Contents
Summary
- Good data quality is necessary – but it does not create a good product experience on its own.
- PIM solves the problem of making product data manageable. PXM solves the problem of making decisions possible for customers.
- A stable data foundation is no proof of a good product experience.
As I write this first blog article, we find ourselves in a time of ever-accelerating "progress". Everywhere – in the media, at work and in private life – AI dominates the headlines.
The impact of AI on experiences involving product data is already noticeable, too. Whether in the form of artificially generated texts in online shops or obviously AI-generated menus in restaurants. The topic surrounds us around the clock.
In my first article, however, I want to take a step back. Back to a time when automation was a topic, but not yet the automation of our entire existence. So before thinking about which processes to hand over entirely to machines, we should ask ourselves which goals we are actually pursuing.
Just because data is complete and processes run fast does not mean that this creates any value for others.
Good data quality is necessary – but it does not create a good experience on its own.
Take the eBay platform as an example. In the selling process, sellers are offered the option of having the description text generated by AI. When the feature was introduced, the AI only created the description text. The results were of an interesting nature. The texts felt artificial and exaggerated. Information was often added, while other relevant information was left out.
From the customer's perspective, too, such texts were hard to read. After the first few words it was clear that they had been machine-generated, and the embellishments suggested it would take ten minutes to filter out the relevant information.
If we now transfer this example to a PIM system, we can see that the following things could mistakenly be treated as equal:
- Data completeness and relevance
- Correctness and comprehensibility
- A clean product model and real decision support
In our eBay example, the product description is maintained and complete for our target channel. However, it has a negative impact on the shopping experience. We are annoyed by the wording and also need a lot of time to get to all the relevant information.
Both factors have an immense impact on our shopping experience. We don't want to shop when we are annoyed, and we want to get to all the information we need for our purchase decision without friction.
Platforms such as PIM systems already achieve a great deal and support us with a wide range of optimization topics. They make our product data manageable, create order and consistency and reduce internal complexity – provided they are used correctly. Without this foundation, a good product experience would hardly be conceivable.
What they do not do, however, is decide for us which information is actually relevant to a purchase for us as customers. Nor do they determine how this information is prioritized and how customers are guided through the flood of information. And ultimately, they do not answer the question of what value the data delivers in the end. We have to make all these decisions ourselves – from the perspective of those who are supposed to make a purchase decision based on the data.
Common misconceptions at the decision-maker level
In my work with clients, I keep coming across the following:
"Let's introduce the PIM system first, and after go-live we'll take another look at the customer experience." This is often encouraged, or even provoked, by short project timelines.
At the decision-maker level, you should therefore deliberately ask yourself before the project starts: "Where do we actually want to go?" It rarely pays off to start running first and set the destination afterwards (unless you deliberately want to explore).
In conversations, statements like these come up again and again:
- "We have product data under control, it is no longer a bottleneck."
- "We have invested heavily in PIM – the foundation is in place."
- "If something isn't working, it's more likely the channels or the UX."
- "More structure automatically creates more clarity."
None of these statements is wrong in itself, but they often imply the view that with a clean data foundation, the work on the customer experience is already done or will take care of itself. This data foundation, however, is merely the prerequisite for focusing on product experiences. It does not do the work of designing the product experience.
To stay with the eBay example: from a data quality perspective, everything was right. The product attribute was filled in and the channel was supplied with the corresponding information. For us as customers, however, the text got in the way rather than helping us.
Where good PIM work reaches its limits
PIM systems are designed to manage product data. They are meant to grow with an expanding assortment and additional output channels, and to ensure that our processes remain efficient.
Good customer experiences, however, do not emerge at this point. They emerge where product data gives us context, enables comparison and prioritizes information accordingly, guiding us elegantly through the purchasing process.
A gap often opens up between these worlds, and it widens in day-to-day project work when focus is lost. What matters is the point at which our success is measured. Our insights into data quality in PIM systems rarely reveal how that data performs in the buying experience.
What decision-makers should take away
Let's return to my thesis from the beginning of the article. Before we consider which tasks we can hand over to AI, it is worth asking: "What experience do we actually want to create with our data?"
For decision-makers, it should be clear that good data structure and data quality merely form the foundation. A good experience is defined by how the data is used and prepared. Looking from the outside in is therefore recommended.