How to implement a data marketplace without creating another data silo

Listen to Murdio’s top experts’ tips on implementing data marketplaces in Collibra in a way that maximizes their value for the business. 

When it comes to the data landscape of today’s enterprise companies, the bottleneck has shifted from volume to access and trust. Companies are sitting on heaps of data, but that doesn’t mean that their employees can access the relevant and reliable data fast. And in a nutshell, that’s what data marketplaces are for.

There is a significant risk that comes with them, though. If you implement a data marketplace as a standalone tool without deep organizational integration, it simply becomes another expensive silo. A data graveyard where data assets go to be forgotten.

A true data marketplace acts as a centralized platform that bridges the gap between data providers and data consumers. It’s an active ecosystem where data products are curated, exchanged, and used to drive analytics and even fuel AI models. To succeed, companies need to look beyond the software license and focus on the underlying processes that allow for seamless data sharing and data exchange. And in this article, we’re talking about just that.

Key takeaways

• A data marketplace is not a standalone tool. You need to integrate it into your ecosystem to avoid becoming another isolated data silo.
• Out-of-the-box templates rarely survive the reality of enterprise compliance. Success requires customizing workflows to match unique organizational processes.
• Effective implementation treats data as a data product, complete with metadata, terms of use, and verified quality, rather than just raw technical tables.
• When moving from “home-made” tools to a professional platform like Collibra, a deep gap analysis is essential to preserve existing value while gaining scalability.
• Seamless connection with modern data platforms like Snowflake and Databricks is mandatory for a frictionless self-service experience for data consumers.

The “out-of-the-box” myth: Why technology isn’t enough

One of the most common misconceptions we see at Murdio is the belief that purchasing a license for a premier tool, such as Collibra, is the end of the journey. Some organizations think that once the platform is installed, the work is done. We’re calling it the “out-of-the-box” myth.

In reality, technology is only the enabler. Based on our expert conversations, a recurring theme is the failure of generic workflows.

Out-of-the-box solutions are excellent for a sales demo. They show a smooth, idealized path for requesting access to a data product. But in a complex corporate environment, no one uses these workflow templates without significant modification. Every organization has unique compliance requirements, legal checkpoints, and technical hurdles that generic setups simply can’t account for.

A critical part of implementing a scalable solution involves a deep gap analysis to identify where your current organizational habits clash with the structured environment of a professional data marketplace. Many of our most successful projects involved migrating a custom Data Marketplace to Collibra. 

Companies often start with a home-made tool, usually a combination of SharePoint, Excel, or a custom-built internal portal. And while custom tools serve an initial purpose, they lack the governance and integration capabilities of a dedicated platform. 

The transition to Collibra requires translating the unique, home-grown processes into a standardized framework. Migration moves metadata, but more importantly, it lets you refine the logic of how internal data and third-party data are requested and delivered.

Here’s an example case study of one of our clients: 

Case Study: Migrating a custom Data Marketplace to Collibra for a pharma company

How to implement a data marketplace in 5 core phases

To avoid the silo trap, we recommend a structured, five-phase approach that prioritizes value at every step. Here’s a rough outline.

Phase 1: Readiness and gap analysis

Before a single line of configuration is written, you need to understand your current state. That’s how we always start. Some essential questions to answer include: What data assets are currently most in demand? Who are the primary data providers? 

During this phase, you want to map out existing processes and identify the gaps between your current manual data exchange and the automated future state.

 

Phase 2: Defining the data product

The next step is to curate your data. This means transforming raw tables into a data product. 

A data product is a dataset with a specific purpose and with additional business context and a way to access it. Whether the data resides in Snowflake or Databricks, it needs to be packaged in a way that a data consumer can understand.

Phase 3: Technical integration and workflow customization

This is where the platform is actually built. And the building involves connecting your data platforms (like Snowflake or Databricks) to Collibra. More importantly, it’s where you replace generic workflows with custom logic for processes like:

• granting and revoking access
• creating data products
• editing access permissions
• etc.

This is necessary if you want to be sure that data sharing is both fast and compliant.

