision fidelity in the boardroom. After the articles, [Remake](#remake-assets) holds the mechanisms beneath the thinking, worth a look wherever a named asset is unfamiliar. [The questions](#faqs) gather the ones directors put to me most often on this subject, each answered in a paragraph with a route back into the relevant article, and [the references](#references) list the external research, standards, and legislation the articles cite, if you want the primary evidence. If you have 15 minutes, read the first article and the questions. If you have an evening, read the core articles in order; they were written independently, but they build to a single position, and the sequence is the shortest route to it. ## Core reading The sequence runs from governing data as an invisible asset, through structuring it so machines can reason over it, to monetising and protecting its value in the AI era. 1. [The Invisible Asset: Why Boards Should Govern Data Like It's on the Balance Sheet](https://mariothomas.com/blog/data-invisible-asset/) (9 minute read, 1 February 2026): Boards steward physical assets with condition checks and ownership, and govern data as if it did not exist. The gap is governance, not technology. Podcast edition: 13 minute listen. 2. [Ontologies and Knowledge Graphs: Why Structure is the Next Data Frontier](https://mariothomas.com/blog/ontology-knowledge-graph-explainer/) (15 minute read, 31 May 2026): Quality tells the organisation whether data is reliable. Structure tells the machine what it means, and structure is where durable AI advantage is now decided. Podcast edition: 17 minute listen. 3. [The enterprise data advantage: Turning information assets into strategic value](https://mariothomas.com/blog/enterprise-data-monetisation/) (15 minute read, 27 January 2025): Data becomes an asset when its structure, meaning, and relationships are known. A 1998 newspaper archive taught me how information turns into value. 4. [From Print to Web to AI: Creating Sustainable Value in the AI Era](https://mariothomas.com/blog/protecting-value-ai-era/) (12 minute read, 17 August 2025): AI answer engines are rewriting how value flows through information, as the web did to print. Value exchange now has to be designed, not assumed. ## Further reading - [Demystifying data monetisation: Insights for private equity portfolio companies](https://mariothomas.com/blog/chief-wine-officer-spring/) (3 minute read, 13 February 2025): Data turns into value when a portfolio company knows what it holds and how to use it: my Chief Wine Officer talk for private equity. - [Unlocking your data with AI: Insights from Monday.com Elevate](https://mariothomas.com/blog/monday-elevate-ai-data-keynote/) (10 minute read, 19 September 2024): From my Monday.com Elevate keynote: the misconceptions about AI adoption, and how to put the data an organisation already has to work. - [The Great Remaking: Why Fast Following Does Not Work When the Gap Compounds](https://mariothomas.com/blog/the-great-remaking-fast-following/) (13 minute read, 15 March 2026): Every previous technology wave rewarded fast followers. The Great Remaking does not: the advantage is operational accumulation that cannot be bought and compounds with time. Podcast edition: 16 minute listen. - [Maximum Fidelity: How Four Indicator Types Strengthen Board Decisions](https://mariothomas.com/blog/maximum-fidelity-four-indicators/) (13 minute read, 12 April 2026): Four indicator types give boards progressively higher decision fidelity: lagging, leading, predictive, and reasoned. Together they represent the most accountable governance instrument available. Podcast edition: 15 minute listen. ## Remake The mechanisms beneath the thinking: the model, diagnostic, methodology, and principle from the Remake Library that turn this briefing into apparatus a Board can use. - **Model: Maximum Fidelity**. The evidence discipline of grading every reading a Board relies on by one of four indicator types: lagging indicators of past outcomes, leading indicators of early signals, predictive indicators of future value, and reasoned indicators that prove what must hold true. [Remake Library](https://mariothomas.com/remake/library/#maximum-fidelity) - **Diagnostic: Five Pillars of AI Capability**. The five capability domains of an AI capability model that cut across every level of maturity: Governance and Accountability, Technical Infrastructure, Operational Excellence, Value Realisation and Lifecycle Management, and People, Culture and Adoption. [Remake Library](https://mariothomas.com/remake/library/five-pillars/) - **Methodology: AI Business Case**. The integrated decision framework that crystallises across an ADAPT engagement rather than at a single stage: strategic alignment established at Align, cost and readiness evidenced at Diagnose, value shaped at Advise, and execution designed at Plan. [Remake Library](https://mariothomas.com/remake/library/#ai-business-case) - **Principle: Well-Advised**. The framework of five strategic priorities, Innovation, Customer Value, Operational Excellence, Responsible Transformation, and Revenue, used to ensure AI investments create balanced value rather than narrow cost reduction. [Remake Library](https://mariothomas.com/remake/library/well-advised/) ## Questions The questions directors put to me most often about their organisation's data, answered from the articles in this briefing. ### Why govern data as an asset when it never appears on the balance sheet? Because the invisibility is an accounting artefact, not a statement of value. IAS 38 prohibits capitalising internally generated data, yet intangible assets now dominate enterprise value, and poor