In Focus

Digital Transformation

Fancy a trip to the metaverse? Already feels a little 2022, doesn’t it? How about a Web3 app, or a ChatGPT personal assistant?

Fads are a constant theme in business. Just as the consumer world gets hooked on tulip bulbs and yo-yos, hacky sacks and Crocs, corporate execs can be suckers for a popular mania. A buzzword gains momentum. Bored journalists ramp up the idea. Pretty soon the concept gains irresistible momentum, prompting boards to fret that they’re being left behind. 

Charlie Munger, Warren Buffett’s investment partner, is a keen observer of the phenomenon. He once noted: “Some years ago, one oil company bought a fertiliser company – and every other major oil company practically ran out and bought a fertiliser company. And there was no more damned reason for all these oil companies to buy fertiliser companies, but they didn't know exactly what to do. If Exxon was doing it, it was good enough for Mobil and vice versa.” 

The result? “It was a total disaster.”

Another case of Fomo?

Much the same thing is happening today in the digital transformation space. The likes of machine learning, Kubernetes and Web3 are generating white-hot interest. According to KPMG’s latest CEO Outlook survey, 72% of CEOs have an “aggressive digital investment strategy, intended to secure first-mover or fast-follower status”. 

But the logic is often nothing more than the fear of missing out. “Fomo is totally a thing at an executive level,” says Tom Grogan, founder and CEO of emerging technology strategy consultancy MDRxTech. He rattles off examples of poorly thought-through fad-chasing. 

“Like so many others, Porsche got excited and launched an NFT project, which went live at the start of the year. It took them so long to launch that they missed the NFT craze by about 18 months. The decision has been widely panned. The NFTs themselves didn’t even sell well.”

The dafter the fad, the worse the strategy, Grogan observes. “A prestigious Italian fashion house spent tens of millions acquiring land in the ‘Oxford Street shopping experience’ of a metaverse project. In return, they ended up with an Oxford Street shopping experience resembling a post-apocalyptic wasteland. There was a handful of visitors at any one time.”

Is it wise to be first?

At the heart of these errors is a belief that cutting-edge tech must be superior. New stuff is fun to play with and there is a social status in being the first. There’s also a sense that innovations will somehow future-proof the enterprise, even if the precise mechanisms are not well understood. 

In practice, though, there’s a trade-off between novelty and reliability. New tech will be glitchy, lack a supportive ecosystem and offer no record of durability. There’s a reason cash machines still run on Cobol, a programming language invented in 1959. The code is primitive, but proven.

Will Lion, chief joint strategy officer at BBH London, advocates caution when approaching anything new and shiny.

“There’s that inspirational story about the fact that you’re the sperm that beat millions of others to get to the egg first. Well done, you!” he says. “The thing is, that’s rubbish. Any biologist will tell you that loads of other sperm cells died getting through to the egg, desperately trying to be first, only to clear the path for you. There’s a lesson here for digital transformations. We have a fetish for the new. It gets clicks, attention and clients to reply to emails. But new is awful. It’s unproven. It’s complicated. It’s risky.”

Lion contends that the smarter strategy is to lag behind the early adopters. Let them bear the costs. And then capitalise on the lessons learnt at their expense.

“I’m here to argue that ‘best’ or just ‘working effortlessly’ beats being first by a long shot,” he says. “Apple has made a whole business out of it. So here’s to being the pioneers behind the pioneers, saving yourself a fortune and turning all their mistakes into gold.”

AI: artificial ignorance?

The explosion of interest in AI is perhaps the biggest fad in digital transformation right now. For all the undoubted promise of the new breed of AI tools, the hype is leading to misapplication.

For instance, judges in the US are routinely using AI to calculate sentence lengths. An algorithm computes the likelihood of reoffending and makes recommendations to the court. One of the most common algorithms for this purpose, known as Compas, has been criticised for being prone to racial bias. What’s more, it’s not clear whether judges understand the complex mechanisms at work. Campaign groups such as the American Civil Liberties Union are lobbying the authorities to limit the use of AI advice until judges are better able to understand the underlying tech. 

