At Lumav, we believe that ensuring high-quality product information is currently one of the highest-impact IT investments an e-commerce business can make.
It is encouraging to see that, alongside conversations about using artificial intelligence in e-commerce, more and more client discussions now begin with the question: “How can we get our product data in order?” This is a positive development, because AI is not a magic solution — it is an accelerator. Poor input data will inevitably lead to poor output.
However, AI is no longer only available to merchants. It is also in the hands of customers.
Consumers are often irrational and convenience-driven, and commerce has always been built around that reality. In the customer’s hands, artificial intelligence becomes a rational and well-informed shopping assistant that helps them make the best possible decision based on facts.
The e-commerce store of the future must therefore be built for both people and machines.
Bots, Crawlers and Product Data
Online stores and product pages are increasingly being read not only by people, but also by bots — Google, AI tools, shopping assistants, comparison engines and other automated systems.
For a human visitor, the information on a product page helps determine whether the product or solution meets their needs. For a machine, that same information is used to extract facts, connect them with other data points, place them into a wider context, compare alternatives and ultimately translate the result back into the customer’s language.
In practice, bots do what we would do ourselves if we had more time, worked faster and had access to more knowledge.
It is possible to make an online store more machine-friendly. In addition to technical configuration, this also requires a new mindset around product data enrichment. For example, a person may be able to infer missing “obvious” information, but a machine usually cannot.
Let’s begin with search — the same place where most online shopping journeys start.
Search in the E-Commerce Store of the Future
Swedbank recently invited us to speak at the e-commerce seminar “From Click to Experience”, where Lumav CEO Silver Kallas shared a vision of what search in online stores may look like in the future. More about this can be read in the summaries published on the Swedbank blog and on kaubandus.ee. One conclusion was clear: effective search once again starts with structured and high-quality product data.
Product data enrichment means making as much relevant product information available as possible — in other words, ensuring that primary product data is accessible and usable. Product data expansion means deriving new information from existing data, or in statistical terms, creating secondary data. This creates additional product knowledge and helps the search engine better understand what the customer is actually looking for.
Neither people nor machines necessarily use the same keywords that the customer enters into the search bar.
The key question is no longer only which keywords the customer uses. The more important question is how the customer searches.
More and more shopping journeys now begin in an AI interface. At the same time, online stores are not disappearing — and neither are physical stores. But imagine a younger customer who has grown up using GPT-style prompts and AI search, and then suddenly has to use your online store’s search function, which only understands traditional product-based queries.
Does your online store search understand the customer’s intent?
The Cost of Intent-Based Search
The answer is intent-based search.
It is worth addressing costs immediately, because many AI projects that start with enthusiasm fail due to insufficient cost control.
There are vendors on the market who claim they can make an online store’s search self-learning for a fixed monthly fee. In practice, this is often based on trial and error — and the test environment becomes your own search box. AI-powered search, meaning the integration of artificial intelligence directly into the customer-facing e-commerce interface, can have unpredictable query-based costs. Without proper control mechanisms, it may quickly lead to unexpectedly high invoices.
When using Lumav’s internal AI search tools, it is not uncommon for a highly useful query to cost 1–2 euros. Experience shows that if a tool is genuinely useful, people will use it more — and costs will increase accordingly. In one client’s case, the search solution became so effective that employees of competitors started using it as their own internal work tool. That is not a cost any merchant would want to carry.
Enriched and expanded product data can be indexed and governed by rules, so the cost of each search query does not depend so heavily on artificial intelligence. The same principle can also be applied to image search by training and enriching an image database.
In short, improving your e-commerce search engine does not necessarily mean that every search must trigger a paid AI query. It is possible to build a unique, controlled and scalable system of your own. Over time, this can become a strong competitive advantage.
Future-Proof E-Commerce Is Built on a Knowledge Base
A large part of the most valuable knowledge in today’s companies is not stored in the ERP, the online store or the marketing platform. It is still stored in people’s heads: how to solve a customer’s problem, which product to recommend, which products work well together and which exceptions apply.
It is true that data can be copied. However, not all information needs to be visible to Google, bots or end customers immediately. Information can be disclosed selectively, at the right time and in the right context, in order to provide a better customer experience.
An e-commerce knowledge base can be expanded beyond enriched product data to include the company’s best practices. In essence, it becomes an always-available senior specialist — working 24/7, not getting sick, not resigning and always ready to answer questions.
We expect that investing in data and building a knowledge base will become an increasingly important factor in assessing the value of a company.
Interested in the topic? Let’s meet for coffee and discuss it further.
* * *
Customer data in e-commerce in the age of AI – a separate topic for another time. It deserves a more detailed analysis, because it involves both ethical and legal considerations.
The greatest opportunities are likely to be found in B2B sales and customer service, where compliance requirements are generally less restrictive and the focus is on understanding businesses rather than private individuals. Customer data can also be enriched, and eventually, like a virtual matchmaker, the right products can be matched with the right customer at the right time — and the process can be repeated.