data quality carries a measurable cost. The disciplines Boards already apply to physical assets (named ownership, condition assessment, maintenance, and impairment testing) extend naturally to data. I set out the case, and the governance architecture for it, in [The Invisible Asset](/blog/data-invisible-asset/). ### We have invested heavily in data quality. Is that enough for AI? Quality was necessary and remains so, but it answers a different question. Quality tells the organisation whether information is reliable; structure tells the machine what it means. A model with access to accurate but flat data can summarise; it cannot reason over relationships that were never made explicit. That is why ontologies and knowledge graphs have moved into Board territory, and I explain both, without the jargon, in [Ontologies and Knowledge Graphs](/blog/ontology-knowledge-graph-explainer/). ### Who should own data strategy and its governance? Somewhere with Board-level visibility, not delegated to IT where the strategic perspective may be limited. For most organisations the [AI Centre of Excellence](/blog/data-invisible-asset/) is the natural vehicle for data stewardship, while the deeper structural layer, the 'chart of entities' that defines what a customer or a contract officially is, usually belongs with the Chief Data Officer. The useful first step is simply asking who owns it now; most organisations find they cannot answer it. ### Do we need perfect data before we can create value from it? No, and waiting for perfection is usually the more expensive choice. When we digitised 271 years of newspaper content, the OCR was imperfect and the metadata minimal, and the syndication market paid for it anyway. Start with the data the organisation already has, prove value through focused pilots, and improve quality where the returns justify it. The approach is set out in [The Enterprise Data Advantage](/blog/enterprise-data-monetisation/) and in my [Monday.com Elevate keynote](/blog/monday-elevate-ai-data-keynote/). ### AI systems are consuming our proprietary content. How do we protect the value? By building value exchange models rather than walls. Bloomberg turned its proprietary financial data into BloombergGPT and premium terminal capability; the Financial Times built a subscription business that funds its journalism. Both show that the organisations investing in high-quality proprietary data can capture the value AI multiplies, through licensing, APIs, and tiered access. I draw out the lessons in [Creating Sustainable Value in the AI Era](/blog/protecting-value-ai-era/). ### Should we approve AI budgets without a data asset health assessment? No, or not without knowing what the money will run on. No Board would approve a facilities budget without a condition report on its buildings, yet AI budgets too often pass without an equivalent assessment of the data beneath them, and 63% of organisations lack or are unsure of the data management practices AI requires (Gartner 2025). The assessment need not be elaborate: which proprietary datasets the proposal depends on, who owns each, and how complete, accurate, and current they are. Plotted by strategic value against present quality, the case either stands or shows where the money should go first. I set out the approach in [The Invisible Asset](/blog/data-invisible-asset/). ### Which data definitions warrant Board attention? The few that decide what the organisation officially knows. The Board should not author definitions; that work is rightly delegated. Its job is to see that the choices are being made, and that the handful touching revenue, risk, regulatory exposure, and customer trust are visible, defensible, and aligned with strategy and risk appetite. What counts as a customer, a contract, an employee, or an incident now propagates into every report, obligation, and AI decision downstream, much as a chart of accounts fixes what every figure means. The Minimum Lovable Governance principle applies: govern those few deliberately and delegate the rest. I make the case in [Ontologies and Knowledge Graphs](/blog/ontology-knowledge-graph-explainer/). ### Should we share our data openly or keep it proprietary? Both; the work is deciding which data belongs in which camp. Public-good data thrives on openness, and sharing non-core data builds goodwill and invites innovation. Proprietary data that took real investment to curate, verify, and maintain needs a value exchange model if that investment is to continue. The models exist: licensing with attribution, API access on Bloomberg's pattern, metered subscription as the Financial Times runs it, and Cloudflare's 2025 pay-per-crawl, which lets owners charge AI crawlers rather than block them. The hybrid I argue for in [Creating Sustainable Value in the AI Era](/blog/protecting-value-ai-era/) opens what costs little to share and builds premium services around what cost real money to create. ## References The primary research, standards, and legislation the articles in this briefing draw their evidence from. - **IFRS Foundation** (April 2001): [IAS 38 Intangible Assets](https://www.ifrs.org/issued-standards/list-of-standards/ias-38-intangible-assets). IAS 38 Intangible Assets, the standard that keeps internally generated data off the balance sheet. - **CFA Institute** (February 2025): [Investor Perspectives: Intangible Assets](https://rpc.cfainstitute.org/sites/default/files/docs/surveys/intangibles-report_online.pdf). Research on intangible assets showing S&P 500 market value running at over four times reported book value. - **Brand Finance** (5 November 2025): [GIFT™ 2025: the annual review of the world’s intangible value](https://brandfinance.com/insights/gift-2025-the-annual-review-of-the-worlds-intangible-value). GIFT 2025, the annual review that put global intangible value at an all-time high of USD 97.6 trillion. - **Ocean Tomo** (9 February 2026): [Intangible Asset Market Value Study](https://oceantomo.com/intangible-asset-market-value-study). The long-running study finding intangible assets represent 90% of S&P 500 market value. - **Gartner** (25 February 2025): [Lack of AI-Ready Data Puts AI Projects at Risk](https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk). 