E-commerce jumps a decade in two months

Has ‘the new normal’ become the most overused phrase in the English language thanks to Covid-19? It has competition: ‘doomscrolling’, ‘you’re on mute’ and, more recently, ‘work event’ should all be in with a shout, but our first nominee has that little bit extra.

Those other three can all actually refer to something: respectively, social media, teleconferences and cake-led ambushes. ‘New normal’ is sometimes so completely void of meaning yet at others so packed as to be beyond comprehension.

This is not just linguistic snark. This problem of definition is already having an economic and technological impact, particularly when it comes to advances in and the adoption of techniques around data science and artificial intelligence.

Determining what exactly the new normal means is increasingly likely to require the use of data – and AI-based techniques. And the task may be so broad that it also demands more collaborative models for research and analysis.

On an individual level, companies are responding, and with good reason. A survey by consultants at PwC found that just over 50 per cent of companies had accelerated their internal AI adoption programmes because of the pandemic, with separate research by Harris Poll/Appen getting almost identical results.

This can be put down in part to the fact that AI work is overwhelmingly digital and lends itself well to a work-from-home model. Even before the pandemic, many AI researchers were collaborating virtually with colleagues not just in their own countries but scattered across the globe.

However, businesses are also facing newer pandemic-based challenges that they see data helping them to address. As lockdowns forced bricks-and-mortar sales channels to close, many had to move more of their sales operations online.

Scott Galloway, professor of marketing at New York University and one of the keenest observers of the evolving economy, notes that immediately before the pandemic reached its global stage in spring 2020, digital channels accounted for 16 per cent of US retail sales. Between March and April of that year that share rose to 27 per cent: “We registered a decade of e-commerce growth in eight weeks,” Galloway writes in ‘Post Corona: From Crisis To Opportunity’.

That share has shown little sign of falling, even as High Streets have reopened. These direct (or Amazon/Shopify-hosted) channels now need to be much more closely observed, as does the trend’s potential impact on town and city centres. And there are other rising urban challenges.

Offices are also an example of how the pandemic is stimulating long-term change, as companies move to models where while staff do not work entirely from home, they at least do so for more of the time.

In April 2021, professional services giant Deloitte announced that it would be moving out of one of its main London offices, Hill House in Holborn. The 185,000 square feet (17,200m2) being dropped there brought Deloitte’s total reduction of its real estate in the UK capital to 250,000 square feet (23,200m2) – roughly a third of its pre Covid-19 footprint.

“We’re reviewing our office space requirements to reflect changes to our ways of working and our sustainability objectives,” the company said.

Already by spring 2021, PwC was able to say that major UK employers plan to cut 9 million square feet of office space overall – the equivalent of 37 skyscrapers – even as Covid-19 becomes a more manageable endemic problem. A third of the 258 companies PwC surveyed said they also planned cuts of around 30 per cent.

“The figures couldn’t be more clear; the shift to hybrid working, with part of your time at home and part in the office, is pretty much embedded into the working culture of many organisations,” says Angus Johnson, PwC’s real estate leader for the UK and EMEA.

Data again is proving important. ‘Hot-desking’ is complex. Leaseholders must balance available space and the need for certain staff to be on site at certain times. An irony here is that one area criticised during the infamous 2019 IPO failure of shared-office company WeWork was its claim to be a technology company because of the information it was gathering about how its buildings were used. At least there, founder Adam Neumann might have been on the right track.

What then about the coffee shops and other retailers that congregate around business centres? Those are just two examples and there are many more, but a third is now most dominant in the thinking of many executives – supply chains. E&T has already looked at this in detail but there are fears of a renewed domino effect because of current events in China, specifically Shanghai, at the time of writing.

With most of the city in lockdown because of China’s zero-Covid policy, the world’s largest container port had, as of late April, about 500 ships waiting to berth. Experts reckon that the currently one-month-long measures will already suffer a multiplier effect that will cause delays in global supply chains for potentially a year.

Companies’ rapid in-house adoption of AI and other data-led innovations is now being fuelled by these factors on top of more traditional business objectives associated with them pre-Covid (such as performance increases in efficiency, sales and net-zero goals).

As Galloway observes: “The pandemic’s most enduring impact will be as an accelerant.”

As the outbreak has placed much broader factors on the agenda, they have highlighted the need for greater collaboration and the way radical shifts in practice internally feed other impacts externally.

Collaboration has been a success story during the pandemic, but it has also exposed weaknesses in how these processes currently work around AI and machine learning. Covid-19 has been the most aggressive stress test of the digital age.

For epidemiology and outbreak tracking, academic institutions have joined with medical research bodies and governments to provide and benefit from modelling and near real-time reporting. These dashboards have helped to inform interventions. Groups such as Oxford-based Our World in Data have made their datasets freely available. Analysts, notably in the UK John Burn-Murdoch of the Financial Times, have used this open-source information to help inform the public with easily understood visualisations.

There have been flaws. There have been concerns that data has not been passed down quickly enough or in enough detail to the local level to guide any necessary measures. In October 2020, this lay at the heart of a stand-off between Prime Minister Boris Johnson and Andy Burnham, the Mayor of Greater Manchester, over the measures central government planned to impose on the region.