Likewise, lots of R&D teams are using ChatGPT to conduct ‘research’, forgetting that the AI engine merely guesses how prompts ought to be answered. ChatGPT is a brilliant inventor – not a researcher – hence the creation of ‘hallucitations’, fake references inserted in convincing pseudo-academic bot prose. 

And in some sectors, such as recruitment, the hasty application of AI is a potential disaster. So says Richard Collins, co-founder of CV Wallet, a CV hosting platform.

“The rapid adoption and use of AI by employers in the hiring process is badly thought out, resulting in bias, inequality and an invasion of privacy,” he explains. “We see companies using AI referencing at the end of the process. These background checks look more like stalking, as they allow employers to cross reasonable boundaries with a significant invasion of privacy. The AI-powered tools collect vast amounts of data on applicants without their consent, often without considering the relevance, accuracy or currency of the information.”

Indeed, our obsession with the latest tech is compounding historic errors, Collins believes.

“It has made an already broken system worse. It’s resulted in indirect discrimination, which has prompted governments around the world to regulate the use of AI in hiring. New York City, for example, is banning businesses from screening job applicants with AI-based tools from 5 July.”

How to dodge disaster

Is there a solution? Charles MacKay chronicled mass hysteria in his 1841 work Extraordinary Popular Delusions and the Madness of Crowds. He wrote: “Men, it has been well said, think in herds. It will be seen that they go mad in herds, while they only recover their senses slowly, one by one.”

His advice was to be wary of popular obsessions and stay aloof. Observe, but don’t get sucked in. And, if seduced by a fad, be the first to recognise the symptoms.

But a digital transformation is a balancing act. The objective is to leap from legacy tech to something more modern. The challenge is to discern the line between what’s a genuinely useful innovation and what’s a trick of the light. Web3, the metaverse or whatever other new tech starts swirling around the hype cycle may capture the imagination of the masses, but this is no guarantee of improved performance.

Instead, we need to recognise that our judgement is warped by cognitive biases and that peer pressure can seduce us all. As Munger himself put it: “Time and time again, in reality, psychological notions and economic notions interplay – and the man who doesn’t understand both is a damned fool.”

The explosion of transformative technology in recent years is hard to ignore. Generative artificial intelligence (AI) is the latest craze, of course, but the metaverse, hyper-personalisation and cloud technologies were making headlines long before that. 

While some businesses might be inclined to exercise caution here, rapid deployment and expenditure – $3.4tn (£2.7tn) across all digital transformation technologies by 2026 – will ultimately pile on the pressure, requiring the more hesitant players to dive in as these new technologies increasingly give early adopters a competitive edge.

The problem is that digital transformations have a habit of raising difficult ethical questions. This may yet be addressed through regulation – data privacy issues, for instance, have been well served by the General Data Protection Regulation and the yardstick it provides for non-EU jurisdictions – but law-makers have generally been slow to catch up with the latest innovations. And key trends and their harms, such as employee surveillance and algorithmic biases, aren’t well covered. 

The same goes for the much-discussed problem of redundancies owing to AI adoption and automation. This has already come to fruition in certain industries and more layoffs seem likely as the technology progresses and more of us are replaced by machines.

How, then, can firms address the ethical issues in their digital transformations? Should they be looking to regulators to provide a baseline, or is it wiser to proactively embed accountability and standards internally? And what might that entail?

Why more tech means more problems

“Nobody wants to do business with a racist or homophobic company, but AI can sometimes raise issues associated with that,” says Adnan Masood, chief AI architect at digital transformation solutions provider UST. Indeed, the reputational cost to businesses when algorithmic technology causes harm can be significant.

Such cases have been well documented. AI chatbots are already notorious for outright racism and sexism. The Cambridge Analytica scandal – where the company collected millions of Facebook users’ data for political advertising – landed the social media giant with a fine of almost $650,000 in the UK for its failure to protect users from data misuse. On the redundancies front, the move to a mostly digital banking platform prompted TSB to cut 900 jobs and close 164 branches in 2021. 

Increasing algorithmic management, typically associated with the gig economy, is also becoming more common in other sectors, from optimising delivery and logistics to tracking workers and automating schedules in the retail and service industries.

