Uniting data to better understand your customer
Omoda is a family-owned business with a passion for shoes since 1875. As you can imagine, back in the 1900s, shopping was a very different experience than it is now and revolved around the in-store experience. However, nowadays the journey of most consumers has changed to an omni-channel journey with touchpoints both offline and online. This makes it difficult to reach and connect with one’s target audience like the old days. Thus, Omoda turned to Dept to help them unify their data sources to better understand their current consumers whilst reinventing their search marketing strategy to help the brand grow their business.
Navigating the current digital ecosystem
The shoe industry is very competitive and demands the knowledge, experience and personal touch that are distinctive to Omoda, allowing them to remain a top player. But guaranteeing this can be challenging due to the current digital landscape and with so much data now available, it can be overwhelming for brands. So Dept stepped in the help Omoda unify their data from various data sources to create multiple consumer personas by implementing machine learning algorithms. At the same time, we helped the company to improve its search marketing strategy.
Unifying data sources to better understand the customer and their needs
Unifying online and offline data
In order to gain a complete view of the brands (potential) customer, it was important to know what they do both online and offline. Therefore, the first step of this process was to centralise the data of the retailer and go from intangible and separated customer views to a complete and scalable customer view. To make this a reality, we used Google Cloud Platform as the main Customer Data Platform (CDP) where all the data sources come together. We combined data from Google Analytics 360, point-of-sales databases and customer data from Omoda’s Customer Relationship Management (CRM) system. The platform also refreshes daily to provide the company with the most up-to-date results. Thus, the CDP offers Omoda a continuously growing database containing more than ten million consumer profiles, representing both visitors and customers and their behaviour online, offline and in terms of returns. This allows the brand to see a complete overview of its customer instead of just one part.
Our next step was to enrich the data with insights that could only be gathered by using machine learning techniques. Thus, we applied a k-means clustering algorithm that was able to find natural existing groups and offer Omoda groups of customers that behave equally across the touchpoints in their customer journey. The k-means clustering algorithm gave us information on the optimal number of groups, which should be optimally different from each other and similar within. Based on the results of the clustering algorithm, we interpreted the groups by analysing the numbers and the differences in scores across all the features.
As a simplified example, we observed that the median of cluster one for the number of offline transactions was much higher compared to all the other clusters. This indicates that these customers are orienting online and buying offline, making them Omni-channel shoppers. This approach resulted in personas like Loyalists, Omni-channel shoppers, Window shoppers, Heavy Returning shoppers and Family shoppers. All with their own patterns in terms of behaviour, assigned to every individual customer or visitor of Omoda, present in the CDP database.
So, with the current set-up, Omoda can predict in real-time for each individual visitor which persona is the most suitable for communication and targeting within the search campaigns, social ads, newsletters, but also the website can be adjusted accordingly.
Applying these persona’s to search engine marketing
In order to use the personas to adjust marketing communication and the website in a scalable way, we needed to add one more building block. This consists of a predictive TensorFlow model that we applied on the enriched data, containing the personas per customer and their behaviour. We taught it to predict a persona for new inputs. Since these predictions are available on CDP, it enables both SEA and CRO specialists at Omoda to start adjusting marketing expressions and website based on a data-driven 360-degree persona.
Additionally, because there is a view of the supply and returns of Omoda via the CDP, it has been decided to make the search campaigns completely feed-based. This means that all keywords and advertisements are created dynamically and paused based on the current Omoda offer. As a result, the advertisements have become more up-to-date and relevant, as a result of which the CTR has increased by more than 25% (average 16% on search ads) and the CPC has fallen by more than 25% compared to the same period last year.
Our team unified all data sources thus allowing to connect offline shopping behaviour to online site behaviours. Through data centralisation via the CDP, Omoda knows its customers better than ever before. Automating campaign management has given the brand more control over their advertising investment, enabling them to improve performance faster. They are now able to cater to consumers needs in a more personalised manner and tailor their brand message depending on the persona they are communicating with.