How Scale‑ups Build an ROI‑Driven AI Strategy
20 Sept 2025
Learn how scale-up leaders can create ROI-focused AI strategies that align with business goals and drive meaningful transformation.
Artificial Intelligence (AI) is no longer a distant futuristic concept; it’s a present-day competitive advantage for businesses aiming to scale up. However, implementing AI effectively isn’t a plug-and-play solution - it requires strategy, clarity, and alignment with business goals. In a podcast discussion with Sazia, a senior manager of AI and data at EY, she shared insights on how SMEs can build an AI strategy that generates real business value while addressing common misconceptions and challenges.
This article distils the key points from the conversation and explores how UK-based founder-led SMEs can take actionable steps to harness AI as a business enabler.
Why Most AI Projects Fail: A Misalignment Between Tech and Business
One of the most striking observations Sazia made was that many AI projects fail - not because the solutions aren’t good enough, but because they aren’t tied to a meaningful business goal. She stated, "Amazing AI solutions can often sit on the shelf because they weren’t connected to a real business need." This disconnect is a major hurdle, particularly for executives who may see AI as a turnkey solution rather than a tool for transformation.
Common Misconceptions About AI Implementation
Sazia highlighted two major misconceptions at the executive level:
AI as a Magic Wand: Some believe AI is a ‘ready-to-go’ solution that doesn’t require foundational work. However, AI requires robust data pipelines, a clear organisational strategy, and well-structured goals.
AI as a One-Team Effort: Successful AI adoption requires collaboration across multiple teams - tech, risk, compliance, legal, and operations. It’s a collective effort rather than the responsibility of a single department.
Building an AI Strategy Aligned with Business Goals
The foundation of a successful AI strategy begins with asking the right questions. Instead of jumping to technologies like large language models or neural networks, Sazia recommends focusing on the following:
What frictions or inefficiencies exist in the business?
What decisions could be improved with foresight rather than hindsight?
What workflows or pain points could AI address to deliver the greatest impact?
By reframing these questions, leaders can identify specific problems that AI can solve, moving away from the "AI-first" mentality to a "transformation-first" approach.
The Role of AI Readiness Diagnostics
AI readiness diagnostics can help businesses evaluate where they stand in terms of AI adoption. Sazia developed a comprehensive AI maturity model that assesses organisations across five dimensions:
Data Foundations: Are data pipelines integrated and usable?
Technology Infrastructure: Does the organisation have the tools to deploy scalable AI solutions?
Talent: Is there a skilled team ready to implement and oversee AI projects?
Use Case Prioritisation: Are efforts being directed at the right problems?
Governance: Is there a robust framework in place to ensure security, ethics, and compliance?
This model doesn’t just identify technical barriers; it also uncovers cultural resistance, such as fear of automation or unclear ownership of AI initiatives.
The AI Roll-Out Journey: Four Key Phases
Sazia outlined the backbone of an AI implementation process, which can be tailored to different industries:
1. Exploration
Define the business problem and value hypothesis.
Identify relevant data assets and set clear goals, such as reducing patient no-shows in healthcare or optimising supply chains in retail.
2. Experimentation
Develop a proof of concept (POC) for a specific use case.
Test the POC in controlled environments to understand what works, what’s missing, and how the business responds.
3. Validation
Evaluate both technical and operational viability, including model accuracy, fairness, and explainability.
Test whether the solution integrates seamlessly into existing workflows and whether stakeholders trust and adopt it.
4. Scaling
Focus on governance, monitoring, and performance measurement.
Train teams to use the AI solution effectively while ensuring change management practices are in place.
Industry-Specific Considerations
Different industries approach AI adoption differently:
Manufacturing: Legacy systems often store rich data, but these must be integrated to support predictive maintenance and supply chain optimisation.
Healthcare: Data is highly sensitive, requiring stringent privacy measures and explainability to build trust.
Retail: The focus is often on improving customer engagement through AI-driven personalisation.
Balancing Quick Wins and Long-Term Vision
When selecting AI use cases, it’s essential to balance immediate wins with long-term transformative projects:
Quick wins: Address smaller problems with high visibility to build trust in AI’s potential. For example, automating call centre transcripts may not deliver massive ROI but can improve efficiency quickly.
Long-term projects: These require more time and resources but offer significant ROI. For instance, optimising supply chains or implementing predictive analytics in manufacturing.
Sazia emphasised, "You may want to start with low-risk, high-impact use cases to build momentum, but keep your eye on larger goals that can deliver exponential results."
Essential Skills for Future AI Leaders
AI leadership requires a blend of technical knowledge, strategic thinking, and ethical responsibility. According to Sazia:
Technical Literacy: Leaders don’t need to code but should understand machine learning fundamentals to ask the right questions.
Cross-Functional Collaboration: AI is inherently cross-functional, requiring leaders to bridge gaps between engineering, operations, and business teams.
Ethical Awareness: Responsible AI principles, such as bias mitigation and data privacy, must be ingrained in the design process.
She also stressed the importance of curiosity, saying, "Be calm in this AI storm and be the one in the room who asks the tough questions."
Key Takeaways
Start with transformation, not technology: Focus on business outcomes and align AI initiatives with organisational goals.
AI readiness matters: Evaluate your organisation’s maturity across data, technology, talent, use case prioritisation, and governance.
AI isn’t plug-and-play: It requires foundational work, collaboration, and cultural alignment.
Experiment, validate, scale: Follow a structured AI roll-out journey, adapting it to your industry’s specific needs.
Balance short-term wins with long-term ROI: Choose use cases strategically to demonstrate quick impact while working toward transformative goals.
Invest in AI leadership skills: Develop technical literacy, cross-functional collaboration, and ethical awareness to lead in this fast-changing space.
Trust through transparency: Build trust by making AI solutions explainable and accessible to all stakeholders.
Conclusion
For SMEs in the UK, building an ROI-driven AI strategy requires clarity, collaboration, and a relentless focus on business impact. By addressing common misconceptions, balancing short- and long-term priorities, and preparing leadership teams with the right skills, businesses can unlock AI’s transformative potential.
To paraphrase Sazia, "Everything possible with AI may not be purposeful. Start with why - and let that guide your journey."
Source: "How Top Companies ACTUALLY Build Enterprise AI Strategy | Ctrl + Shifter Podcast #5" - SmartDev - AI Powered Software Development, YouTube, Aug 20, 2025 - https://www.youtube.com/watch?v=01rzioo3loA
Use: Embedded for reference. Brief quotes used for commentary/review.