Meanwhile, consumer-sourced rating systems are increasingly being used to evaluate workers. This can create a culture in which pernicious problems such as sexual harassment take root, as workers stay silent to avoid a bad rating. 

These tech-fuelled quandaries demonstrate why up-to-date regulation is important. So far, there has been relatively little movement on this beyond data privacy, although there are indications that AI regulations may soon be in the works.

The only way is ethics

The regulatory outlook is complicated further by the fact that digital transformation is a moving target. Even the existing data privacy guidelines may fall short over time because we’re often unable to foresee where prescriptive detail will be required. 

Take the right to explainability of how an algorithm comes to a decision: “That’s virtually impossible when it comes to black-box models such as neural networks,” Masood says. “You cannot explain how a neural network works.”

The good news is that, as problems become more apparent, the conversation about digital ethics is getting louder. Companies are increasingly weighing up how to manage these ethical dilemmas themselves, including by self-regulating and even prioritising ethics above sales. For example, after the killing of George Floyd and the Black Lives Matter protests in 2020, IBM declared that it would no longer provide facial recognition products to police forces for the purposes of mass surveillance and racial profiling.

Some observers are even questioning whether regulators are capable of stepping up to address AI, or whether it should be left to the tech players. One of them is Oyinkansola Adebayo, founder and CEO of Niyo, a group of brands focused on the economic empowerment of Black women through tech-driven products. 

“A lot of the regulators are not doing the job – they’re not part of that field of AI models and they tend to have a fear-based mentality,” she argues. “Regulation is stifling innovation. We need a collaborative approach with the people building it, to challenge the build as it happens rather than at the borders.”

Why humans are still central

One way for businesses to start straightening out ethical issues in their digital offering is to ensure they aren’t perpetuating a skewed view of the world. 

“Less than 2% of the tech industry is made up of Black women specifically,” says Adebayo, who contends that addressing the gender and racial imbalance in tech workforces would result in a greater diversity of thought. This, she believes, should help to ensure that fewer ethical problems slip through the net.

Rehan Haque, founder and CEO of Metatalent.ai, also stresses the importance of human capital in any digital transformation. When he built his company, he focused on upskilling, reskilling, cross-skilling and redeployment to equip people to handle emerging technologies.

“Humans were the most important thing from an investor’s point of view. And then it was technology,” Haque recalls.

That’s all well and good, but will it be enough to assuage customers’ worries? To help firms keep pace, could AI-based tech be put to work to help on the ethical front? 

It’s a prospect offered by a set of principles the EU has been working on for more ethical approaches to AI, also known as ‘ethics by design’. Similar to the concept of privacy by design, companies are encouraged to build respect for human agency, fairness, individual, social and environmental wellbeing and transparency into their AI models, as well as the familiar principles of privacy, data protection and data governance.

But trusting technology to solve technological problems could lead to a whole other set of concerns, notes Professor Keiichi Nakata, director of AI at Henley Business School’s World of Work Institute.

“One of the ways to identify whether certain work has been done by AI is to use AI to check,” he says. “Of course, it’s a cat-and-mouse game, because both sides will improve and become more evasive.”

So it seems that regulators are too slow to keep up with emerging technologies, while the tech itself can’t solve all the ethical problems that are being created. But we cannot afford to ignore the harms threatened by widespread and unchecked digital transformation either.

Guidelines for ethical design could be helpful, but it will take time for the tech industry to adopt and implement these. In the meantime, it’s up to businesses to engage the expertise of a broad range of players to manage ethical risks as their digital transformations take shape. 

In one of the largest cloud migrations ever seen, consumer goods giant Unilever recently announced that it has gone all-in, having shifted all of its 400-plus household brands to Microsoft’s Azure cloud platform.

According to the company, the move should help to accelerate product launches, enhance customer service and improve operational efficiency. It will also help its sustainability efforts, by curbing carbon emissions from its own tech stack.

“There are very few companies of our size and legacy that can claim this level of cloud implementation,” says Steve McCrystal, Unilever’s chief enterprise and technology officer. “We have confidence that we’re set on the right course and that this will deliver a step change in speed and flexibility.”

