David Floyd, vice-president at global investment firm Permira, talks to Real Deals about how the business is harnessing advanced analytics to drive competitive advantage in its funds’ portfolio companies, and outlines how Permira is using AI tools internally to support investment teams.

RD: How is Permira using data and analytics to support value creation in the portfolio?

David Floyd: Permira established an advanced analytics team within the value creation team about five years ago, with its primary function being to identify ways to use technology to help management teams to drive value creation at the portfolio company level.

We strive to find the most impactful use cases for data and AI in portfolio companies, show companies what they can do with their data and then demonstrate proof of concept to the management team. We will get involved in some forecasting if there’s a clear bottom-line opportunity, but top-line growth is the main priority.

We do not mandate that all portfolio companies use the same software or follow the same implementation blueprint. The aim is to help each company find the right tool and right process for the right situation. There are, however, recurring themes that we have observed and we regularly see advanced analytics utilised to drive improvements in pricing, customer churn and personalisation. 

We also have a significant focus on knowledge sharing. If you take the example of AI, we have built a community of more than 260 senior stakeholders across our portfolio who can collaborate through various forums, roundtables and a new digital engagement platform.

This is proving a very fertile community and helps our portfolio companies stay up to date on the latest – and rapidly evolving – regulatory environment relating to AI. It also allows these companies to take part in topical training sessions, deepen partnerships with vendors and develop collaborative exercises between portfolio companies. 

At the same time, the portfolio companies have been continually experimenting and sharing knowledge about various GenAI tools – such as Enterprise ChatGPT, GitHub Copilot, AWS CodeWhisperer – to find ways to improve productivity. 

RD: How are you approaching the topic of AI more specifically in the portfolio?

Floyd: Broadly speaking, our strategy is to work with and encourage our portfolio to experiment with GenAI applications and, as part of that, leverage the portfolio as a distributed laboratory to rapidly prototype solutions and then share the learnings within the portfolio. As part of this, we are encouraging all companies to ring-fence teams to ensure a focused approach.

There is currently an average of three GenAI development projects currently live in each portfolio company – more than 200 in total – with new initiatives being identified continually, either by dedicated R&D teams from within business functions or through company ‘hackathons’. In a tech portfolio, this average figure can be as high as 10. 

RD: With regards to your work on the firm’s internal technology capability, what is the focus and what are you hoping to achieve?

Floyd: During the last 12 to 18 months, one of the major areas of focus has been to work closely with our CTO to look at how Permira can drive performance advantage for the firm and the Permira funds from generative and predictive AI. We have explored everything from deal origination, due diligence and value creation through to portfolio management, financial reporting and ESG compliance.

We have built an internal platform for running AI tools securely across all of our internal data. The platform delivers similar output to what you would expect from a tool such as Microsoft Copilot, but we decided to do the build in-house so that we could learn about the technology as fast as possible, deploy it quickly and not be tied into any one vendor. Developing the platform internally has also helped us to deal with the information security challenges that come with deploying this kind of technology, which are significant for any private equity manager. 

The platform is called Gaia; it is live and working. The initiative started as an experiment to see if we could use Generative AI effectively, and we are now in the production phase. We have seen teams across the whole organisation use it. 

It is still early days but initial indications suggest that every individual using the platform saves between 5% and 11% of their time every week. The exploration and testing of ways to use the technology to automate more simple tasks, as well as handle autonomous workflows, is an ongoing process.

We have also developed a second tool in-house – Prism – which produces a visual and predictive readout of a company’s customer churn ‘health’, as well as its upselling and downselling performance. Pulling together that information supports a much more informed management conversation.

In addition to the platforms that we have developed ourselves, we have also explored a few complementary third-party technology options. Hebbia, Glean, and Dataiku are the tools we are most familiar with – Hebbia and Glean serve as AI-powered enterprise search tools, similar to Gaia and M365 CoPilot, while Dataiku operates as a broader platform for deploying AI/ML models in automated workflows.

It should also be said that as part of all of this work, we have worked closely with our legal department to ensure that all AI tools operate mindful of relevant compliance frameworks and procedures.

So, if I am on a deal team and I log in to one of the tools, how can they support me through the deal process? 

Tools like Gaia assist dealmakers with finding and synthesising relevant documents at speed. A piece of analysis that would have been a huge undertaking because of the time it would take to assemble the relevant documentation can now be done in a couple of hours. It might also help to compile a quantitative picture of a company or opportunity out of the mass of available commentary and qualitative data. 

These are common applications for generative AI. The AI finds the needle in the haystack and uses natural language processing-style analytics to pull out something structured from something unstructured.

Tools such as Prism provide deep insights into a company’s revenue flows and areas of risk or opportunity early in the diligence process, to support evaluation and to provide value to management teams.

We also have a proof of concept for using AI to not only identify top-performing companies in particular sectors, but also to recognise characteristics that may not be immediately obvious but can predict growth and outperformance. 

RD: Are there any particular highlights or success stories from your team’s work that stand out?

Floyd: Personalisation has been a powerful lever. Customers increasingly expect a personalised service from brands; this is especially true at the luxury end of the market. More than half the growth (and high tens of millions of euros per year in direct-to-consumer revenue) at one of our luxury goods portfolio companies, for example, has been due to the personalisation programme that we put in place.

Improving customer retention in the portfolio has been another highlight. Increasing retention by 5% can lead to a company’s profits growing by anywhere between 25% and 95%. It has been interesting to observe how translatable this model is, and the success of the model is what led to the development of Prism.

Finally, pricing is an interesting area and can yield exceptional results when done well. One of our portfolio companies saw gains in the tens of millions of euros after using technology and data analytics to analyse how it was using promotions and signals at the best times to market promotions to customers.

RD: What are some of the biggest challenges you have encountered when it comes to the implementation of new technology in a company, and how can these hurdles be navigated?

Floyd: The challenge that comes up time and again is the siloing of data. What we are trying to do is to find signals in data but what we often find, particularly for early-stage companies, is that managers are focused on their own KPIs and are not really interested in what other parts of the business are doing.

Therefore, when you are trying to leverage data analytics and AI, implementing cultural change is central to the success of the project. Often you will want to change the culture of the company completely, but that is not realistic. The key is to identify the culture that exists in the company and adapt, making subtle changes to make data and AI work for the business.