Sportland’s eCommerce team had spent years building more than 100 recommendation blocks by hand, each one a careful guess at what a customer would want next. Then they switched on Bloomreach’s AI engine, and its recommendations converted 6x higher.
We covered this story in our recent panel discussion with Sportland and Bloomreach. Sportland’s marketing and eCommerce leads walked through their customer data platform, what it cost them in time and attention, migration to Bloomreach, and what they would do differently today. Read the rest of the article for the highlights, or here’s the recording if you prefer to watch:
Where Sportland started
Sportland opened its online store in 2020, during the first store closures, which the team itself calls late. The following years went into the eCommerce foundation, then customer experience, then performance, and new markets. Personalization only came up in the middle of that journey, when the email tool and the data behind it could no longer keep up with what the team wanted to do.
The data about customers was there, in spreadsheets, in the ERP, in the Adobe Commerce backend, and in the email platform. What was missing was any connection between them. And with five markets, Estonia, Latvia, and Lithuania plus smaller operations in Finland and Poland, every flow, web layer, and recommendation engine was built three or four times over.
One more thing a multi-market retailer should know. The biggest market has several times the customer data of the smallest, so personalization left to itself widens the gap between strong and weak markets. It has to be designed against, or the best work ends up where it is least needed.
CDP project goals
CDP was set to bring more revenue from email and a better return on ads. It would also improve automation, multi-market support, and behavior-based journeys. Email moved from scheduled newsletters toward messages triggered by what a customer does, running on their own. They were also going to build audiences from Sportland’s own customer data and push them to Google and Meta. Personalization was never the goal as such. A new customer, a repeat customer, someone into one brand or one sport, someone about to lapse, each should get different communication and a different share of the budget.
What a CDP does
A customer data platform brings everything known about a customer into one profile, from every channel and source. Marketing automation acts on that profile, deciding where a journey starts, what gets personalized, and when to stop.
In Bloomreach, the AI layer, Loomi, sits inside the platform and takes over decisions that used to be made by hand. It finds the best segment for a campaign from different combinations, writes copy in a set tone of voice, ranks products per visitor, and in what Bloomreach calls contextual personalization, it sends each customer the test variant that people like them responded to, even when that variant lost the overall A/B test.
The next step is agents. In the webinar, the demo showed an abandoned-search campaign, with scenario, emails, and dashboards, built from a single request. One Bloomreach client puts the time saved on campaign creation at around 92%.
Sportland’s CDP setup
| Area | Before | After |
|---|---|---|
| Paid media | Audiences inside the ad platforms | Audiences built from Sportland’s own data, refreshed daily and pushed to Google and Meta by rules |
| Website | More than 100 rule-based recommendation blocks, roughly the same products for everyone | Loomi personalizes every block per visitor, with fallback rules for visitors it cannot identify |
| Manual product recommendations | Automatic recommendations from the same engine, plus fully automated campaigns and flows | |
| Customer | Known one channel at a time | One profile that the site, email, and ads all read from and write to |
Follow one customer in Tallinn through it:
- The customer sees a Sportland ad for road running shoes on Instagram →
- On the homepage, they get what is popular, because they’re still anonymous →
- The customer creates a Sportland ID →
- Their anonymous history merges into their profile →
- From that point, product pages show alternatives in this customer’s size that are in stock, and search results are ranked for them specifically →
- If this customer leaves at payment, they get a reminder, and 9.6% of those reminders turn into an order →
- After delivery, comes a survey and picks chosen for this customer, with a cap on how often they hear from the brand →
- Two months of silence, and they gets the new season’s version of the shoes they bought, in their size and brand →
- A year on, Sportland still knows which group this customer belongs to and writes to them accordingly.
One campaign plan across markets
The campaign side did not get more complicated. Sportland works from one group-wide campaign plan, so the three Baltic markets usually run the same campaign at the same time across email, paid media, web, and stores. The copy is translated and the imagery is local, with each market’s own athletes and influencers.
The personalization is underneath – a running campaign carries one message, but a heavy runner and an everyday runner get different emails, recommendations differ per market, and the automated flows keep running regardless of the campaign calendar.
Results
- +62% more revenue on the same ad spend
- +24% orders from automated flows
- +24% conversion rate on personalized recommendations
- +9.6% of checkout reminders end in an order
- Automated flows are a small share of what Sportland sends and bring over 30% of its email revenue.
What Sportland would do differently
- Define the first few business problems before exploring what the platform can do. A platform this broad is easy to get distracted by, so pick a use case with a real customer problem, the data to act on it, and a result you will measure. Start small, test, understand why it works, then widen.
- Plan for ownership on day one, because the platform launched fast, but changing how the organization works with it took far longer than expected. A partner matters early, but the long-term goal is for nobody outside the company to know the setup better than the company does. Sportland is now moving from built for us to built with us.
- Get the architecture right before activation. Flows convert within two weeks, so everyone wants to launch them first, but how markets are separated, how identity resolves, and what gets tracked cannot be undone cheaply later.
A project like this runs in three phases, and each needs a different team:
- Build covers data mapping, integrations, migration, and testing, and carries the biggest decisions.
- Setup and run covers local coordination, reporting, optimization, and content, and for Sportland that is where scandiweb did much of the running in the first period.
- Control is where the internal team owns the platform end to end and calls the partner for specific things. Sportland now handles around 80% of the work internally, and the advice is to decide who will own that last phase before the build starts.
Want an outside view on your own personalization setup? Get in touch with a CDP strategist and book a free discovery and data audit.