2025 finding that 63% of organisations lack, or are unsure of, the data management practices AI requires. - **Institute of Directors** (2025): [AI Governance in the Boardroom](https://web.archive.org/web/20251121075957/https://www.iod.com/app/uploads/2025/09/AI-Governance-in-the-Boardroom-1c7612e872fa3fce3f9d6cad78b0b4ba.pdf). AI Governance in the Boardroom (2025), including director concern over missing data governance frameworks. - **PwC** (2025): [The Fearless Future: 2025 Global AI Jobs Barometer](https://www.pwc.com/gx/en/issues/artificial-intelligence/job-barometer/2025/report.pdf). Global AI Jobs Barometer 2025, evidencing threefold revenue-per-employee growth in AI-exposed industries. - **Financial Reporting Council** (2024): [UK Corporate Governance Code 2024](https://www.frc.org.uk/library/standards-codes-policy/corporate-governance/uk-corporate-governance-code/). The 2024 UK Corporate Governance Code, whose expectations press on explainability. - **legislation.gov.uk** (2025): [Data (Use and Access) Act 2025](https://www.legislation.gov.uk/ukpga/2025/18/contents). The Data (Use and Access) Act 2025, giving individuals the right to contest solely automated significant decisions. - **McKinsey & Company** (1 December 2017): [Fueling growth through data monetization](https://www.mckinsey.com/business-functions/quantumblack/our-insights/fueling-growth-through-data-monetization). Research finding high performers three times more likely to earn over 20% of revenue from data monetisation. - **Cloudflare** (1 July 2025): [Introducing pay per crawl: Enabling content owners to charge AI crawlers for access](https://blog.cloudflare.com/introducing-pay-per-crawl/). Cloudflare's 2025 pay-per-crawl mechanism, letting content owners charge AI crawlers for access rather than block them. - **arXiv** (30 March 2023): [BloombergGPT: A Large Language Model for Finance](https://arxiv.org/abs/2303.17564). BloombergGPT, the 2023 paper on a large language model trained on proprietary financial data, the example behind the case for owning the data an AI system runs on. - **Harmonic Security** (31 July 2025): [GenAI Data Exposure: What GenAI Usage Is Really Costing Enterprises](https://www.harmonic.security/blog-posts/genai-data-exposure-report-fa6wt). Harmonic Security's 2025 research on what employees paste into generative AI tools, the exposure that makes data governance a Board matter. ## The ideas beneath this briefing Ideas I've named and matured writing about Data Strategy: what each one means, and where it started. - **Chart of Entities**: The governed set of core entity definitions that, like a chart of accounts, determines what every AI system in the organisation treats as true. [Read more](https://mariothomas.com/blog/ontology-knowledge-graph-explainer/) - **Unified Platform Experience**: An approach that adds value through an abstraction layer over legacy systems, achieving quick wins and faster time to value without massive system overhauls or full platform integration. [Read more](https://mariothomas.com/blog/chief-wine-officer-spring/) - **Data Loop**: A compounding cycle in which AI-integrated workflows generate higher-quality, better-structured operational data that feeds back into an organisation's AI systems, improving their performance and, in turn, producing still better data over time. [Read more](https://mariothomas.com/blog/the-great-remaking-fast-following/) - **Process Redesign Loop**: The institutional capability for redesign itself: accumulated knowledge, cross-functional relationships, and governance that make each successive restructuring of work around AI faster, compounding organisational learning capacity rather than just productivity. [Read more](https://mariothomas.com/blog/the-great-remaking-fast-following/) - **Talent Loop**: The compounding advantage that accrues as people develop tacit, experiential AI-collaboration capabilities through sustained practice in redesigned workflows; competencies that cannot be hired or trained quickly and that attract further capable talent. [Read more](https://mariothomas.com/blog/the-great-remaking-fast-following/) All concepts: https://mariothomas.com/glossary/concepts/ ## More Board Briefings More complete resources on AI and emerging technology for the Boards that need the full picture. - [AI Strategy](https://mariothomas.com/briefings/ai-strategy/): Approving good AI projects is not a strategy, and the Board's move is from accumulating pilots to a strategy it owns. - [AI Infrastructure](https://mariothomas.com/briefings/ai-infrastructure/): AI infrastructure now runs from the power station to the protocol, and energy access decides what an organisation can do with AI. - [AI Business Cases](https://mariothomas.com/briefings/ai-business-cases/): Business cases built for predictable payback misread AI, whose value arrives in parallel, late, and elsewhere; Boards need different instruments. - [AI Governance](https://mariothomas.com/briefings/ai-governance/): Governance people route around fails to govern; the task is governing AI the Board cannot fully see without strangling adoption.