Thinking Global, Acting Local: How Adidas is Personalising CX at Scale

For global brands scaling operations into new markets, adapting to regional preferences and local processes is critical. Success cannot be achieved by throwing a blanket over diverse customer groups. Of course, an overarching global strategy is key to any broad, large-scale expansion programme, but it must be coupled with a deep understanding of regional dynamics and engagement behaviours. Navigating this complexity is anything but simple, but for the brands that get it right, competitive advantage awaits.

It is this promise that sparked a digital and cultural transformation within the customer service function at Adidas. It’s a process under the stewardship of Stijn Bannier – the brand’s Global Digital Product Director – and here he takes a deep dive into how the change initiative is taking shape. Stijn breaks down the role AI is playing in the transformation, the procedures Adidas have in place to avoid technological missteps, the moments when they needed to course-correct their approach, and so much more.

Simon Hall: Stijn, thank you for joining me! Adidas is on a transformation journey right now, evolving its digital CX strategy in a way that trades a global uniform approach for one enriched by local adaptation and cultural sensitivity. What were the key factors that led to this shift in thinking?

Stijn Bannier: There are two elements to this.

First: Our product strategy and architecture are inherently global – designed to serve every region where Adidas has an e-commerce or customer service presence. Everything we build applies to the whole world. That said, each market comes with its individual nuances. In the realm of e-commerce, this might mean working with region-specific delivery carriers, or offering locally preferred payment methods, or adapting to various legal and policy frameworks. On the customer service side, it’s about tailoring to local communication preferences – whether that’s through Facebook Messenger, WhatsApp, KakaoTalk, or other regionally popular channels. There are so many different distinctions that we must embrace.

Second: Beyond features and functionality and looking at the bigger picture, it's essential to our sustainability and long-term growth that our brand’s operation in each region reflects both its global image and the identity of each market. While Adidas is a multinational brand, its expression differs slightly across regions – how it sounds, how it feels, how we market it. This balance must be carefully represented across all our channels – be that our website, our landing pages, our voice channels, our customer service channels, or our digital chat services. The goal is to align our global brand vision while adapting to the localised expectations and cultural context of each region.

SH: AI is at the core of your strategic pivot. Can you break down how your agents on the frontline are using AI technologies to deliver customer service?

SB: We’re integrating AI into our customer service framework in multiple new and meaningful ways – all of which are designed to reshape how we engage with consumers and attend to their needs.

For simple customer service tasks such as tracking the status of a package, or processing a return, or verifying order details, we’ve introduced an AI-powered chatbot engineered for speed and maximum impact. It handles those common, repetitive, predictable queries that sap agents’ time, freeing them up to focus on areas where they can add true value and really move the needle on customer care.

On the flip side to that, we’re bringing AI into the agent experience to empower our frontline teams to handle the more complex and challenging issues with greater efficiency – anything to do with refund processing, resolving payment discrepancies, or facilitating returns for specific items from a larger order, etc. These scenarios typically require deeper contextual understanding and decision-making that goes beyond pre-programmed logic. We’re embedding AI to support our agents at every step in the process – these tools are helping to streamline workflows by summarising customer interactions, classifying cases, and suggesting potential solutions in real-time. It’s all to ensure they can focus on delivering empathetic, accurate, and effective solutions.

In short, the approach is two-pronged: we’re automating basic service with chatbots and deploying AI to enhance agent-customer interactions.

SH: Localised, multi-market AI implementations remain few and far between and represent a daunting proposition. Talk to us about the technology challenges associated with this: With many languages and markets involved, how did you manage the complexity of rolling out new tools across different regions?

SB: Our global architecture is built on a foundation of working with best-in-class vendors. From CRM systems to chatbot solutions, whatever it may be, we prioritise partnerships that offer the most advanced and effective technologies on the market.

We structure the implementation of these technologies in a way that maintains global consistency while ensuring simplicity of use. A perfect illustration of this is the launch of our chatbot: It features a low-code, no-code administration panel, enabling our product and design teams to manage it and condition it independently without developing everything from scratch. Leveraging the tool in this way helped us engage with market-level customer service teams and guide them on their digital transformation journey. Through targeted training they each developed fluency in the system – they learnt to identify customer intents and refine response strategies and adjust language for local relevance – all minus the need to build entirely new flows.

And why is that so important? Although our non-native English employees all understand the language, it will never be their mother tongue. For instance: From a Latin American standpoint, while it’s entirely possible to translate all our conversational elements from English to Spanish, we must consider the regional differences – Argentinian Spanish differs from Chilean or Colombian Spanish. It’s crucial, therefore, that we account for these nuances and empower our regional teams to fine-tune the language and make it more relatable and accessible on a local level – we want it to feel tailored and natural to all our different audiences.

SH: Establishing cultural sensitivity and cultural nuance within AI applications requires deep insight. What methodologies do you use to ensure that the tone and language of your communication tools meet the expectations of diverse customer bases?

SB: Indeed, how consumers perceive a message can vary significantly from market to market – what's considered appropriate or engaging in one may not translate to another. The tone of communication, the use of emojis, or other subtle elements: They’re all variables. Local teams have the best understanding of the intricacies and that’s why their input is so valuable.