And Unilever is not alone. As more businesses strive to modernise, large-scale cloud migrations are becoming more common. But few companies ever reach Unilever’s scale and even fewer plump for a full overhaul of their data storage strategy. 

That all makes this particular cloud migration project remarkable, according to Dean Clark, CTO at global digital transformation provider GFT.

“While firms in other industries have accomplished similar cloud migrations – such as Capital One’s migration of 8,000 apps, and a similarly large ongoing Google Cloud undertaking by Deutsche Bank – Unilever’s project stands out as especially impressive in the consumer goods industry,” he says.

What can other firms weighing up their own digital transformations learn from Unilever’s experience of doing it on such a huge scale?

The case for going all-in on the cloud

One key thing to remember here is that the Unilever project began back in 2019. The latest announcement represents just the most recent phase of a lengthy digital transformation. 

Even so, the scale and speed of the move still make it stand out from the crowd. After all, with its infrastructure supporting more than 400 consumer brands and being used by more than 3 billion people daily, this was no small undertaking.

Support for the migration came from Accenture. Nicole van Det, senior managing director and Unilever’s global account lead at Accenture, believes that going all-in on cloud technology is a crucial requirement for consumer goods companies.

“Whether it’s gaining access to flexible and scalable infrastructure, accelerating innovation across the business and with partners, curbing carbon emissions or transforming the use of customer analytics and artificial intelligence, cloud is a must-have for any company serious about digital transformation,” she says.

Watch the cost of cloud

But Unilever’s wholesale move to the cloud comes at a time when other firms are arguing in favour of saving money by shifting their IT back on premises, or at least maintaining a hybrid IT model. In October, for instance, David Heinemeier Hansson, CTO of Basecamp, announced in a blog post that the firm would be turning its back on the cloud, thanks in part to a bill reaching $3.2m (£2.6m) a year.

Nonetheless, McCrystal says that Unilever is unwavering in its commitment to the cloud, given the significant cost savings it has enabled. These are most apparent across Unilever’s network and support structures. It has been selling data centres it no longer needs, for instance.

“The cloud allows us to pay for server capacity only when we need it,” he says. “We have been able to put a lot of servers into standby mode.”

Geoff Barnett, head of global multi-cloud advisory at IT consultancy Crayon, says that some firms may well look to shift their large-scale IT deployments back on their premises. But he adds that they will generally be companies that have built their applications and services in a way that uses simple building blocks that are similar across all their software.

“Adopting an on-premises environment can make sense in this case, because the business can standardise around a single consistent platform that can be maintained with a high degree of automation and a small number of people,” Barnett explains.

“On the other hand, when a business is buying in a lot of commercial off-the-shelf software that has different dependencies in terms of platforms and other components (or when a business has developed software using different platforms, frameworks and architectures), the cost of supporting this variety of needs can often be more easily met when the management of some of these platforms can be handled by the cloud provider.”

Barnett adds that organisations are coming under significant pressure to do more with less, owing to the tough macroeconomic environment.

“IT cost optimisation has never been a higher priority, with 90% of businesses recently labelling it as such,” he notes. “They are reassessing where and how they spend their money – and that means the efficiencies and economic benefits of cloud technologies are becoming increasingly appealing to them. Globally, 56% of businesses are spending more than $1m a year on the public cloud alone.”

Sell your cloud vision to the whole team

Of course, any cloud migration – especially one for an organisation that has as much data as Unilever – will run into problems. Employees may lack all the skills required to support the design and migration of key processes, Clark suggests. Selecting the right partner can help here, as it may offer tailored training to help people develop vital cloud skills.

Similarly, businesses may find themselves needing to reprogramme staff to operate in a new way, according to Barnett. This is particularly true within the parts of the organisation that work directly on supporting or developing applications and IT services.

McCrystal maintains that “aligning IT and the business around the same vision and objectives provided coherence and clarity throughout our cloud journey”. He adds that, by working with a mix of internal teams, Accenture and Microsoft, Unilever has “created an environment to bring out the best of everyone working on the project”. 

“This meant being mindful of different cultures, time zones, ways of working and technologies. This, combined with having a single goal across the whole of Unilever enabled incident-free, on-time and in-budget delivery, landing industry-leading success.”

What could be easier, right?