The problem with online returns: solved 75% faster with AI

Project overview

Retailers across the globe are under mounting pressure from the financial, operational and environmental impact of product returns – particularly from online sales. Driven by pressing client demands in retail, Zühlke launched a proof-of-concept experiment to examine how AI, when combined with our Cybernetic Delivery Platform (CDP), can dramatically accelerate the discovery phase, improve solution quality, and drive business outcomes.  

In just five days (75% faster than usual) we uncovered several root causes of online returns, generated insights and prototyped the most promising solutions.  

This research demonstrates that the real value of AI assistants lies not in incremental, task-level gains, but in transforming how interdisciplinary teams can collaborate to unlock substantial productivity improvements – especially in the early phases of digital product and service delivery. Responsibly applied, enterprise-ready AI cuts through large datasets to diagnose and solve complex challenges. This helps clients reduce costs, accelerate time to value and make faster, more informed decisions. 

Go-to-market approach for retail: reimagining discovery

In a go-to-market effort for the retail sector, our Digital Consulting team felt driven rethink how we address one of online retail’s most costly and wasteful: returns. 

Previous projects in this area indicate that a discovery phase of this nature typically spans thirty days. But, drawing on our rich consulting and engineering experience, we deliberately reframed this constraint as a catalyst for innovation – and a strong opportunity to test how the Cybernetic Delivery Method^TM (CDM), as part of our Cybernetic Delivery Platform, could power AI-augmented delivery. 

This methodology speeds up early product ideation and discovery phases without compromising quality, while also bringing tangible value to the client through faster insights and data-driven choices. 

This reframing of the challenge became the foundation for how we structured our entire approach, helping our team demonstrate how we can dramatically accelerate the discovery phase in all client projects. 

By turning data overload into actionable insights, we enable faster, more informed decisions for both technical and business teams. 

Why this matters - not only to retailers

Every year, online retailers around the world are haemorrhaging value. The numbers speak for themselves: 

• £6.6 billion worth of online purchases are returned annually in the UK alone
• 40% of clothes bought online are sent back
• Each return costs between €6 and €18 to process
• In 2020 alone, returns generated 2.6 billion kilograms of landfill waste

These figures are more than statistics; they are alarm bells. For our retail customers, this is no longer a marginal issue. It is a core threat to profitability, sustainability goals, and customer satisfaction.

The overarching objective: establish a clear, shared understanding of the status quo. Fast.

Every successful consulting engagement starts with a few key questions:  

• What business outcomes are you trying to achieve?
• What is the real problem that prevents you from reaching your goal?
• Why does solving it now matter to your business? 

Answering these requires a shared, data-driven understanding that unites every relevant stakeholder – from business leaders to engineers. 

This is where AI-enhanced discovery makes a difference: it accelerates clarity, improves collaboration and sets the stage for scalable innovation.  

Our focus: 5 levers for more efficiency in the discovery phase

We first set out to address critical inefficiencies found in project workflows – especially during discovery phases – by applying AI to streamline and enhance key activities with a focus on five areas:

• Saving time: Significantly reduce the duration of the discovery phase, particularly during research, prototyping and user testing, by using AI tools in place of conventional approaches
• Augmenting workflows: Replace repetitive and mundane tasks, freeing up teams' time for more value-added work.
• Ensuring accuracy: Test a variety of AI tools, and select the most accurate when compared to traditional methods.   
• Anticipating pitfalls: Proactively address ethics and responsible AI adoption, with a focus on fairness, bias and privacy.
• Identifying risks: Ensure careful handling of sensitive data inputs to prevent confidentiality breaches.  

Adding value with CDP: an AI-enhanced discovery

We structured the project around two workstreams: deliver actionable, user-tested ideas to solve an industry-wide problem for retailers and demonstrate the business impact of smart AI adoption. 

To do this, Zühlke’s experts applied the Cybernetic Delivery Platform (CDP) – our modular, self-service system designed to embed AI accelerators into digital product and service delivery. 

The CDP is modular, stakeholder-friendly, and has reliability, security and adaptability built into its DNA. And it can be rolled out in full or selectively – via modules like the Cybernetic Delivery Method™ (CDM) – depending on each project's needs. 

First result: Discovery in five days instead of one month

The discovery phase typically takes up to thirty days. Using the Cybernetic Delivery Platform’s CDM module, we reduced this to just five. AI-supported tools embedded in the platform enabled us to:

• Undertake comprehensive end-customer and market research, rapidly synthesising and distilling data into key insights
• Generate and validate personas
• Translate concepts from ideation into wireframes and prototypes
• Accelerate prototype testing with stakeholders and end customers to enable data-driven decisions on development priorities
• Create project artifacts – like playbacks and retrospectives 

This resulted in a 75% acceleration in delivery of the discovery phase, without compromising quality or results.

Second result: Solving the issue before it scales

Time saved in the discovery phase allowed us to go beyond status assessments and focus on strategic insight: we stepped back to examine not just what happens with returns, but why they happen – and how clients could prevent them. 

Through customer research and data analysis, we uncovered two key insights: 

1. Sizing tools are often underused because they lack personalisation and user trust.
2. Consumers are often unaware of the environmental impact of their returns. Even small nudges that highlight this can influence behaviour and encourage more sustainable choices.

The how: The methods and tools that cut discovery time by 75%

Traditional discovery approaches often struggle to keep up with evolving data sets, even when AI is involved. That’s why we applied Zühlke’s Cybernetic Delivery Method™ (CDM) – a structured but adaptable approach within the CDP to design AI-enhanced solutions.  

CDM prioritises: 

• Human insight
• Systemic thinking
• Ethical innovation
• Patterns that offer consistent solutions

CDM supports the full product lifecycle – ensuring consistency, traceability and responsible AI use. Our team combined the CDM module with a curated set of AI tools that ensured human oversight and kept purpose front and centre. 

With CDM, we’ve already achieved reliable, repeatable productivity gains of 30% in AI-enabled projects. We’re now pushing that figure to 50% and beyond. And, in this instance, reduced the time required for the discovery phase by 75%. 

Broader application: Beyond retail

While this experiment focused on online retail returns and adding value for retailers, the approach and outcomes are relevant far beyond this sector. Any industry facing fragmented data, legacy systems or slow discovery phases (including logistics, healthcare or financial services) can benefit from AI-enhanced consulting workflows. 

The modular nature of Zühlke’s Cybernetic Delivery Platform makes it adaptable to a wide range of settings. Whether the goal is to modernise infrastructure, improve customer experience or uncover hidden inefficiencies, structured and responsible use of AI can drive meaningful results – quickly and reliably. 

Conclusion

AI assistants undoubtedly enable greater efficiency in routine engineering work. Tools like GitHub Copilot boost individual productivity and free up time for what really matters, like understanding the client’s problem space more deeply. But, overall, these task-level productivity gains remain incremental – rather than representing a fundamental shift in how work gets done. 

The real value of AI assistants emerges when they’re used to transform how interdisciplinary teams work, unlocking significant productivity gains at the departmental or organisational level. This is particularly powerful in the early phases of digital product and service delivery, as this case clearly demonstrates.

In short, AI doesn’t just speed up discovery, it strengthens it. 

Thoughtfully applied, AI enhances clarity, accelerates delivery and supports better business outcomes. The Cybernetic Delivery Method – Zühlke’s Cybernetic Delivery Platform (CDP) module – proved transformative in this instance, helping us solve a complex challenge for our retail clients with speed, precision and measurable impact.