Louise Herring, a partner with QuantumBlack, AI by McKinsey, specializes in the intersection of AI, digital transformation, and business strategy. Her path from chemistry to business consultant to AI leadership highlights her love for solving tough problems and working with a wide range of people. For Louise, successful AI isn’t just about cutting-edge technology — it’s about the human side: teamwork, collaboration, and helping businesses adapt in smart and meaningful ways.

In this interview, Louise gives us a peek into the world of AI transformation, sharing insights from her wide-ranging experience across industries. From generative AI (GenAI) adoption to the future of AI in sectors like retail, her perspective is sure to resonate with business leaders and AI practitioners alike. Watch the video or read on for more.

0:00

/6:28

Louise Herring on how Enterprise AI is progressing.

Can a company compete without AI?

Today, maybe. Tomorrow, no. It's very clear that the position is rapidly changing. The potential and value of AI are so high that being too slow to adopt it will leave companies at a disadvantage. Whether it's in terms of their productivity, or more profoundly, their competitive position and the customer experience they provide — ultimately, no, they won't be able to compete without AI.

How do you use AI in your daily work life?

I use AI to solve the problems of AI. It's a big part of every client conversation. But in terms of my actual work and supporting clients, this has changed dramatically with the rise of GenAI. Particularly with tooling that the firm has invested in, such as Lilli, our bespoke GenAI-powered knowledge management tool. It means that we have much more information available, synthesized at our fingertips. It makes prepping for client conversations and learning new concepts across different sectors easier.

What is the hardest part of moving from POCs to production?

The challenge I see many companies encountering at the moment is problem definition. What are companies actually pointing their generative AI applications at? There have been a lot of hammers looking for nails — framing small experiments to learn, which has value. But the challenge is that even if every experiment was wildly successful, it won't deliver much value because it wasn’t aimed at solving a big problem. To drive impact, companies need to focus at a domain level — beyond just a use case here or there — reimagine what a part of the organization could look like, and systematically build towards that. A domain can be a functional area or, as a rule of thumb, we talk about 10 to 15% of the company in some aspects. For example, I’ve seen companies take a customer care point of view and reinvent that from end to end. The point is, it needs to be a meaningful shift.

How should companies track AI implementations as they scale?

Tracking success for AI implementations requires a multi-layered approach, as the metrics will change depending on the stage of maturity. First, companies need to be clear about the value they're trying to achieve — what does success look like from a business performance perspective? It’s not just about the technology working, it’s also about moving the right business metrics. As you scale, you also need to focus on enabling infrastructure, such as the talent, data, tech, and operating models that are crucial for successful AI deployment. This includes thinking about productivity gains and understanding how freed-up time is translated into business value. Companies should have different success indicators for each phase and ensure that there’s room for course correction along the way.

Who is responsible for the ROI of AI solutions?

If I look at the maturity of understanding around ROI, I think it's low on the whole, partly because it's relatively fragmented and also unclear as to who should be worrying about it.

On return, it comes back to: what are the metrics that need to move to drive business performance? That's the number one question that needs to be solved by the management team. What is it that they would look for in order to say, ‘“Yes, this is changing my business.”

Then, of course, you have the investment side of the equation, not just the return. How do you optimize for the problem? If you take the model, for example, does it really have to be incredibly fast, or actually, can you have a slightly lower performing, but much more economical approach? There's a lot of costs tied up on the risk side of the conversation that is underestimated in particular. How are you going to manage the risks? What are the people costs around that, and what does that mean for scaling? There's still quite a journey to go on overall.

If we look at the retail sector in particular, what are the biggest AI opportunities and challenges?

In retail, there are significant opportunities across the entire value chain — from product design to marketing to customer service. One exciting development I have seen is in the luxury space, where AI is transforming the discovery and inspiration phases of the customer journey.

0:00

/2:19

Louise Herring on the future of retail with AI.

However, the challenges are equally pronounced. Retailers face high stakes in managing customer relationships, and any missteps can damage trust. There’s also the challenge of integrating AI experimentation without disrupting day-to-day operations. Retail is a highly competitive space with thin margins, so balancing innovation with consistent trading performance is a constant tension.

Even with these challenges, the upside is likely worth it. For example, we partnered with Cohere to create the Virtual Shopping Assistant (VSA). Available to deploy in just 6 to 8 weeks, VSA interprets product queries and suggests relevant items to enhance shopping. The resulting impact metrics were impressive: 10 to 20% increase in conversion throughout the product discovery journey with a 98% accuracy in product recognition and attributes.

What are three pieces of advice you have for anyone working with AI technology?

The three pieces of advice I have are: focus on the problems that matter, make it a business transformation, and keep it simple. First, you will never see the value from any type of AI if you're not actually answering questions that make a difference to the organization. This requires a top-down vision aligned with business goals. Second, do this arm in arm with the business — AI has impact as a business transformation, not just a tech initiative. It needs sponsorship, ownership, and accountability at all levels, especially from P&L owners. Lastly, keep it simple. Avoid too much complexity in the tech stack or across projects. Focus on scaling with clarity and efficiency rather than creating fragmentation within your systems.

0:00

/2:37

Louise Herring on working with AI technology.

What are your hopes for GenAI in the near future?

I have hopes for this technology on two fronts. First, that it can demonstrate transformative benefits quickly — both in terms of customer experiences and how companies are run. There's huge potential for AI to help organizations ask great questions and make better decisions. Second, that it is deployed well, with ethics taken seriously. Risks need to be addressed early, ensuring that organizations can benefit from AI without succumbing to the challenges and potential pitfalls that come with it.

About Louise Herring

Louise Herring leads QuantumBlack, AI by McKinsey in the United Kingdom, Ireland, and Israel. She helps organizations scale impact from digital and analytics, helping them capture opportunities through new growth platforms, rethinking consumer engagement models, and improving organizational performance.

About McKinsey and Cohere

In 2023, Cohere announced a strategic collaboration with McKinsey, a global leader in strategy and technology consulting. McKinsey and Cohere aim to drive clients’ business performance through tailored end-to-end AI solutions. The collaboration is led by QuantumBlack, AI by McKinsey, the firm’s industry-leading AI arm.

Mastering private AI deployments

Is your business ready for a secure, private AI deployment?

Download the eBook

Braving AI: A conversation with McKinsey’s Louise Herring