AI is predicted to transform how retailers operate. We delve into how it will impact the major functions within retail.

When reading about the impact of artificial intelligence on the retail sector, it’s easy to go down the doomsday route.

Deloitte predicts that 60% of retail jobs are at a high risk of automation in the next 20 years, higher than any other industry. PwC forecasts that 42% of UK wholesale and retail jobs could be automated by the mid-2030s.

As we stare down the barrel of the fourth industrial revolution, which analysts believe the advent of AI will usher in, how will jobs across retail change?

Accenture retail managing director Vish Ganapathy says: “Artificial intelligence in retail will effect everybody from merchants, the store, HR, accountants, category managers, planners, the supply chain and marketing.”

Ganapathy says the rise of AI means “some functions will be disintermediated but I don’t think that translates to people being disintermediated”.

He believes AI will deconstruct each role into execution tasks with a high degree of predictability, which can be automated and improved with AI, and tasks that require emotional intelligence and instinct.

How will the day-to-day reality of retail jobs from the shopfloor to the boardroom change as AI advances? We take a closer look at functions across the retail organisation to find out.

Warehouse workers

Automation is already having a huge impact on the retail warehouse. Just look at Ocado. The grocer’s highly automated fulfilment centres, which are powered by swarms of robots zipping along grid systems picking and packing goods, have proven so successful that retailers across the world are lining up to partner with Ocado.

PwC chief economist John Hawksworth says the repetitive nature of the work done in a warehouse make it ripe for automation to disrupt.

Ocado warehouse andover3

Retailers around the world want to use Ocado’s automated warehouse technology

“The reason automation is relatively easy in factories is because it is a very controlled environment, which is where automation works best.

“Warehouses are a little less well controlled but you can still see, as robotics advance, jobs could start to go from there, particularly as robots become better at manual dexterity type tasks.”

However, the highly automated nature of warehouses in the future means the humans who work in them will take on a more highly skilled, supervisory role.

“Your value as a worker will definitely be enhanced if you know how to work around the robots,” he says.

Hawksworth believes that in years to come, human roles in retail warehouses will primarily be divided into two types.

The first will be an engineer-type role, which will supervise the robots and oversee repair and maintenance as needed. The second will be coders who write the software and control the processes of the robots in the warehouses.


Doug Stephens, founder and president of consultancy Retail Prophet believes AI will carry out many duties today fulfilled by store staff.

He says: “Most inventory management functions, and many managerial tasks such as scheduling to productivity, will become redundant as AI begins performing this work with infinitely greater speed and accuracy.”

While AI is unlikely to wipe out specific shopfloor jobs entirely, the advances of this technology coupled with robotics increasingly being used to stack shelves, clean floors and unload inventory, make it inevitable that the number of jobs required to run a store successfully will fall in the years ahead.

retail robot

Advances in robotics will lead to a decline in humans working in stores

Stephens argues that this is not necessarily a bad thing.

“If anything, AI may have an oddly humanising impact on retail by moving people away from data-driven work and more toward creative and interpersonal skills,” he says.

The role of the store manager is also likely to change.

“The role of the manager will become far less of a backroom occupation and will rely on talented, creative and outgoing people to motivate staff and add a personal touch for consumers,” says Stephens.

Deloitte Digital chief disruptor Ed Greig says AI can help store managers make decisions and empower their staff.

“Artificial intelligence should be used in store to allow managers to see the potential impact of decisions which they are considering making in real time,” he says.

“One option could be the ability to type in ‘I want to see the sales data from this product or region arranged in this specific way’ and have it presented to you immediately.

“If you can empower shopfloor colleagues to see the impact of what they are doing and surface that in a way that is easily understandable, it would make a difference.”

Greig says AI could be used to suggest potential offers or points of sale promotions that could perform well based on other stores or regional and demographic demand.

HR and finance

The rate of job losses in head office as a result of AI and automation will be much less extensive than in a retail warehouse – but the impact this technology has on what tasks are carried out will still be significant.

“AI will remove the mundaneness from our daily routines,” says Greig, who adds there is “massive” potential to free up employees to focus on more “high-value tasks”.

