The secret recipe for getting real value out of Gen AI

Are you finding Gen AI overhyped and not delivering the results you anticipated? Do you feel pressured to show value from Gen AI experiments but frustrated by slow progress?  

You’re not alone – achieving meaningful value from Gen AI is a significant challenge faced by many organisations today. 

Through partnerships with organisations from various industries across South Africa (banking, insurance, telcos, and more), we’ve identified five key insights: 

1. Most organisations are bullish about the potential of Gen AI  
Over the last six months, BSG has held two Gen AI roundtables with clients. Across these sessions, there was clear enthusiasm for Gen AI’s impact on the bottom line, despite some unique risks. Organisations believe that Gen AI can and will drive meaningful business value. 

(Source: BSG Gen AI roundtable)

2. New Risks Emerging with Experimentation 
Most organisations lack experience in deploying Gen AI, which has led many to adopt an experimental approach. This experimentation exposes gaps in skills, partnerships, and understanding of the application of Gen AI technology. Many organisations end up with numerous use cases but limited success due to challenges in integrating Gen AI with existing systems, capabilities, and data sources. 

3. Difficulty Launching Meaningful Solutions 
Organisations struggle to move beyond traditional “system” development lifecycles, often mixing agile and waterfall methods in Gen AI projects with their standardised “systems” adoption and change management practices. These projects require ongoing adaptation and input, not static implementations. Without this mindset, companies risk rigid, low-value Gen AI solutions that don’t meet expectations. 

4. Lack of Business Impact Without Clear Metrics 
Many Gen AI experiments don't have a direct impact on the bottom line; it is hard to quantify that without solid business leadership. (Ref: McKinsey, IBM and others) 

One of the hardest disciplines to achieve is to be ruthless about what you are experimenting with in terms of its potential value contribution. Being able to articulate the impact and the metric you are using to define the value contribution is one aspect and being able to use that metric as a primary selection criterion for your business, is the other secret. This is still very difficult to get right, and mostly because of indirect involvement by the owners of the value metric (generally referred to as “the business”). 

5. Persistent Data Challenges 
Organisations are discovering that the old "data problem" haunts them even more with Gen AI. (Ref: BSG experiences in banking, insurance, and health industries, Gartner) 

If you’re an IT or business veteran, you know that data best practices – such as democratisation, ownership, master data models, and single-source customer records – are difficult to achieve. Many Chief Data Officers have left their companies due to the political and technical complexities around data management. With businesses running data operations 24/7 across multiple geographies, these challenges often result in a fragmented data strategy. Implementing Gen AI doesn’t eliminate these issues; in fact, it can compound them by introducing unstructured and third-party data sources. Beyond selecting the right use case and expertise, having access to the right data remains critical to unlocking Gen AI's value. 

As Gartner notes, "CIOs will begin to spend on GenAI, beyond proof-of-concept work, starting in 2025. More money will be spent, but the expectations that CIOs have for the capabilities of GenAI will drop. The reality of what can be accomplished with current GenAI models, and the state of CIO’s data will not meet today’s lofty expectations."  

The secret recipe: An integrated 9-point Gen AI success stack

There are for sure more issues that affect an organisation’s ability to unlock value from Gen AI, shaped by context and maturity. And dealing with the above is not simple, as it in most cases requires an integrated approach that gets you the success you are looking for.  

BSG has understood that right from the start and have over time constructed an integrated “9-point Gen AI success stack” that we believe will get you the results you are seeking. The figure below, is a summary of the stack. 

(Ref: Gen AI success stack based on Klarna and various other sources and BSG experiences) 

BSG's 9-point Gen AI success stack

1. Gain a deep understanding of Gen AI: Know the technology’s capabilities fully before diving into solutions.

2. Explore broad applications: Look beyond current products to see how Gen AI could empower your organisation to disrupt and differentiate.

3. Encourage experimentation with clear vision: Encourage Gen AI experimentation, starting in the line of business areas. Make sure you have an AI vision to help drive the experimentation culture and give it enough senior weight to make it an identifiable “moment” in your organisation.

4. Be brave - empower top talent: Incentivise innovative thinking, especially where it challenges existing constraints.

5. Focus on quality, not quantity: Select fewer, high-value use cases rather than spreading resources thinly across too many. Be very intentional about the value that you are seeking. Let the value impact guide your selection criteria.

6. Enable the right tools and data: Don’t allow the lack of IT capability to block your experimentation, prepare platforms for experimentation, even if it means using external infrastructure (radical and controversial, we agree, but true).

7. Don’t underestimate the value of telling your Gen AI story: Use marketing to build buy-in and highlight benefits, making the initiative visible (not a secret dark room exercise).

8. Leverage strategic partners: Make sure you have partners helping you. This journey is one about learning, capability building and discovery. Make use of consulting, technology, platform, and solution provider partners that understand you, have a trusted track record and can deliver a successful outcome.

9. Implement responsible AI: And lastly, but for sure not the least, there are approximately 12 lenses that you need to view your solution through to achieve a “responsible AI solution” status. Aspects like transparency, bias checking, privacy and many more must form the basis for your solution. Look out for our next article on this topic.

At BSG, we strive to be a proactive force for meaningful change. Our 9-Point Gen AI Success Stack is designed to help you overcome barriers and achieve substantial value with Gen AI.