糖心视频

Key Messages

• Retail is set to be amongst the top industry beneficiaries of AI
• AI predicted to improve retail efficiencies by +59% by 2035
• Ca. 50% of retailers already claim to use AI 鈥渋n some form鈥�
• But only 28% view AI as a top investment priority
• And 40% lack the investment to implement AI strategies
• Exploration to date restricted to major retail players
• Dictated by the need to deliver quick and measurable ROI
• Not all AI implementations are 鈥榝uturistic鈥�
• Most focused on solving routine inefficiencies across the business
• AI is not a silver bullet
• AI will not single-handedly solve the sector鈥檚 structural issues
• Vanity projects will gain most attention鈥�
• 鈥�but more 鈥榤undane鈥� interventions are likely to have the greatest impact
• AI could be key in quantifying the true value of the physical store鈥�
• 鈥� and aid the setting of realistic and sustainable (turnover) rents.

Hot on the heels of last week鈥檚 AI safety summit at Bletchley Park, 糖心视频鈥檚 own musings on the potential impact of AI on the both the retail and retail property markets.

The fact that Rishi Sunak鈥檚 event was dubbed a 鈥渟afety鈥� summit (and Elon Musk鈥檚 declaration that AI was 鈥渙ne of the biggest threats to humanity鈥�) perpetuated the notion that AI is principally the ultimate threat to our very existence. Meanwhile, the release of a 鈥渘ew鈥� Beatles song and video created through the use of AI also served as a timely reminder that AI can equally be a force for good.

The narrative around AI will always be hyperbolic, the realities far more nuanced. And this is certainly true of its potential role in the retail sphere. AI will most definitely shake up retail and be a major driver of structural change. At the same time, it also has to deliver clear returns and ultimately avoid becoming nothing more than a series of vanity projects.

AI 鈥� the basics

First of all, AI is not a new phenomenon 鈥� it has been around for decades, quietly changing the world around us since its emergence in the 1950s. Although 65% of retailers claim to understand AI 鈥渨ell鈥� or 鈥渧ery well鈥�, there is actually no agreed single definition of AI. Rather, AI is acknowledged to be a 鈥渇amily鈥� of different technologies and systems, capable of performing tasks normally requiring the intelligence of a human. Examples of AI technologies within this 鈥渇amily鈥� system include 1. Algorithms: 2. Machine Learning, and 3. Deep Learning.

The government is anticipating AI will change the UK economy on an unprecedented scale, providing a £400bn boost by 2030. For the average UK worker, productivity gains are expected in the region of 100 hours per year, but some sectors are set to benefit more than others. GPs and teachers, for instance, could save 700,000 hours usually spent on administrative work through AI, freeing up £8bn worth of public sector resources.

And, according to Mr Musk, 鈥渢here will come a point where no job is needed鈥�...

AI and Retail

To date, retail has lagged behind more 鈥楢I mature鈥� industries, such as finance, science and the military, causing it to miss out on potentially major productivity benefits. Currently, just over a quarter of retailers (28%) regard AI a 鈥榯op investment priority鈥�. There is, however, growing acknowledgment amongst retailers that AI warrants further investigation and many are weighing up the level of investment to allocate.

A study by Accenture identified retail as one of the top 4 sectors to potentially benefit most from AI (from a total sample of 16 industries). AI could deliver a 59% boost to profitability across the Wholesale & Retail industry by 2035, compared to a wider average of 38%. Those who move fastest can reap even greater rewards: early adopters expected to enjoy 8% higher profit margins.

But how exactly will AI drive profitability increases for the retail sector? The theory goes that by augmenting AI with its human workforce, productivity and efficiency improvements will drive profitability. For instance, self-service checkouts enabling one staff member to handle multiple transactions at once, reducing staffing costs. With AI technologies capable of optimising both operators鈥� 鈥渂ack end鈥� (automated warehouses) and 鈥渇ront end鈥� (customer service chatbots), the accrual of even small implementations across retailers鈥� value chain could lead to major gains. All basic retail disciplines can be revisited using AI technologies, such as what to stock, when, and how much.

Experimentation to date

Well-known examples of AI deployment have typically been the most glamourous and futuristic, such as Amazon Go鈥檚 checkout-less stores, and Ocado鈥檚 automated robot warehouses. It鈥檚 debatable whether these deployments have been wholly successful - Ocado pausing rollout of its new distribution centres in February following a record £500m annual loss, and Amazon closing several Just Walk Out-enabled stores in July. The price of an education?

