Platforms Don't Make Money. Products Do.

Why Data as a Product Has Become a Board-Level Conversation

Over the past decade, enterprises have poured extraordinary sums into modernising their data estates. Lakehouses have replaced legacy warehouses. Governance frameworks have matured. Dashboards proliferate. And yet, in boardroom after boardroom, the same question keeps resurfacing: if our data platform is so advanced, why hasn't it moved the needle on revenue?

This is a conversation dominating the agendas of many Chief Data Officers, Chief Digital Officers, CIOs and Heads of Data & AI. The infrastructure is in place. The pipelines run. The data is, technically, ‘available’ - and availability is where most organisations stop.

Across many years of data strategy engagements, we have watched “data monetisation” settle into the same slot in almost every deck we are shown: the final slide, pointing at some distant and undefined future. It rarely gets more specific than that.

The reason is simple: platforms don't make money. Products do.

The Limits of Treating Data as Infrastructure

A data platform is a capability. It stores, processes, governs and moves data, all of which is necessary and none of which is sufficient. Treating data solely as infrastructure, as an asset to be managed rather than a product to be delivered, caps its commercial potential at ‘cost avoidance’ and ‘risk reduction’.

What's changed is the level at which the conversation is now happening. Boards are asking:

• What products can we build from our data - for internal teams and external customers?
• What decisions can we automate, at scale, with confidence?
• What new revenue streams can this unlock?

These are product questions and answering them requires a shift in mindset from managing data as an asset on the balance sheet, to designing, shipping and iterating on data as a product with customers, value, and accountability attached.

What ‘Data Product’ Actually Means

There's no shortage of definitions. DJ Patil's, Data Jujitsu: The Art of Turning Data into Product ¹ describes a data product simply as "a product that facilitates the end goal through the use of data." Zhamak Dehghani's Data Mesh frames data products around discoverability and accessibility. Both are useful, but the definition that resonates most with practitioners borrows directly from classic product management: Marty Cagan's four-part test for what makes something a product at all.

To qualify, an offering needs to be:

• Valuable - it solves a problem someone actually has, internal or external.
• Usable - consumers shouldn't need to wade through ten thousand pages of documentation to extract value from it.
• Feasible - buildable with the skills and technology the organisation actually has.
• Viable - affordable to produce, and compliant with the laws and contracts that govern the data underneath it.

It's the same test every one of us applies, consciously or not, to everything we buy. If it fails on any dimension, the thing isn't a product for you, however well it performs on the other three.

This reframes the starting question. Instead of asking "what data do we have?", we ask: does this solve a problem that a business, a department, or an individual has today?

Every organisational strategy ultimately alters at least one of three levers: increase revenue, reduce costs, manage risk. A data product should move at least one of those levers. If it doesn't, however impressive it looks, it's shiny. And shiny sells once before it tarnishes.

Why Forward-Thinking Organisations Are Making the Shift

The organisations pulling ahead in financial services, insurance, retail and healthcare share a pattern: they stopped waiting for a ‘monetisation moment’ and started packaging what they already do internally as products for internal customers first.

Once a data team learns to treat its own stakeholders as customers, with defined needs, agreed service levels, and value it can measure, it has already built most of what external monetisation demands. It has a roadmap, a support model, and someone accountable for when the product breaks. Selling to an outside buyer becomes a pricing question rather than a rebuild. This is also the point at which a data function stops being carried as a cost centre and starts operating as a profit centre.

Retail offers one of the clearest illustrations

A retailer's traditional customer is the shopper. But aggregated data on how products perform - by region, by demographic, by season, by basket composition - has value to a second customer entirely: the retailer's own supply chain. Suppliers gain benchmarked insight into how their products perform against category peers. The retailer gains a high-margin revenue stream. And the collaboration between them produces sharper personalisation and more accurate demand forecasting for both sides.

Tesco's Clubcard is the textbook case. According to dunnhumby (2025) more than 24 million UK households now hold one, in a country with just over 28 million households² and dunnhumby - wholly owned by Tesco since 2011 - turns that into the Tesco Media and Insight Platform, selling audience access and closed-loop measurement back to the brands whose products sit on Tesco's shelves. Tesco does not disclose what the platform earns. The category it created is not so shy: retail media networks took $174.2bn of global advertising budget in 2025, and WPP expects that to grow by a further 11.3% in 2026 ³. A loyalty scheme became an advertising business.

Not every data product needs an external buyer to justify itself. Consider a retailer whose shop-floor staff are employed by concessions and partner brands rather than by the retailer, but who still need the tools to deliver excellent clienteling: personalised service, and accurate attribution of sales against targets, commissions and incentive structures. Solving that identity and attribution problem once and packaging it as a reusable product rather than a one-off integration, produces something that compounds in value with every team that adopts it.

