From idea to impact: creating successful AI Proof of Concepts

Key considerations to maximise your AI business value

Contact our experts

The potential applications of AI in business today are virtually limitless. Organisations across all sectors – including yours probably - are brimming with ideas for AI-powered improvements. From improving your customer experience to optimising your operations or automating routine tasks, the possibilities are endless. However, with so many options, the crucial challenge becomes identifying which ideas are worth pursuing and how to effectively scale them to a Proof of Concept – or as we at BDO prefer to call them ‘Proof of Value’. 

This is the first article in our two-part series on AI implementation. While this article focuses on developing and validating AI ideas through proof of concepts, it's important to note that a proof of value alone doesn't create business value – it's the foundation for successful production implementation, which we'll explore in our follow-up article. 

In this article, we will explore a couple of key considerations for selecting, developing, and validating your AI ideas and turning them into proof of concepts: why is one idea ‘better’ than the other? And on what aspects should you base this decision?  

Balancing top-down and bottom-up approaches

When thinking about developing AI ideas, organisations typically follow either a top-down or bottom-up approach. Top-down makes sure your idea is aligned with your business goals but often produces concepts that aren't feasible or don't address practical challenges. On the other hand, bottom-up leverages the knowledge and experiences of your operational employees but can lead to scattered efforts with limited strategic impact. 

A balanced approach combines the strengths of both methodologies and focuses especially on the strategic alignment. Having ideas in line with your organisation’s vision and ambitions is essential for ensuring that AI investments contribute meaningfully to your business goals rather than becoming isolated technological experiments.  

This strategic foundation is critical, as we explored in our recent article on creating long-term AI value in your organisation, sustainable AI success requires a capability-first approach that balances technical excellence with business integration. By integrating strategic direction with operational insights, you can identify AI initiatives that are practically viable and deliver real business value. 

Simplify before digitalise

Before applying the power of AI to any business process, it’s crucial to firstly optimise your processes and data. Streamlining workflows, removing unnecessary steps, and clarifying decision points creates a solid foundation for effective AI enhancement. When evaluating processes for potential AI implementation, ask yourself: 

  • Is the current process clearly defined and consistently followed?
  • Are there unnecessary complexities that could be eliminated?
  • Are decision points and criteria well-documented?

This process documentation approach has proven valuable beyond immediate operational benefits. For instance, a real estate development company recently mapped their processes for a new application landscape selection. This documentation exercise unexpectedly positioned them to deploy AI solutions much faster in the future, as the foundational process clarity was already established. 

High-quality, well-prepared data is equally as essential for a successful AI implementation than streamlined processes. 
Even sophisticated AI models will produce poor results if they are trained on inadequate data. 

A comprehensive data readiness strategy requires attention to five foundational layers. 
These elements can be implemented pragmatically and incrementally, tailored to your specific AI objectives and organisational context:

🏛️
Governance
Create organisational structures for effective data management across your enterprise and stakeholders. This includes establishing clear ownership, accountability, and decision-making processes for data-related initiatives.
⚙️
Processes
Establish standardised processes for data management, including exception handling, maintenance protocols, archiving procedures, and quality control mechanisms that ensure consistency across your organisation.
📊
Content and quality
Develop data standards, comprehensive inventories, clear definitions, traceability systems, quality measures, and integration protocols. This foundation supports both current and future AI applications.
🔧
Systems and tools
Implement the technological infrastructure needed to support governance frameworks, streamlined processes, and quality control measures. This includes platforms for data integration, monitoring, and analytics.
🎓
Culture and data literacy
Foster an organisation-wide understanding of data value and best practices. This cultural foundation ensures sustainable adoption and continuous improvement.

The key is to prioritise these layers strategically based on your immediate AI goals. Start with the elements most critical to your initial AI initiatives and build upon this foundation as your AI capabilities mature. 

Investing time in data preparation significantly increases the likelihood of success and pays dividends throughout the implementation process. A recent project with a major telecommunications provider illustrates this principle perfectly. Facing challenges with operational excellence across their extensive facility network, the company needed to monitor data quality and identify human errors within their facility management system. 

By implementing a comprehensive data quality surveillance solution, they established continuous monitoring capabilities with detailed error categorisation according to predefined rules. This investment in data governance and quality has created a foundation that will enable AI initiatives to deliver significantly greater value when deployed across their operations. 

Organisations that simplify processes and structure data before applying AI typically achieve higher success rates and faster implementation. 

Generating strong AI ideas – the Double Diamond methodology

Coming up with effective AI initiatives requires you to build a structured engagement across all your organisational levels. We recommend using the Double Diamond methodology during AI discovery workshops to ensure comprehensive exploration of potential opportunities. 

