Home » What we do » Data solutions » Data science

Data science.

Data for next-level targeting and optimisation

A customer’s value, their propensity to behave in a certain way, and many of their other attributes and categorisations take time to manifest. But when we train a machine learning model to accurately predict these, at the moment the customer is first acquired, these data points become game-changing for audience targeting, bidding and optimisation. Machine learning has a lot more to offer: visual recognition techniques, for example, can audit ad creative at scale, and provide data for machine-driven creative design direction.

Our approach

Machine learning is more than enabling predictions. Wherever possible, we develop models that reveal insight too: we attribute predicted customer value, for example, to each dataset variable. In developing models, we’re able to expand datasets by using visual or speech recognition to transform ad creatives into structured data. We also emphasise feature engineering, where variables are combined or transformed in order to improve the performance of modelling, and arrive faster at an effective model. This is where our hands-on digital advertising knowledge sets us apart.

The Croud difference

Your data is only as valuable as your ability to activate it. As an independent advertising business with market-leading expertise, Croud can maximise its value. We deliver real-world effectiveness.

We’re marketing data natives: we understand the practical application of every feature in marketing data, so we build effective models faster. We use tools including Tensorflow, scikit Learn, Google Cloud Platform, OpenCV, AI Platform, SHAP, and Streamlit. Tim Collier – Head of GMP Partnerships, UKI, at Google – wrote about our Creative Intelligence product, calling it, “a cutting edge use of data technology.”

Our areas of expertise

LTV modelling

A single acquisition misrepresents value: real value is loyalty. We’ll analyse LTV, and enable predicted metrics for audiences and optimisation.

Creative Intelligence

Our proprietary tool machine-directs the creative design of Display assets. It also predicts the performance of new creative before launch.

Propensity

We’ll predict user behaviour, whether they’ll convert, or what they’ll add-to-basket next. It will power personalisation, product recommendations, and remarketing

Customer segmentation

Your customer data may include hundreds of attributes. Clustering customers will reveal segmentations, for use in audience and content strategy.

Automated categorisation

Machine learning can categorise language and images at scale and speed. Typically, our automated categorisation supports SEO and content planning.

Google Sales Partners

Get in touch

Learn more