There have also been instances where the media and the public, still coming to grips with what data science involves, have misread, misinterpreted, or even mischaracterised information. On 7 March 2021, leading media outlets simply failed to note a caveat in the latest data from the UK Health Security Agency that showed no deaths because data processing had missed the deadline.

A third factor – and one common to any such project that seeks to have a global reach – has been data variability. Not all countries report their data in the same format, and definitions vary. On that last point, the most frequently cited examples concern how different countries define causes of death (was a victim suffering a separate condition that was already a morbidity when tested positive?) and distinguish between symptomatic and asymptomatic cases (with assumptions of significant differences between Europe and Asia, notably China).

Outbreak data has made ‘teachable moments’ of granularity, communication and data consistency. How do you pull all this together so that you can feed in good information and get valid insights?
Meanwhile, the private sector has also set up several collaborative initiatives that have looked at the economic implications of the pandemic. These too have done valuable work but at the same time shown some of the other limits that these exercises still encounter, while facing many of the same challenges as their counterparts in public health.

The Emergent Alliance was one such project. Led by Rolls-Royce through its R2 Data Labs digital research arm, it brought together 54 organisations including companies such as IBM, Google Cloud, Microsoft, Bosch, Mott MacDonald and Reed as well as bodies like the Greater London Authority, the National Physical Laboratory and the University of Coventry.

Its aim was to support “a better-informed, and thus accelerated, economic recovery to benefit businesses and individuals”, according to R2 Data Labs group director Caroline Gorski.

“The Emergent Alliance wouldn’t have existed without the Covid-19 pandemic, but it was a galvanising crisis that enabled a group of disparate organisations to come together and collaborate, overcoming competitive and anti-sharing behaviour,” she says.

Between April 2020 and May 2021, it developed a set of applications around travel, local economic impact modelling and reskilling (see box below). The project was taken from concept to its active development phase in just one month.

It became a finite exercise, or at least seen as such by many of those members once its set of what were considered most deliverable and pressing tools had been developed. A further limitation was that volunteers worked on the alliance’s projects in their spare time or on furlough.

“As the roll-out of the Covid-19 vaccines commenced, people began to see signs of recovery. It became inevitable that volunteers and member organisations would prioritise resources to the direct needs of their own businesses,” the alliance’s wrap-up report says.

As a result, members chose not to fund the project further, though its work has been posted on GitHub for others to use for current and future challenges. As other economic trends have moved to the fore since its closure, an argument exists for another similar venture that could adapt to a shifting pandemic-driven agenda.

The Alliance’s legacy does also include what it learned in the process of delivering data-science-based benefits for general use and through collaboration pulling on a very broad set of competences, with companies that were sometimes competitors sharing the outcomes.

Again, it encountered issues with education and understanding, specifically in its relationship with the UK’s Cabinet Office, which was leading the national Covid-19 effort. “While trusted arbiters of insight were informing decision-making through epidemiology models, the Emergent Alliance presented machine-learning outputs,” its report notes. “While reports were viewed and interest taken in the results, the techniques weren’t recognised as those which might inform government policy.

“It proved an important lesson in knowing your audience, understanding at the outset who is going to evaluate your work and by what standards and expectations.”

A similar point is made more strongly in separate research published by the Turing Institute after a series of workshops on ‘Data Science and AI in the age of Covid’ regarding how various submissions were received.

“Many workshop participants observed that the lack of transparency in policy decision-making made it difficult to know which research studies  had ‘cut through’ or even been considered by government and expert advisory groups when deciding on policy interventions,” it says.

“Policy makers were also not transparent about which data were important: it was known that Google and Apple data were central to policy-making, but the detail and quality of those data were unknown, while other data science was not independently scrutinised.”

More positively though, the alliance does foreground the need for collaboration to get accurate results when tackling issues that pull upon wider activity than that within or directly attached to a particular company.

“To produce datasets rich enough to formulate accurate predictions, the Emergent Alliance needed a diverse, collaborative alliance where analytical skills could be pooled and shared to everyone’s benefit,” it concludes. And toward this goal the initiative secured high-level support from the Bank of England. It sees a need for more research of this type.

At the same time, the alliance did put experts in charge of projects. Reed, as the UK’s largest online employment agency, led its reskilling project. For that on air transport it turned Satavia, a specialist in aviation impact analysis.There are seeds in the alliance’s work.

What is this ‘new normal’? As Covid-19 does hopefully become less pervasive, the risk of future pandemics remains because we live in an increasingly globalised world. The process of bringing together established public health practices and new technologies is maturing. What that process has exposed are issues around data and how new processes are understood and communicated.

Those same challenges will now apply continuously because economic and technological change has, as New York University’s Galloway says, been “dosed with an accelerant”. Trends are gathering pace that were not immediately seen as likely to persist when the pandemic began. Risks that need greater mitigation have also been highlighted.

These changes feed into so much of how we live that it is unlikely that they can be managed and interpreted by a single organisation, not even one with as deep and wide a set of insights into society as a Google, and as much technological muscle.

As with public health, we need to turn to high-level collaborations. Leveraging the many data sources and sources of expertise that will help us interpret these trends looks unavoidable. And, if these collaborations have involved recognised leaders, they might make governments take notice.
Covid-19 has arguably provided a template for all this.

Are we ready to use it? The willingness to do so may be what we should first see as ‘the new normal’.