Many e-commerce companies invest heavily in personalisation engines, expecting AI-driven recommendations, dynamic content, and tailored experiences to drive higher conversion rates. But often, after months (or even years) of development, the results are underwhelming—engagement remains stagnant, revenue uplift is marginal, and customer frustration grows.

So, what’s going wrong?

We spoke to CTOs, CDOs, and Heads of Engineering who have been through this journey. Their insights reveal common pitfalls and critical lessons for turning struggling personalisation initiatives into real business impact.

1. The problem isn’t your AI—it’s your data

Many organisations blame their recommendation engines when personalisation falls flat. The reality? Bad data is the real bottleneck.
Common issues we’ve seen are:

• Data silos – Customer data is scattered across platforms, preventing a unified personalisation strategy
• Duplicate and inconsistent records – A customer might be treated as multiple different users across touchpoints
• Lack of real-time data – Personalisation models rely on outdated batch-processed insights, leading to irrelevant recommendations.

Lesson: Personalisation is only as good as the data behind it.

2. The tech is built, but the business can’t use it

One CTO shared how their engineering team built a personalisation engine, but marketing and product teams couldn’t easily access or test recommendations. The result? Underutilisation and minimal impact.

Common missteps are:

• No business-friendly interface – Marketing teams need low-code/no-code tools to adjust personalisation strategies without engineering dependencies.
• No A/B testing framework – Without proper experimentation, it’s hard to measure whether personalisation is actually working.
• No clear ownership – Is personalisation owned by engineering, data, marketing, or product teams? Often, it’s unclear.

Lesson: Personalisation needs cross-functional collaboration, not just a strong tech stack.

3. The infrastructure can’t handle scale

A CDO at a leading e-commerce retailer described how their personalisation model performed well in a test environment but crashed under real-world traffic. The culprit? Bottlenecks in the data pipeline and inefficient query execution.

Common scalability pitfalls are:

• Recommendation engines querying live databases instead of using precomputed embeddings
• Inefficient data pipelines causing latency in real-time personalisation
• Infrastructure not optimised for high-traffic events (e.g., Black Friday).

Lesson: A system that works in a pilot phase can still fail in production if scalability isn’t considered.

4. Lack of flexibility stifles innovation

Several CTOs we spoke to described how early technology choices limited their ability to innovate later. Many personalisation platforms were designed with specific use cases in mind but couldn’t adapt as customer expectations evolved.

Signs of an inflexible system:

• Personalisation logic is hardcoded into the platform, making it difficult to test new models.
• Scaling personalisation globally requires expensive rewrites due to tightly coupled architectures.
• Integrating new AI models or third-party data sources is slow and complex.

Lesson: Rigid architectures and legacy data systems slow down personalisation efforts.

4. ROI expectations are misaligned

Many CDOs highlighted a disconnect between leadership expectations and real-world personalisation ROI timelines. Executives expect immediate revenue lift, but in reality:

• Personalisation takes time to optimise—data models need continuous tuning
• Success metrics must be realistic—personalisation might improve customer lifetime value (CLV) more than immediate conversions

Long-term investment is key—AI-driven personalisation isn’t just about quick wins but creating sustained differentiation.

Lesson: Personalisation is not a one-time implementation—it requires ongoing optimisation.

Final takeaways for CDOs and CTOs

If your personalisation engine isn’t delivering results, the problem isn’t the algorithm—it’s the data, infrastructure, and operational alignment.

• Fix your data first—invest in real-time, high-quality, unified customer data
• Make personalisation accessible—give marketing and product teams the tools to optimise and measure results
• Design for scale—ensure your infrastructure can handle real-world e-commerce demands
• Future-proof your architecture—adopt a modular, flexible data infrastructure that allows personalisation strategies to evolve without costly rework
• Align ROI expectations—personalisation is a long-term investment, not a quick win.

By applying these lessons, e-commerce leaders can turn underperforming personalisation into a true competitive advantage.

Want to accelerate your data readiness for personalisation? See how we can help.