Phase 4: Pilot and feedback loop

Never launch to the whole company at once. Select a high-value Collibra use case instead. Perhaps a specific analytics team or a group working on AI models. 

Let them use the self-service features of the data marketplace and gather feedback on the data shopping experience.

Phase 5: Scaling and building an ecosystem

Once the pilot is successful, you scale by onboarding more data providers and integrating external data or third-party data. This phase focuses on making the platform the default destination for any data needs within the organization.

4 architectural pitfalls to avoid during implementation

Even with a phased approach, several architectural traps can derail your progress, so let’s quickly go through them.

1. Don’t over-complicate metadata collection. When you try to ingest every piece of metadata from your data platforms at once, it leads to one thing: noise. Instead, start with the data that you know people often use. For example, ask Data Analysts or Application Owners about the most commonly requested data, and go from there.
2. Don’t launch a platform without enough data products. If a data consumer searches the data marketplace and finds nothing three times in a row, they will never return. You need a critical mass of internal data available at launch.
3. Don’t ignore the provider experience. If it’s too difficult for data providers to list their assets, they simply won’t do it. The workflow for publishing a data product needs to be as streamlined as the workflow for requesting it.
4. Your data marketplace shouldn’t be an island. It needs to integrate with where the data lives. If your organization relies on Snowflake for warehousing and Databricks for data science, the marketplace has to seamlessly pull technical context from them so that the metadata stays current.

How to drive data marketplace platform adoption

Adoption is the ultimate metric for success. And it takes a smart strategy to get people to use the data marketplace just as you want them to. Here are a few tips from Grzegorz Jabłoński, one of our top experts in data marketplace implementation:

• The self-service interface needs to be intuitive. Searching for a data product should feel as easy as searching on Amazon.
• Use trust signals like certification badges, top-rated assets, and clear governance owners. When data consumers see that a data product is “Gold Certified,” they’re more likely to use it for critical analytics.
• Shift the culture from data hoarding to data sharing. Highlight success stories where the data marketplace significantly reduced the time-to-insight for a specific use case.
• The biggest barrier to adoption is the access wait time. Use the workflow engine to automate approvals where possible. If a user requests a low-sensitivity data product, the platform should grant access in Snowflake or Databricks automatically.

“If the platform is technically perfect but no one uses it, the ROI is zero. To drive adoption, you need to treat the data marketplace like a consumer product and really nail the user experience.”

Grzegorz Jabłoński, Collibra Solutions Architect

How to build a business case for a data marketplace platform

Securing a budget for a data marketplace implementation requires moving beyond technical jargon and focusing on business outcomes. A strong business case can focus on four pillars:

• Operational efficiency 

Calculate the time currently wasted by data consumers looking for data and data providers answering repetitive questions. Think of it as a kind of data discovery tax, reduced by a centralized platform. When you enable self-service, you free up your most expensive data engineering resources from manual fulfillment tasks and data specialists continuously answering questions from people who don’t know where to find the data they’re looking for.

• Accelerated time-to-value 

If it takes three weeks to get access to a dataset, the business opportunity might have already passed. A data marketplace shortens this cycle from weeks to minutes, allowing for more agile decision-making.

• Risk mitigation and governance 

Manual data sharing is a compliance nightmare. When you centralize data exchange within a governed platform, third-party data licenses are respected, and sensitive internal data is only seen by authorized eyes. And you reduce the risk of regulatory fines and data breaches.

• Monetization and cost optimization 

For some organizations, the data marketplace can even facilitate external data sharing with partners, potentially creating new revenue streams. Internally, it helps identify redundant datasets, allowing the organization to stop paying for third-party data feeds that are already being purchased by another department.

• Data democratization

People who are not technical experts can now easily find, access, and understand important data. And this benefits the entire business.  

The bottom line

There’s one key shift in perspective that needs to happen for a successful data marketplace implementation: you need to move from a tool-based approach to an ecosystem-based one. Otherwise, you’re creating another silo, and that’s counterproductive for any enterprise.

Need help building or updating a data marketplace? Call us – we have experts who know how.