“The kinds of jobs we are looking to automate are the things that people don’t like to do anyway”

Vish Ganapathy, Accenture

Ganapathy believes AI will have the most significant impact on finance and human resource departments in the short term. He says in the next five years, 50% of an accountant’s day-to-day tasks could be automated, and over 70% of those carried out by an entry-level HR employee.

“The kinds of jobs we are looking to automate are the things that people don’t like to do anyway, like approving payments for a vendor or on paperwork relating to a new recruit,” he says.

Greig concurs: “A great example is being able to type in a query in natural language. Say you need the sales data from this country arranged this way. This could be presented immediately using artificial intelligence. Right now that would currently be a multi-week process.

“It is not necessarily sexy, but if you eliminate all the time spent on that sort of stuff it frees people up to solve the problem they are actually there to solve.”

In the medium term, AI will take over the tasks that dominate the time of junior employees across HR and finance departments, such as raising purchase orders or onboarding new starters. This means fewer roles will exist in these departments in future, but those that remain will hold more strategic significance.


The speed of being able to access data will also impact the work done in retailers’ marketing departments.

Greig explains that when retail marketers launch a new campaign the process of collecting feedback and reporting on the findings takes  months. AI will shift this to a real-time process allowing marketers to respond and tweak campaigns.

“From a marketing point of view, AI will give retailers instant access to the effectiveness of what they are doing, with reports automatically generated on a weekly or even daily basis,” says Greig. “You can action it in real time, so your job becomes less about reporting and more about experimenting.”

Many of the day-to-day activities of the retail marketing department, such as social data gathering, A/B testing on emails and product promotion, will be superseded by AI.

This inevitably means fewer jobs, as demonstrated by Zalando’s decision last March to axe 250 roles from its marketing department in favour of AI and algorithms.

The roles that remain in retail marketing divisions will operate in two basic strands. The first will be more junior and revolve around a team that oversees the algorithms used to gather information and ensure they are capturing the right data as effectively as possible.

The increased efficiency of these departments will mean there is far more data to analyse at far more regular intervals. Therefore, the second type of new marketing roles will comprise senior marketers who will combine creative and strategic thinking to ensure the greatest return on investment based on the information that AI delivers.

Buying, merchandising and design

Ganapathy says the ability to dedicate more time to experimentation and less time to process will also change how merchandisers work.

He gives the example of work Accenture did with a department store retailer, which included reimagining how its merchandising team work. It found the team’s workload could be cut 30% using AI and automation.

“We discovered a lot of the merchants are spending a lot of time in relatively mundane stuff like approving a PO, making sure product attributes are correct, making sure a vendor is uploaded on the finance system and the latest instructions and manuals related to a product are on their vendor portal site. What we did was took those tasks, deconstructed and figured out which ones humans can and should do and which ones machines can take over,” he explains.

However, rather than simply using the time savings to justify cutting jobs, the retailer opted to allow those people to focus on bigger, more strategic projects.

London College of Fashion’s head of fashion innovation agency Matthew Drinkwater believes AI will make store merchandising much more hyperlocal.

Yoox mirror

Yoox Net-A-Porter used AI tools to help its design team create a new range

“You will get to a point where AI and machine learning understand exactly what is selling in real time, adjust orders accordingly and predict what will sell in the future with a great deal more accuracy than a human can, which will mean no more discounting on products that missed the mark or running out of bestsellers,” he says.

As a result, retailers will have much leaner buying and merchandising teams in future. Drinkwater says having a computer science degree to develop the algorithms used to monitor trends and move stock accordingly will be vital.

However, Stephens insists a degree of creativity will still be crucial.

He explains: “The linear nature of AI makes it virtually incapable of lateral creativity or innovation. Consequently, the role of fashion buyer will become much more creatively centred – recognising nuanced expressions of style in the market, creating unique brand and product adjacencies and developing emotionally driven collections of products will become the primary occupation. While data may inform that creativity it will not, at least for now, replace it.

“Any occupation that relies more heavily on either physical ability, such as visual merchandisers, or creative ability, such as store designers, will remain in healthy demand.”

In product design, AI will not replace roles but will help enhance creativity, experts argue.

Ganapathy explains: “If I’m a fashion designer trying to figure out two seasons in advance which colours are going to be in, AI can trawl through social images, buying patterns and colours trends and suggest that lilac, for example, will be in. That is a piece of intelligence which can be infused into my decision process.”

