Leveraging AI in eCommerce: Sustainability, Due Diligence, and Roadmapping

Brave Bison recently teamed up with our partner Klevu to co-host the Discovered Fringe Breakfast at eCommerce Expo 2024.

PLAY VIDEO

...

During the event, Carole Breetzke, Business Solutions Director at Brave Bison, joined the panel on Sustainability, Due Diligence, and Roadmapping. The discussion centred on building robust, future-proof eCommerce strategies by leveraging AI, featuring insights from James Gurd (Digital Juggler), Andy Eva-Dale (Tangent), Sezin Cagil (Selfridges), and Carole Breetzke (Brave Bison). Read on to explore the key takeaways.

In today’s fast-paced digital world, AI is transforming eCommerce in exciting ways, helping businesses optimise customer experiences, increase conversions, and reduce costs. However, adopting AI requires careful consideration to ensure the technology aligns with business needs and delivers real value. In this post, we explore key discussions around integrating AI in eCommerce, from vendor selection and proof-of-concept testing to sustainable practices and governance. Here are 6 key actions you need to take to successfully integrate AI into your eCommerce strategy and drive sustainable growth.

1. Choosing the right AI partners

When selecting an AI vendor, it's essential to thoroughly vet their technology. A few key questions can help guide the evaluation process:

What is the algorithm built on?

How long has the technology been in use?

What makes this technology stand out?

AI technology, especially in areas like search, merchandising, and personalisation, is mature and widely used in e-commerce. But, for newer innovations, it’s crucial to dig deeper into the technology’s unique selling proposition (USP). A balanced approach of trust and caution should be adopted—sometimes it's worth taking a leap of faith on emerging tech, but due diligence remains vital to avoid costly mistakes. Ensure the vendor has credible evidence to support their claims and always check their track record with conversion rates and ROI.

Adopting AI shouldn’t lead to drawn-out processes like endless RFPs (Request for Proposals). Instead, brands need to leverage modern low-code or no-code environments to pilot AI projects quickly. While speed is essential, it’s equally important to assess the technology carefully, ensuring that it solves specific business problems.

2. Test AI with proof-of-concept trials

Proof-of-concept (POC) testing is essential when selecting an AI vendor or technology. Many modern partners are willing to collaborate with development teams to run small-scale POCs, allowing businesses to assess the potential of a technology before committing fully. POCs provide valuable insights into the partner's capabilities and allow you to see how their solutions fit your business needs.

That said, it’s important to budget for POCs and be creative in how you run them. For instance, some businesses have used A/B testing platforms to build front-end POCs without involving their development teams. While developers may not always approve, sometimes innovative workarounds are necessary to move projects forward.

3. Conduct thorough due diligence on AI vendors

When selecting AI vendors, look for those who are transparent and have an open ecosystem. Modern eCommerce platforms are becoming increasingly complex, with omni-channel and varied use cases requiring integration across different systems. Vendors should offer flexible integration options, such as apps, SDKs, and libraries, to ensure their solutions can work with your existing systems.

Furthermore, consider vendors that use widely adopted technologies. This will give you the flexibility to switch platforms in the future without losing the work you’ve already invested. The goal is to select a partner who is not only suitable for today’s needs but can grow with your business over time.

4. Establish strong AI governance and compliance practices

Governance is a critical aspect of AI adoption. AI-driven tools need to align with the overall business strategy and broader operational goals, not just individual team requirements. Effective governance also involves compliance with data privacy regulations and ensuring robust guardrails are in place to protect your brand’s reputation. AI that lacks proper safeguards, such as chatbots that go off-script, can quickly damage customer trust and brand perception.

For example, if an AI tool pulls incorrect or embarrassing information, it can cause significant harm. To prevent such issues, it's essential to set clear boundaries on how AI is used and ensure teams are properly trained in its governance. AI should be implemented with a clear understanding of its capabilities and limitations, and businesses must take a careful, measured approach to avoid potential pitfalls.

5. Avoid common AI pitfalls in eCommerce

When integrating AI, it’s easy to get caught up in the hype and rush into projects. Here are four common pitfalls businesses face:

Value: AI must drive measurable value, such as cost reduction or increased profits.

Cost: AI is expensive to run in production, especially when dealing with large volumes of data or user requests.

Brand Protection: Without proper safeguards, AI could harm a brand’s reputation. Businesses should guard against issues like AI "jailbreaking" by ensuring their models have the appropriate protections in place.

Performance: AI performance goes beyond speed—it’s about how well the model understands the specific business problem.

AI models need to be fine-tuned, especially when using techniques like retrieval-augmented generation (RAG), where a base model is tailored with specific content. Moreover, businesses should avoid using resource-heavy models like ChatGPT-4 for every use case, as smaller, more specialised models can achieve similar results with less cost.

Many eCommerce businesses struggle with aligning AI initiatives with their overall business roadmap. One significant challenge is the lack of internal expertise in AI. It's essential to educate teams and stakeholders about the capabilities of AI, including training them to use tools like ChatGPT in day-to-day operations.

The data challenge is another critical hurdle. AI thrives on quality data, and organisations must have systems in place to handle and process data efficiently. Otherwise, they risk deploying AI solutions that underperform or fail to meet expectations.

6. Adopt sustainable AI practices

As AI technology evolves, so too does the demand for sustainable practices in coding and deployment. Sustainability in digital operations is becoming a growing priority. From optimising the brightness of images to reducing the energy used by backend systems, there are various ways businesses can reduce their carbon footprint. A key tool for measuring this is CO2.js, an open-source JavaScript product developed by the Green Web Foundation. This tool helps measure energy consumption on the browser level, factoring in elements like image size and font usage.

For those using cloud services like AWS or Azure, businesses can now track their carbon emissions through the cloud provider’s billing data. With EU regulations mandating scope 3 emissions reporting, companies will also need to account for the environmental impact of their supply chains. Sustainable design practices, such as using static site generation and optimised front-end design, are also emerging as standards. Microsoft and W3C have both published sustainable design frameworks to guide developers in reducing energy consumption through better coding and infrastructure practices.

Examples of successful AI integration in eCommerce

Many businesses have successfully implemented AI to enhance their operations. Here are some examples:

Search and merchandising tools: AI-powered search and merchandising tools have delivered impressive results for several companies. Astrid & Miyu, for instance, saw a 22% decrease in bounce rates, while Paul Smith experienced a 73% increase in search-led revenue. Similarly, Cambridge Satchel saw a 64% increase in conversion from AI-driven product recommendations.

SEO optimised content: AI tools have been used effectively for generating SEO optimised product descriptions, allowing brands to scale content creation at a fraction of the time it would take manually.

Customer service chatbots: Chatbots that answer specific customer inquiries—such as stock availability or delivery times—have helped reduce pressure on customer service teams while improving customer satisfaction.

Final thoughts

AI is undeniably transforming eCommerce, but businesses must approach its adoption with care. Thorough due diligence, careful vendor selection, proof-of-concept testing, and sustainable design practices are all essential to making the most of AI. By focusing on real business problems and ensuring that AI aligns with broader business goals, eCommerce brands can harness the full potential of this transformative technology.