Paweł Kurek is the founder and CEO of the Polish outdoor furniture and garden equipment company Meblobranie.pl. He has nearly 20 years of experience in the e-commerce industry and focuses on driving growth through digital innovation and data-driven marketing strategies.
Our outdoor furniture business used to be at the mercy of the weather. A sunny forecast could send sales soaring, while a week of rain could bring them to a standstill. Adding to that challenge, we face fierce competition from large, established brands with physical stores where customers can try the products, while we operate exclusively online.
We invested heavily in digital advertising and saw an increase in clicks, but sales weren't keeping pace. To spend our ad budget smarter, we needed to stop guessing where our value came from and start measuring real business impact.
To get a true picture of our marketing performance, we partnered with a marketing agency digitalSharks to build a Marketing Mix Model (MMM). In simple terms, an MMM is a data-driven analytical approach that shows how different marketing channels — along with external factors like weather, seasonality, and competitor actions — contribute to sales. Here is how we discovered a powerful mix of media channels and used control variables to add to our creative strategy.
Unlocking the true picture of business performance
Using Google's open-source MMM, Meridian, we embarked on a three-month journey to build a model tailored to our business. We fed it data from all our advertising platforms, encompassing Google Ads (Search, Performance Max, Demand Gen), Display & Video 360 (YouTube, Rich media creative), Meta Ads, and Microsoft Ads.
But we didn't stop there. We layered in crucial external variables, like daily temperature data from key Polish cities and the timing of public holidays, to control their impact on our business performance.
The model’s building process involved constant refinement: our initial model included around 140 variables, which we then reduced to 35. We would build one model, test its accuracy against historical sales data, and then adjust the variables to improve its predictive power.
Discovering the value of awareness campaigns
One of the first and most profound insights from our MMM was the true value of our brand awareness campaigns. In a standard approach that assigns all credit for a customer conversion to the final touchpoint, channels like YouTube often appear to underperform because they don't typically drive immediate sales. Our model, however, told a different story.
It revealed that our YouTube campaigns were playing a critical role, creating initial interest and desire that would later be captured by other channels. The MMM was able to quantify this additional value, giving us the confidence to continue investing in building our brand through video.
The goal is to help our marketing operate even more efficiently — and that is exactly what I love about MMMs.
Another benefit was determining the optimal capping for our ads, particularly on YouTube and programmatic campaigns that included rich media creatives.
“Our marketing mix model helped us define the right capping level, said Aneta Krystosik, partner at digitalSharks. “This was essentially telling us the optimal saturation point for our ads.”
On top of that, the model didn't just evaluate channels in isolation; it revealed their synergies. We discovered a powerful combination that became the engine of our growth.
YouTube created the inspiration, while Search captured intent when people started looking for terms like "garden sofas" or "best outdoor dining sets". Performance Max then worked across all Google's channels to reach those high-intent shoppers and guide them toward a purchase.
Using dynamic ads to urge a purchase
To isolate the external impact on sales from media-driven effects we initially used the weather and day-of-week variables primarily as control variables within the MMM. However, this modelling process inspired a powerful creative strategy.
We began using real-time weather forecasts to create dynamic ads. If it was raining on a Wednesday but the weekend was forecast to be sunny, our ads would automatically update with messages encouraging people to order now to enjoy the upcoming good weather.
Here is one example: the ad on Google Display Network shows the current rainy weather and a sunny forecast in two days, saying that if customers buy now, they can get a discount.
Refining the model to improve efficiency
Our data-driven approach transformed our entire marketing strategy. We started spending our budget more strategically — we knew how to allocate it across specific campaigns and channels that effectively built brand awareness while maximising the overall return on investment. By spending our budget smarter and using a new approach to our creative ads, we achieved 46% year-over-year revenue growth and a 66% increase in orders.
The main lesson for us was that building a sophisticated model isn't a one-time project, it's a commitment to a new way of thinking.
“You need time and passion to do this, but if you build a good model, it’s worth the effort,” Dawid Krystosik, head of analytics at digitalSharks said.
Based on a year of experience, we are continuing to refine the model's data and explore additional variables. The goal is to help our marketing operate even more efficiently — and that is exactly what I love about MMMs.
Our next strategic goal is international expansion, and the model is central to that plan. Instead of entering a new market with guesswork. With such complex user behaviors and the multitude of channels through which we can reach them, it’s challenging to efficiently allocate the budget and set the right ad frequency on our own.
Today, MMM serves as our guide — helping us identify channels that may not seem profitable at first glance but are essential to achieving the best overall return on investment for the business. We’re no longer just reacting to the weather; we’re turning it into an advantage.