Lesser-known deployments may perhaps be regarded more mundane and have gone undetected by typical consumers. Some 50% of retailers claim they already use AI in some form. This is not surprising considering many retailers hold vast amounts of customer data from their loyalty schemes. Some of the best operators (Tesco, Next, Argos) have been deploying this data for years, but AI technologies are now taking this to a much higher level.

For example, Morrison鈥檚 deploys AI to assess historic sales data of individual stores, alongside local weather reports, to predict future demand - reducing shelf gaps by a substantial 30%. ASICs, the footwear brand, invested in Aura Vision in 2016 to conduct analysis across its physical estate. The London-based tech startup uses CCTV to track customer movement, providing insight on store layout and product engagement, to improve conversion of footfall to sales.

It's not just retailers themselves adopting these technologies. The New West End Company (NWEC), a retail body representing 600 retail and leisure operators across London鈥檚 major streets, use of AI helped it drive £100m in additional income by keeping shoppers spending. Previously relying on analogue street surveys, the AI aggregation of mobile, spend, and flight data provided a better profile of visitors. Insights identified some of the most affluent visitors came from less visible markets, prompting it to encourage retailers to employ staff with wider language skills.

The AI endgame: improve profitability

Whilst the actual benefits of some AI deployments may be unclear (one third of UK consumers believe customer service has worsened, with 50% citing chatbots as a source of frustration) 鈥� the ultimate goal should be driving profitability. A clearly defined ROI (return on investment) is required to prevent ventures becoming vanity projects.

However, retailers appear alert to these risks. 67% of retailers allegedly possess 鈥榗lear AI investment plans鈥�, but 40% see a lack of available investment as the second biggest hurdle to implementation. The need to demonstrate quick returns on investments may stifle AI progress in retail 鈥� but will force stakeholders to fully consider how AI tools could best solve their problems.

AI and the relevance to retail property

At first glance, it may feel like AI benefits are limited to the digital retail realm. But as outlined in our 10 Structural Failings in the 鈥楻etail Renaissance鈥� report, the transition to multi-channel retailing and increasing convergence of online and physical stores will be a pivotal facet of the sector鈥檚 evolution.

AI may also prove an effective way to manage many of the sector鈥檚 wider structural issues too. Operators could minimise 鈥楤rand Devaluation鈥� (Structural Failing #8) by gaining a more intimate understanding of their customers using data analytics.

Perhaps slightly controversially, it could also tackle 鈥榃ider Cost Inflation鈥� (#5) by reducing unnecessary staff costs. There is a fear AI could eliminate jobs across the sector, but the technology will still need to be carefully balanced with human instinct and experience. Retail does serve human consumers after all, not robots.

Amazon鈥檚 UK Fresh stores, whose format stripped staffing to a bare minimum, are an excellent case in point. Amazon clearly has access to some of the best AI capabilities, and yet the closure of two stores this year perhaps highlighted an overreliance on data, to the detriment of other basic retailing disciplines 鈥� analysts pointing to 鈥榗linical鈥� stores in locations where customers were perhaps not the most tech-savvy. In effect, giving customers what they thought they wanted 鈥� but, in fact, they didn鈥檛.

Better management of the 鈥楿gly Tail鈥� (#3) could also be achieved, with AI helping unveil the true value of the physical store. To effectively implement AI, retailers will have to undertake a huge amount of housekeeping, consolidating siloed data spread across their physical and digital estates. Data will have to be maintained on an ongoing basis, enabling up-to-date or even real-time access to individual stores鈥� KPIs. This would go some way in solving the notoriously difficult quantification of physical stores鈥� online 鈥榟alo effect鈥� in deciphering what is a sustainable (turnover) rent.

Although good in theory, the reality is most retailers will likely remain suspicious of data sharing provisions in leases which could lead to a ratcheting of their costs. Furthermore, forensic capabilities to investigate the impact of store closures (e.g. sales uplift to neighbouring stores), would likely embolden retailers鈥� mantra that 鈥榥o store is sacred鈥� in landlord negotiations.

The bottom line: AI is just one tool in retailers鈥� armoury

AI is not a silver bullet. Not every problem can be, or should be solved by AI. Where AI is implemented, it will require the guidance of humans, who must ensure the technologies effectively tackle clearly defined business objectives.

And more Beatles records. Please.

To access Emma鈥檚 full article, including a fascinating case study on Shein and its use of AI, please download the full 鈥楻etail Renaissance鈥� report.