Online retailer Yoox Net-A-Porter has done just this with the launch of its own-label range in November.

The retailer used its proprietary AI tools as a starting point to gather content across social media and fashion retailers in key markets.

These results of these searches were used alongside data on products sold on its website, customer feedback and top trend searches to create a mood board for YNAP’s design team to work from when creating the range.

“In the short term, there will be a collaborative relationship between designers and data scientists, but in the longer term, the two roles will fuse together”

Matthew Drinkwater, London College of Fashion

However, there is an argument that AI is capable of designing products better than humans. Indian fashion etailer Myntra, owned by Flipkart, used AI to design a T-shirt that outsold the equivalent human-designed products.

Drinkwater says: “Anyone who is still working on paper [in fashion design] needs to be really terrified.”

He predicts that in 15 years the essential skills of a fashion designer will not lie in illustration or textiles, but in data.

“In the short term, there will be a collaborative relationship between designers and data scientists, but in the longer term, the two roles will fuse together in a creative data scientist or creative technologist role,” he says.

This will effectively mean a fashion designer’s primary role will be as a sense checker for the designs AI generates, rather than creating designs themselves from scratch.

Technology and ecommerce

While the employee numbers across the majority of retail departments are likely to diminish as a result of AI, the technology division is likely to increase in size.

As AI becomes central to many retail functions, technology will become a “quasi-colleague” in its own right – which means the IT and technology team will have to work much more closely with employees across every other business function, says DynamicAction co-founder and chief scientist Michael Ross.

“Data is the oil which fuels AI’s engine, so retailers are going to be required to have a really interesting mix of different roles in their IT teams to get the most out of the tank,” he says.

Ross expects to see a chief data officer at board level but warns that re-engineering the whole technology team will be necessary to get the most out of AI across all parts of the business.

He expects a big spectrum of new roles, such as people who work out how and where to collect the data required for each function, data engineers that work out how to store data and “knit it together coherently”, engineers that focus on building out the raw data into useful features for various business functions, and algorithm designers.

“Business people across all divisions need to be upskilled in tech terms. The key role will be a translator who sits between the business and the AI functions”

Michael Ross, DynamicAction

But Ross argues the most critical role day to day will be a translator.

“Business people across all divisions need to be upskilled in tech terms. The key role will be a translator who sits between the business and the AI functions being deployed,” he says.

The data translators will essentially act as a bridge between various business functions and the IT team.

Ross adds: “The translators will become fundamental to help retail leaders ask good questions of the AI at the business’ disposal and then ensure that the IT employees take those questions and transform them into understandable and tangible business insight.”

While Ross says the size of a retailer’s technology team will increase as AI develops, he believes its ecommerce division will likely shrink as the marketing and merchandising roles in the department converge.

He predicts the ecommerce team will transform to be much more like a “highly automated banking trading floor” with employees given end-to-end responsibility for a particular product category and how it is marketed and sold online.

Ross says AI will be used to provide these employees with real time information on metrics such as predicted sales, website traffic and inventory numbers. They will then use their judgement on “additional levers which can be pulled to increase these metrics”, such as targeted promotional emails or social messages.

The chief executive

AI will be used across retail on everything from driving efficiencies to bolstering creativity. But what are its implications for the leader of a business?

For the time being, Massachusetts Institute of Technology professor emeritus Frank Levy believes AI has little place in the boardroom.

“If you think about [the role of chief executive], at its crux is the calculation of how much and what kind of risk to take on as a business,” he says.

Driverless car, autonomous vehicle

Driverless cars illustrate the difficulties of getting AI to make complex decisions

“The whole notion of risk is that you are in an area of uncertainty and make your best judgement. By contrast, most AI needs a lot of data and relevant experience to train in strategic business decisions and even then there is still a level prediction baked in.”

He gives the example of the development of driverless cars to illustrate the limits of AI.

“One of the reasons autonomous cars are so difficult is you are trying to create software that can react to any kind of situation a driver can confront and that is a hugely difficult task,” he explains.

“A lot of the CEO’s job is to confront situations that haven’t been confronted before, and if you can do that without a CEO why would you need one at all?”

AI may not usurp the role of the chief executive, but its predicted impact on every function in retail means it will become a central part of a business leader’s life in the years to come.