.jpg)
But transitioning to a genuinely data-driven model isn’t merely technical. It’s a complete cultural transformation for a business that requires moving away from 'we've always done it this way' and toward a culture of continuous experimentation, where even the most logical assumptions are put to the test.
In this article, we provide a practical examination of what that transformation looks like, grounded in the real experiences of two major Nordic retailers who are living it.
The 4 Foundations of a Data-Driven Organization
Building a data-driven culture is a multifaceted challenge that requires alignment across several pillars of the business. Skipping a step is a reason why so many data initiatives ultimately fail. The four steps to becoming a data-driven organization are:
1. Break Down the Data Silos
A common hurdle in large organizations is that data often ends up in silos. Purchase data sits in one system, marketing data in another, and technical performance data in another. Ask any digital commerce professional whether they recognize this pattern, and most will likely say yes.
To gain a holistic view of the customer journey, teams need to see how different touchpoints interact, where customers drop off, and what happens before and after each engagement moment. It’s simple: data that cannot be connected cannot be acted on.
2. Secure Executive Buy-In
For a data-driven approach to really take root, it must be supported from the top down. If a CEO does not advocate for data-backed decision-making, reports will often fall flat and fail to influence actual business strategy. When leadership actively advocates for evidence-based decision-making, it changes the culture of the entire organization and empowers employees to see themselves as contributors to a shared standard of proof.
3. Democratize the Data
When a single analyst is responsible for all reporting, efficiency is easily lost. Lowering the barrier to data access changes this dynamic. When UX designers, growth specialists, production managers, and e-commerce leads can each interpret data relevant to their own work, they take ownership of their metrics in a way that top-down reporting never achieves. Each person owns their delivery and can make informed adjustments without waiting for a centralized report.
The goal is not to turn everyone into an analyst. It is to give everyone enough visibility to ask better questions.
4. Build a Two-Stream Workflow
Organizations that use data effectively tend to adopt a “two-stream” approach:
-
The short-term stream: a weekly rhythm, involving several different teams reviewing real-time insights and making immediate seasonal or tactical adjustments.
-
The long-term stream: strategic meetings where teams come together and use data to prioritize major site improvements.
These two streams serve different purposes and involve different people. Combining them by trying to do strategic planning in a weekly operations review or using long-cycle data to make time-sensitive calls can reduce the value of both.
Case Study: Home Furnishing Nordic (HFN)
Home Furnishing Nordic (HFN), the Nordic brand behind major brands including Trademax, Chili, and Furniturebox, has integrated data so deeply that it now drives its entire development roadmap. The company’s journey illustrates what happens when an organization fully commits to evidence-based development from the top down.
Building a CRO Team
The most consequential decision HFN made was establishing a dedicated CRO team consisting of analysts, growth experts, and technical staff. The common goal across the organization was to prioritize all development based on data-driven insights.
The CRO team brings together an analyst, growth specialists, and technical staff who work through a structured process: analyzing site data for anomalies and drop-off points, forming hypotheses, building test variants, and measuring results in terms of Revenue Per Session (RPS). Only “winning” tests that show genuine uplift progress to full development.
This matters because most ideas turn out to be wrong. HFN's early hypothesis was that roughly one in three tests would be a winner. In practice, fewer than half of all tested changes yield positive results, meaning that a substantial portion of development effort in organizations that don't test is simply wasted.
A 34% Conversion Rate Increase
By strictly adhering to a data-backed strategy and measuring everything through Revenue Per Session (RPS), HFN has seen a 34% increase in overall conversion rate. This number represents the cumulative impact of a disciplined, systematic process applied consistently over time, and it makes the case for investment in data infrastructure far more powerfully than any theoretical argument.
Case Study: Lindex
For Nordic fashion retailer Lindex, data has become the ultimate tool for resolving internal conflicts and reacting to the rapid shifts of the fashion industry.
The Benefit of Using Behavioral Data
One of the most practically impactful shifts at Lindex has been the use of behavioral data (most specifically visual session data and interaction map) to resolve internal disagreements.
The homepage is an everlasting battleground in retail organizations. Marketing wants to promote a current campaign; buying teams want their category front and center; the e-commerce team has a view on what converts, etc. But without data, who wins the conversation comes down to seniority and persistence.
With behavioral data, on the other hand, the conversation changes. Showing a purchasing or marketing colleague visually exactly how customers move through a page, where they stop, what they ignore, and where they abandon, produces a different type of discussion. It becomes harder to argue for a placement that the data shows customers are simply not engaging with
The Two-Stream Model in Practice
Lindex has applied the two-stream model to its data workflows:
-
The company has weekly sessions in place that bring together purchasing, paid media, and channel teams to review current performance data and make immediate adjustments. This makes it easier to respond to what is happening in the season right now.
-
Longer-cycle working meetings are also part of business, where developers, product owners, UX designers, and e-commerce leads bring together behavioral data and business priorities to form hypotheses and create a prioritized development backlog.
Separating these two rhythms prevents the company’s short-term issues from getting in the way of long-term priorities.
Automation and “The Segment of One”
Looking ahead five years, the manual analysis we see today will be largely replaced by hyper-automation. We are moving toward a "segment of one", a future where AI-driven systems run thousands of parallel tests to create a personalized experience for every individual user in real-time.
This evolution doesn’t mean that human roles will disappear. However, they will change. Instead of performing manual work, professionals will have to focus on high-level strategy and interpreting the complex signals provided by automated systems. In this new landscape, the ability to build and lead a data-driven organization will be the most valuable skill in e-commerce.
Organizations that have already built data cultures where experimentation is normal, where multiple roles can interpret data, and where the infrastructure for measurement is in place, will be able to absorb this shift more easily than those starting from scratch.
The Final Word: Culture is Everything
Both case studies in this article show the same underlying pattern: organizations that invest in data foundations, testing disciplines, and a measurement culture not only see significant improvements but also structural advantages that emerge over time.
The best results are the outcome of small, validated improvements made possible by an organization that has built the habit and infrastructure for continuous experimentation.
The retailers that will succeed the most aren’t the ones with the largest budgets or the most advanced AI tools today. They’re the ones who have made the largest cultural change internally, building teams and processes that know how to learn from data, challenge their own assumptions, and act on what the evidence shows.
