From Data Chaos to Confident Decisions: How Retailers Are Rethinking Data, AI and Attribution

Ed Lawson: There is no shortage of data in retail today, yet many organisations still struggle. What are the clearest signs that something is not working?

Lee McCance: It tends to show up in how people behave rather than in the data itself. When teams genuinely trust their data, they move quickly and make decisions with a sense of confidence. You see a clarity in how they operate. When that trust is not there, everything slows down. People start questioning the numbers, checking them repeatedly, and asking for validation from different sources. Over time, that creates hesitation across the organisation. It affects not just campaign decisions but also longer term strategic thinking. Instead of focusing on what to do next, teams get stuck trying to prove that what they are looking at is even correct.

Praveen Kumar: From an organisational perspective, it becomes very obvious when there is no alignment. Different teams begin reporting different figures for what should be the same performance metrics. Marketing may present one version, finance another, and analytics something else again. At that point, you are no longer discussing performance or growth, you are debating the numbers themselves. Without a shared and trusted version of the truth, it becomes extremely difficult to run the business in a coherent way.

Ed Lawson: AI has become a major focus in recent years. Has it helped address these challenges, or has it made them more pronounced?

Lee McCance: It has certainly made them more visible. AI is incredibly powerful, but it relies entirely on the data it is given. If that data is flawed or inconsistent, the outputs can appear very convincing while still being wrong. That is what makes it risky. You end up in a situation where decisions could be based on insights that feel authoritative but are not actually reliable. In that sense, AI has raised the stakes. It has not introduced a new problem, but it has made it much harder for organisations to ignore the importance of getting their data right.

If the foundation is not solid, then AI cannot deliver consistent or meaningful results. So in many ways, it has reinforced the importance of fundamentals.”

Praveen Kumar: There has also been a shift in how organisations are thinking about AI. Initially, there was a lot of excitement and experimentation, but now the focus is moving towards real outcomes and measurable impact. Businesses are asking what value AI can actually deliver. The answer always comes back to the same point, which is the quality of the data underneath. If the foundation is not solid, then AI cannot deliver consistent or meaningful results. So in many ways, it has reinforced the importance of fundamentals.

Ed Lawson: Marketing data is often described as particularly difficult to manage. Why is that the case?

Praveen Kumar: The main challenge is fragmentation. Marketing data comes from a wide range of sources, including different platforms, channels, and partners. Each of these sources has its own structure, its own definitions, and its own way of reporting performance. Bringing all of that together into a single, coherent view is not straightforward. On top of that, ownership is often unclear. Multiple teams contribute to the data and rely on it, but no single team is fully accountable for its quality and consistency. That lack of ownership creates gaps, and those gaps are where problems tend to emerge.

Lee McCance: Even when organisations manage to bring the data together, there is another challenge, which is usability. Data can be technically accurate but still not very helpful if people cannot easily access it or understand how to use it. The real objective should be to enable better decisions. That means presenting the data in a way that is clear, actionable, and relevant to the people who need it. Otherwise, it remains something that is observed rather than something that drives change.

Ed Lawson: How did you approach these challenges in practice?

Praveen Kumar: The starting point for us was to step back and clearly define the problem we were trying to solve. Before that, different teams were using their own tools and generating their own sets of numbers. That led to a great deal of confusion and a lot of time spent aligning rather than acting. We focused first on creating a consistent structure by bringing the data together into a unified framework. That gave us a shared foundation across the organisation. Only once that was in place did we look at tools and automation. Taking that approach helped us avoid adding unnecessary complexity and instead created clarity and alignment.

Lee McCance: That sequence is important. There can be a temptation to start with the tool, but without a clear understanding of the problem, that often leads to more fragmentation rather than less. The real goal is not simply to centralise data, but to make it genuinely useful. That means moving beyond reporting and towards optimisation, where the data actively supports better decision making across the business.

 ownership is often unclear. Multiple teams contribute to the data and rely on it, but no single team is fully accountable for its quality and consistency.”

Ed Lawson: Attribution remains a persistent challenge for many organisations. How do you think about it?

Praveen Kumar: Attribution is not something that you complete and then move on from. It is an ongoing process that needs to evolve over time. Customer journeys are increasingly complex, and they rarely follow a linear path. A single purchase can involve many different interactions across channels, often over an extended period. Because of that, any model you build needs to be flexible and open to refinement.

Lee McCance: Traditional approaches, such as last click attribution, do not really reflect how decisions are made in reality. They tend to oversimplify what is actually a very nuanced process. Rather than searching for a perfect model, organisations should focus on developing an approach that reflects their own customers and their own channels. That involves forming hypotheses about what drives performance and then using data to test and refine those assumptions over time. There is no universal answer that works for everyone.

Ed Lawson: What role does first party data play as these challenges become more complex?

Lee McCance: First party data is becoming increasingly valuable because it provides direct insight into customer behaviour. The signals are typically stronger and more reliable because they come from your own interactions rather than from external sources. As the broader data landscape becomes less predictable, that direct relationship with the customer becomes a significant advantage.

Praveen Kumar: Organisations that have invested in building and understanding their own data are now in a much stronger position. They have more control over the quality of their data and more confidence in how they use it. That makes it easier to adapt to changes in the market and to make informed decisions.

Ed Lawson: Finally, what does success look like in this area?

Praveen Kumar: For us, success is about creating clarity, speed, and alignment. We want teams to be able to make decisions quickly, based on data that everyone understands and trusts. That reduces friction and allows people to focus on improving performance rather than debating the numbers. It also means less reliance on manual processes and a more streamlined way of working across the organisation.

Lee McCance: Ultimately, it comes down to confidence. When people trust the data, they are far more likely to act on it.

Rather than searching for a perfect model, organisations should focus on developing an approach that reflects their own customers and their own channels”