MongoDB: How AI can Target Shoppers During the World Cup

In case you somehow missed it, England lost to Argentina in the FIFA World Cup 2026 semi-finals this week. I watched the second half on the train home from London with a group gathered around the phone of the man sitting beside me (who had brilliant 5G connection with EE – perhaps a story for another day). 

While it was tough to accept football won't be coming home, the comradery was unmatched. In fact, the woman sat across from me said how she’d be cancelling her annual leave for Monday, since we won’t be in the final. It’s probably the same for a lot of us and the disappointment of millions of fans has a knock-on effect on retail. 

Behind the scenes of our favourite e-commerce apps, delivery services and supermarkets, a different kind of scramble takes place. Within seconds of the final whistle, consumer behaviour shifted with the demand for celebratory party platters and cases of beer evaporated, replaced by a collective desire to pack up, head home or order comfort food.

I woke up the next morning to see this shift first hand in my inbox, which was flooded with marketing emails titled ‘It’s not coming home 💔’ – cleverly offering an extra 10% off already ongoing sales to soften the blow. 

But just how do retailers respond to these overnight emotional swings? Genevieve Broadhead, Global Industry Lead at MongoDB, advocates for how modern technology allows brands to adapt their storefronts, promotions and supply chains in real time to match the unpredictable chaos of live events.

From historical data to real-time reactions

Historically, retailers predicted the future by looking at previous customer behaviour. 

“Retailers would have used past sales to judge how much stock they needed for a particular year,” Genevieve says. “Understanding how many beers or how many jerseys we sold in the previous World Cup would have helped them to understand what they might sell in the next.”  

While historical modelling can help estimate baseline demand, it fails entirely when the whistle blows. The true differentiator today is the ability to react to what Genevieve calls "signals in real time".

Imagine you are watching a match at a friend’s house. “If your team loses, unfortunately, a lot of people will pack up and go home,” she says. "If your team is winning, you might want to continue the party. You might jump onto Uber Eats and get a few more beers or pizzas."

For a delivery app or supermarket, capitalising on these split-second decisions requires highly-dynamic tech. 

Genevieve asks: "How can we make sure that apps that sell in real time are able to promote the right kinds of things to people? It’s all about understanding for that match day, what people usually eat or if someone is winning or losing, are they more likely to want a beer or are they more likely to want a kebab, for example?"

Up until recently, processing these insights was too slow. 

“It used to be very difficult as all analytics were done in data warehousing systems or the kind of analytic systems where you’re looking at it once every few hours and asking one query at a time,” Genevieve says.

Today, databases like MongoDB’s bypass this lag by unifying operations and analytics. 

“MongoDB’s technology is perfectly designed to run not only the transactional workload – so checkout searches and inventory management – but also runs analytics in real time based on what’s happening in real time, be that a weather event like the current heat wave, or when someone scores a goal in a match."

Moving from broad demographics to the segment of one

One of the most exciting aspects of modern real-time retail is how generative AI is contributing to personalisation. Instead of manually building promotional campaigns for every possible sporting outcome, retailers are using the tool to generate targeted content on the fly.

“AI-driven personalisation is really interesting,” Genevieve says. “When you log into an app – and I think the food delivery apps are a great example of this – you’re going to see personalised banners and categories.”

She paints a vivid picture of this in action based on the England-Argentina game, which took place the night before our interview: 

“If you’re supporting Argentina and it’s the day of the match, you might go in and see white and blue. You might click on a supermarket app and suddenly see a section that is selling steak for your Asado and special Argentinian beer. Generative AI is really changing the way personalisation can be done because it can be done every single day for every single World Cup match.”

Furthermore, AI allows retailers to move away from outdated demographic assumptions and toward highly individualised marketing.

“We’re moving from the broad sweep generalisations into segment marketing,” Genevieve explains. 

“In the old days, maybe you wouldn’t have received a lot of World Cup pieces just because you’re a woman and, typically, the audience for a lot of football is traditionally male… What retailers do now is really look at your buying patterns to understand what you’re buying, when, and perhaps why.”

If an algorithm detects that you regularly buy crisps, dips and beers on match nights, it profiles you as a match viewer, regardless of your age, gender or postcode. It builds a profile based on your active preferences, contributing to a personalised consumer experience.

Managing the supply chain chaos: the Mexican Jersey Effect

While digital storefronts can be updated in milliseconds, physical supply chains are notoriously rigid. What happens when an unpredictable trend catches a retailer entirely off guard?

“Some things are predictable,” Genevieve points out. “England are in the World Cup so retailers are going to sell more England jerseys.”

However, social media has introduced a new layer of volatility. The Mexico national team jersey has been Adidas’ best-selling team kit globally this year, largely due to the hype surrounding Mexico as one of the host cities for the World Cup.

“The most sold jersey this year worldwide was the Mexican jersey because of all of the positive social media feedback that people were seeing on Instagram and TikTok. It fed an urge for Mexican jerseys that a lot of brands had not predicted.”

While you cannot instantly fabricate a physical product out of thin air, real-time data allows retailers to optimise the stock they do have.

“If you’re in a situation where you happen to have those jerseys in a warehouse somewhere, you can use technology to change your shipping and distribution networks to get them in front of customers in store or make them available online for quick next day delivery,”  Genevieve says.

By running analytical workloads on secondary database nodes in real time, retailers can quickly route inventory to physical stores or dispatch hubs where demand is spiking, transforming a potential logistical nightmare into an omni-channel win.

Scaling through the spikes without breaking the bank

For retail engineering teams, the ultimate test is surviving sudden, unpredictable traffic spikes without the website crashing or overspending on idle servers. 

Whether it is a World Cup penalty shootout, a Wimbledon final, or Taylor Swift fans buying out local bead supplies for the Eras tour, modern consumer demand is incredibly spiky.

In the past, retailers had to provision expensive data centres year-round just to handle Black Friday traffic. Today, modern cloud architecture allows databases to autoscale on demand.

“MongoDB will scale up and down automatically during and after a peak in demand,” Genevieve says. Crucially, it does this by scaling horizontally rather than vertically.

“Typically systems will just scale vertically,” Genevieve says. “So, it means retailers may have one unit that just gets bigger and bigger and bigger. Of course, that means you’re creating a single point of failure. 

“MongoDB is built in a really clever way with its distributed architecture that you can also scale horizontally, which means you’re splitting up one unit into many units, which means you’re splitting up the traffic between all of those different nodes.”

This architecture keeps systems online during peak chaos while keeping infrastructure costs highly efficient.

The power of the memory layer

As we look to the future, the integration of retail and AI will only deepen. Genevieve believes the next major frontier is the memory layer in conversational and agentic commerce.

“The most important thing that makes them really usable and accurate is the memory layer,” she concludes. “Once you have a memory layer, it can understand what your previous buying behaviour has been, what kind of things you like to buy, what kind of questions you’re asking – everything to do with who you are. And so the next time you come, you’re going to get a more personalised response.”

Ultimately, whether England wins or loses, the retailers who come out on top will be the ones who can listen to their data, understand their customers as individuals and pivot their entire operations in the blink of an eye.