This methodical approach involves collecting diverse ideas through different brainstorm sessions. These workshops bring together every stakeholder in your organisation (C-level executives, team leads, process owners,…) to each bring their own unique perspectives. By capturing insights from different organisational perspectives, you develop a more comprehensive understanding of the challenges and opportunities within your organisation and define concrete AI ideas. 

Want to know more about the Double Diamond approach or want us to guide you by organising an AI Discovery workshop in your organisation? 

Contact one of our BDO experts for more info!

Strategic selection: impact vs. AI readiness

With multiple AI opportunities identified, the next step is evaluating each idea by comparing impact versus AI readiness. 
We recommend using an impact versus AI readiness quadrant to plot each opportunity – ideally targeting ideas with high impact and high AI readiness, such as well-defined processes with quality data already in place. When selecting your initial AI projects, resist the temptation to tackle the most ambitious ideas immediately. Instead, to build momentum, it’s recommended to develop a couple of quick wins to demonstrable the value of AI to the full organisation and key stakeholders. This will offer you extra support to implement broader, more ambitious AI initiatives.  

A typical ideal quick win: 

  • addresses recognised pain points 
  • has clear, measurable value 
  • can be implemented relatively quickly uses readily available, quality data. 


For initiatives that show high impact potential but low AI readiness, define specific follow-up actions to bridge these gaps: 

🔧
Technical readiness
If data quality or infrastructure isn't sufficient, establish improvement roadmaps including data governance implementation, system upgrades, or skill development programs.
🏢
Organisational readiness
For cultural or process barriers, create change management plans that include stakeholder engagement, training programs, and gradual adoption strategies.
🎯
Strategic readiness
When alignment with business objectives is unclear, refine the business case and establish clearer connections to organisational priorities.

Tip: Both successful as well as unsuccessful outcomes provide valuable learnings for future initiatives. Don’t just focus on the positively received proof of concepts.

Defining success: from technical proof to business value

Before developing and implementing your AI ideas, you should also focus on defining clear success metrics. Without predefined criteria, it's difficult to objectively evaluate outcomes and determine when an initiative is successful.  

In defining this success, the distinction between proof of concept (PoC) and proof of value (PoV) is critical. As we see it, a PoC validates technical feasibility, while a PoV confirms actual business value and user adoption. For example, an AI-powered contract analysis tool might achieve 95% technical accuracy (successful PoC) but fail to be adopted by the legal team due to poor integration with their workflow (unsuccessful PoV). 

Define both technical performance metrics and business value metrics before implementation begins. Technical metrics might include accuracy rates and processing speed, while business metrics should cover time savings, error reduction, and user adoption. For a comprehensive approach to measuring AI value across multiple dimensions and timeframes, see our detailed guide on AI ROI: a pragmatic measurement approach.

Developing and implementing AI proof of concepts is an iterative process requiring ongoing assessment and adjustment. Rather than a one-time deployment, a valuable idea involves continuous cycles of evaluation and refinement. When you want to continuously improve these concepts, think about implementing: 

  • Regular performance reviews against predefined metrics 
  • Feedback mechanisms for end users 
  • Processes for retraining models with new data 

After experimenting with proof of concepts, it’s time to move on to production deployment. Another complex process involving several critical steps. One essential aspect to keep in mind is – just like we said at the beginning of this article – to make sure the implementation remains aligned with your strategic business objectives. This alignment check should be performed at each milestone to maintain the scope of your project and ensure the final solution delivers the intended business value. The most successful AI implementations are in line with this strategic coherence from initial concept through full production deployment. 

Ready to turn your AI ideas into value? 

A balanced approach to AI implementation combines strategic vision with operational pragmatism. By taking into account the key considerations mentioned above, you can significantly increase the success rates of your AI implementations and turn your AI potential into profit. 

BDO's pragmatic AI journey

Successful AI implementations require expertise across strategy, process, technology, and governance. Our BDO experts can guide you across all implementation stages. What’s more, we don’t automatically apply AI when alternatives might be more effective, we focus on the right solution for your unique business needs. 

Our multidisciplinary approach focuses on three core elements: 

  • Strategic alignment: ensuring AI initiatives support business objectives 
  • Practical feasibility: validating that proposed solutions are implementable 
  • Measurable value: defining and tracking business impact

BDO Belgium's AI Services

BDO's experts can help you implement this pragmatic, results-focused approach to ensure you not only measure AI value but maximise it through proper implementation and governance.

Our multidisciplinary team combines financial expertise with practical AI implementation experience to deliver solutions tailored to your organisation's specific needs. Contact our team today to discuss how we can help you navigate the complex landscape of AI investment and value measurement.

Check our service page

Contact our BDO AI experts

Send us your questions, and our experts will provide the answers you need.

Interested in other AI related topics?

Discover our other AI articles here!