Customer case · Høie

From gut feeling to machine learning.

Høie is one of Norway's oldest bedding brands. Marketing ran on experience, last-click ROAS and instinct. Now the budget follows a model that gets checked against actual sales every month.

Book a demo Read the story

Industry: Retail, home textiles
Market: Norway
Model: Retrained monthly

Two colleagues in conversation at a table in a Nordic office — +28% revenue growth in a key month, at the same ROAS

Illustrative photography, not a photograph of the Høie team.

The challenge

Three questions that kept coming back.

For years the marketing decisions were built on a mix of experience, last-click ROAS and gut feeling. The team was good at their job, but the reports could not settle the arguments.

Which channels drive total sales, not just the last click before purchase
Where the ceiling sits: when a channel stops paying for the next krone
How the budget should be split for the highest total return

What we built

One model across every channel.

Odins collects and structures Høie's marketing data across all channels continuously, and the marketing mix model is retrained and tested every month. Forecasts, scenarios and channel effects are shown plainly, so they are easy to discuss internally and with agencies.

Budget and channel scenarios simulated in minutes
Channel effects measured on total sales, not the last click
Forecasts verified continuously against real sales data
A living decision tool, not a static report

The results

What changed, in numbers.

+28% Revenue growth in a key month Spend increased on the model's recommendation, and ROAS held at the same level.

Monthly Model retraining Forecasts are made before the outcome is known, then checked against actual sales.

One Model across every channel Digital and offline measured together, instead of channel by channel.

From gut feeling to machine learning.

Three questions that kept coming back.

One model across every channel.

What changed, in numbers.

The same clarity and control, on your numbers.