When it comes to digital conversational support, we continue to rely on traditional chatbots to uphold precise control. By using scripted flows and content, we can guarantee everything is fully aligned and predictable and we can eliminate the possibility of errors or hallucinations. This is particularly important for protecting our brand reputation and avoiding the type of backlash other brands have experienced with more experimental AI tools.

We harness the generative component of AI in the backend: We use it to optimise flow structures and craft new intent models. And from an agent perspective, we’re utilising AI to help enhance live conversations. AI is undoubtedly a key focus area for us in the contact centre. So much so that we have established a dedicated AI architecture board that incorporates legal personnel, privacy personnel, and other important stakeholders to ensure everything is properly vetted and approved. It may appear complex and elaborate, but this structure helps us avoid any potential missteps through the various ideation, launch, and scaling processes.

The corporate environment we have often means navigating bureaucracy. The upside, though, is that it forces us to think critically and collaborate across teams, across markets, and by the time a tool is released, we know it’s been evaluated and refined by multiple functions. As a global brand, minimising risks is key.

SH: Rewinding now on to the initial phases of the transformation. Were there any moments early on when you needed to course-correct or adjust your strategy based on customer or employee feedback?

SB: We’re constantly monitoring and assessing whether we’re moving in the right direction.

A few years ago, there was a growing movement internally to develop chatbots. At the time, the big promise – much like the one we’re seeing now with agentic AI – was that chatbots could replace human agents and manage customer interactions independently. Back then, however, generative AI was still a distant concept, and AI technology in more general terms was much less advanced. The company implemented a static chatbot that failed to meet customer expectations – in simple terms, customers were unable to get the support they needed and that led to a negative impact on NPS. The chatbot couldn’t deliver on its initial promise, and we saw an increase in customer contact, either because customers couldn’t get to where they needed to go or because the chatbot frustrated them further.

That was the first signal for course correction.

As we progressed, it also became apparent that we were expanding too quickly. We had set out to capture every type of customer query within our chatbot flows, but the ambition outpaced our capacity. In the realm of conversational design, there’s a tendency to focus on creating intricate, comprehensive chatbot interactions that can handle anything and everything. But it’s not effective. We realised that customers were looking for more straightforward support on this channel, and so we pivoted our strategy accordingly – to play to the strength of our digital chat and utilise it in the simple moments. It comes down to balancing conversational design with practicality. Chatbots have limits. Our goal is to guide customers in the right direction as quickly as possible, and since that shift in focus, we’ve seen significant improvements in both the chatbot’s performance and our ability to position it strategically.

We place great importance on overseeing our processes directly. Our priority is to address customer problems and then explore how AI can help. By focusing on the opportunity to improve the customer experience, we’re constantly ready to course-correct as needed. We don’t begin with technology. Instead, we remain flexible, adapting our approach based on what we learn, ensuring that our solutions remain aligned with customer needs and business goals.

SH: Bringing AI models into customer service is obviously a significant undertaking, requiring a progressive, phased approach. What governance structures have you put in place to ensure scalability and the long-term sustainability of all your localised AI capabilities?

SB: It all comes from that global framework we have in place – that centralised infrastructure.

Looking at customer challenges from that overarching perspective allows us to build conversational solutions for specific problem areas across the board – for example, we can look at everything around order status, or everything around returns and exchanges. It’s an approach that powers scalability – we can roll out changes throughout different markets quickly and the workflows can be deployed in multiple languages.

In essence, the focus begins with general customer problems, not language nor localised messages.

SH: Let’s dive into the impact the transformation has had on the business, both in qualitative and quantitative terms. What specific indicators have helped you assess the success of your new approach?

SB: Our key metric is Cases to Order (CTO) – a measure of how many support cases are generated per 100 orders. We’re seeing a clear improvement in this area owing to the new technologies we’ve introduced into the contact centre: Fewer cases are escalating to agents, which drives profitability while also giving our frontline team members the additional capacity they need to focus on the real, difficult questions. The end result is more loyal customers and better retention. We see that combination of elements – CTO and retention – really improving.

And then there’s the impact of AI internally. Average handling times are coming down and our agents have expressed higher levels of satisfaction in interviews. We also look at improvements in NPS – it is, of course, a customer rating, but we look at that rating as a reflection of their interactions with both our chatbot and live agents.

SH: Last question to bring us home, Stijn: Have you seen significant differences in engagement levels or feedback across regions? What success stories stand out from this transformation?

SB: We’re proud of the global reach we’ve achieved and the positive outcomes we’ve had working with teams in different regions.

Personally, I’m particularly fond of Latin America – we have extremely dedicated and passionate people working on the chatbot over there. But every region brings something different to the table and we see tremendous engagement once we get started. All our regional teams inspire us to constantly seek new ways to enhance our approach, work more cohesively, and challenge one another to elevate our efforts to the next level.

Our journey began in the North and South American markets, and by the end of 2024, we had expanded into Europe. Now, as we approach the final phase, the remaining markets are coming on board in the next couple of quarters. I’m excited to see the results, and by May and the CCW UK Summit, we’ll likely have insights to share.