Keeping Up With Data #113
5 minutes for 5 hours’ worth of reading
The private equity market saw a slowdown in deal activity and fundraising in Q3 of 2022, with exits falling by 67% according to EY. This is due to depressed valuations and a widening disparity between buyer and seller pricing expectations. As leverage becomes increasingly expensive and portfolio multiples contract, operational value creation will drive the majority of private equity returns.
In this environment, data is the new leverage. However, many investors and management teams lack the capability to realise the data opportunity, making data domain experts and strategic data consultants essential for unlocking the value of data through due diligence, diagnostics, strategy and project delivery.
This recognition led to the formation of DataDiligence.
One of our value-seeking principles forming the bedrock of all that we do at DataDiligence is: Data has economic value.
Data having economic value is the theme of my reading list today. I read about the “fourth surge” of the data-driven economy, where we’ll see a significant shift in how data is transformed into true economic value worldwide. I also found valuable benchmarks for the economic benefits of AI implementation, which can help set realistic expectations. And what companies should keep in mind when making AI and ML an integral part of their products or services.
The relative benefits to individual companies may seem marginal (especially given the effort required to succeed and risks involved), but they add up on a global scale.
To truly unlock the economic potential of data, a business-first attitude and holistic approach is essential. Keep in mind that success with data and analytics is rarely just about data and analytics, but we’ll delve into that another time.
Today’s reading list looks at data-driven economy, building defensible machine learning companies, and the state of AI.
- The “Fourth Surge” of the Data-Driven Economy and How it Will Transform Our World: While the amount of data is growing every year, the impact it makes on the economy and the real world comes in surges, which Brett Hurt describes in the article. The first surge was triggered by banks starting to really leverage data in 1970s. The second one was around the SaaS and massive move of data into cloud. Third one was linked to cybersecurity and data protection. And the fourth one powered by a convergence of tools and approaches that the author calls the ‘Double Helix of Data’. Now, as a silver lining to the massive layoffs, the data talents is spreading from tech giants to more traditional sectors such as manufacturing and health care, which are about to get truly data-driven. And as the author concludes: “Sure, there’s a slowdown in Big Tech. But the piece of the digital economy iceberg we see above the waterline should not mask the huge mass below that’s been waiting for an appropriate infusion of talent in order to surface.” (data.world)
- Building a Defensible Machine Learning Company in the Age of Foundation Models: With the recent frenzy around ChatGPT, triggering executives and entrepreneurs in thinking how to adopt AI in their business, it is worth reading about complexities of building a defensible ML company. In the past, building a long-term defensible ML company was structured around two principles — the costs of ML and defensibility flywheel. The first one claims that the cost of ML should not ideally exceed the user utility. The defensibility flywheel starts with a value-driving product generating more data, improving models, resulting in superior UX, generating more value and accelerating the flywheel effect. Has this changed in the age of foundation models (think Dall-E 2, Stable Diffusion, or ChatGPT)? Not really. ML companies should still focus on building a “value-driving product and build up as much distribution as possible, with or without ML, initially. Companies, in general, should never be ML-first.” (Viet’s Monologues)
- The state of AI in 2022 — and a half decade in review: What is the state of AI in the corporates? The adoption has plateaued between 50 and 60 percent. The average number of AI capabilities embedded in products and operations sits around 4, with robotic process automation at the top. What I find useful is a benchmark on the cost savings and revenue increase through AI. Around 10% companies are seeing larger than 10% increase in revenue, and around 5% see decrease in cost larger than 20%. Useful to keep in mind when reading vendors’ marketing collateral claiming “2.3x increase in sales” or similar. But for large organisations, even 5% is a massive win, often delivering amazing ROI. (McKinsey)
Enjoy the weekend and remember that keeping up with data is easier than catching up.