Why Built On Databricks

None of this happens on ambition alone. It requires a foundation capable of turning fragmented data estates into governed, trusted, reusable products at the pace AI now demands – and a route to market once those products exist.

Most data product ventures survive the build but falter at the handover; the moment the product has to reach a customer who runs a different stack, procures through a different process, and will not accept a nightly file drop.

The Databricks Data & AI Platform addresses both halves. Unity Catalog provides the governance and discoverability backbone that lets teams find, trust and safely reuse data products instead of rebuilding them. The lakehouse architecture removes the false choice between data and AI workloads. Lakeflow simplifies the engineering required to keep products fresh and reliable.

Distribution is the harder half, and it is where the platform earns its place in the monetisation strategy. OpenSharing – the open protocol Databricks pioneered as Delta Sharing and contributed to the Linux Foundation – lets you serve a live product to a recipient who is not a Databricks customer at all, without copying and without a replica ageing in someone else’s environment. Databricks Marketplace turns that share into a listing and lets buyers acquire against commitments they have already made, collapsing a procurement cycle that would otherwise run for months. Clean Rooms let you sell the insight where contract of regulation prevents you selling the data. All of this is governed through Unity Catalog, so the controls that satisfy your regulator are the controls that satisfy your customer.

When a Data Product Becomes a Business

For a growing number of organisations, the internal product turns out to be the prototype – what begins as a supplier insight portal or an attribution service acquires external customers, a price, a roadmap, and a support model. At that point you have stopped being only a Databricks customer and started being a software business with Databricks inside it.

Built On Databricks is the partner program designed for that transition. It provides architecture validation against the Partner Well-Architected Framework, direct access to Databricks product teams, marketing investment, and joint go-to-market with a sales organisation already present in your prospects’ accounts. In practical terms, it converts a data product from something you have built into something Databricks has a commercial interest in helping you sell.

Few data strategies plan for that step, which is why so many arrive at it unprepared – with a product that works, customers who want it, and no commercial route to either.

Getting There is the Hard Part

Between an internal data product and a commercially viable one sits a set of problems the platform will not solve on your behalf. How do you isolate one customer’s data from another’s inside a single governed estate without running a workspace per client and watching your margin disappear into overhead? How do you attribute compute cost to the customer who generated it, so that you can set a price and know what margin sits behind it? How do you meter usage and enforce entitlements, when not everything in the platform bills by the query? What does support look like when your consumer is external, and your service level is contractual rather than collegiate?

This is the work Advancing Analytics does. We have built out the Partner Well-Architected Framework as a pattern library – covering the governance, tenancy, cost management and scale requirements that validation tests. We have an approach to subscription and entitlement management for products built on Databricks, because metering and billing is where data products most often stall between pilot and revenue.

The Executive Opportunity

For CDOs, this is now a leadership capability as much as a technical one.

The work starts with identifying which existing data assets have product potential, and assessing each candidate against value, usability, feasibility, and viability. A framework of that kind does two jobs: it surfaces the opportunities worth funding, and it gives you a defensible reason to say no to the shiny distractions.

Governance then must be balanced against commercial ambition: you cannot sell your data as-is without giving away trade secrets or breaching contractual and regulatory boundaries, but overly conservative governance kills product velocity just as surely.

Finally, the strategy has to be sold internally before anything is sold externally - to peers at the executive table, who need to understand the risk alongside the return, and the teams who will build and operate the products day to day. Strategy shapes future commercial direction and buy-in at both the top and the bottom of the organisation is what separates a data strategy that is written and one that is executed.

The organisations that will define the next decade of enterprise data aren't the ones with the most advanced platforms. They're the ones that have learned to build and ship products on top of them.

---

Keen to explore more? Join me at Big Data LDN!

I’m hosting a workshop for senior data leaders at Big Data LDN on 24 September:

From Data Asset to Data Product: Built on Databricks

Work through examples of data productisation and monetisation, evaluate your own organisation's product opportunities, and learn how leading enterprises are turning data investments into measurable business value.

Places are limited and reserved for senior business and technology leaders.

Register your interest today to secure your place.

You’ll need to be registered for Big Data LDN too: Register Here

If you are already further along with a product identified, or a Built On Databricks validation in front of you, you can talk to us directly about the architecture, entitlement and operating model work that sits between the two.

---

References

¹ Patil, D.J. (2012) Data Jujitsu: The Art of Turning Data into Product. Sebastopol, CA: O'Reilly Media. Available at: O'Reilly Data Jujitsu (Accessed: 26 August 2026).

² dunnhumby (2025) Tesco Media Upfront 2025: Brands discover how Tesco is redefining retail media. dunnhumby, 10 October. Available at: Tesco Media Upfront 2025: Brands discover how Tesco is redefining retail media (Accessed: 26 August 2026).

³ Faull, J. (2026) Retail Media knows its special status on plans is over. And that's a good thing. The Drum, 29 January. Available at: The Drum article (Accessed: 26 August 2026).