Let’s meet Manuel, digital campaign manager!
Manuel is a JOT account manager who is optimizing campaigns for a client at global level. He is very skilled in analyzing data, identify the key indicators and implement optimization strategies to spend the budget efficiently. However, due to the number of variables (clicks, impressions, conversion rate, average position, date, time, device, platform, …) and the disparity of performance depending on the keyword category (travel, finance, home and garden, energy, sports, leisure,…) and country (above 20 for each client), Manuel hardly achieves the objectives defined by the clients in terms of visits and impressions.
He realized that there exist several external conditions that strongly influence the interests (queries) that society search in the main platforms like GOOGLE and BING. For example, events like Saint Valentine’s day, Christmas, Easter and the weather (both seasonal variations and weekly forecast). These initial assumptions have been confirmed in two very small scale showing that:
- People are more likely to search for take away food on rainy days
- People used to search for sports during a world championship
- People stops searching for shaving machines after Valentine’s day
Therefore, Manuel wonders that if these experiments can be replicated at global level, combining third party data sources (weather and events) with advance data preparation and analytics, it could be possible to boost the performance of the campaigns thanks to a better prediction of the user behavior, enabling Manuel, as account manager, to achieve the expected performance objectives.
As mentioned, Manuel is an expert in digital marketing, so the main challenges are motivated from his need of advance data management solutions in terms of:
- Preparation: Extract, transform and load the performance data to a common storage environment.
- Enrichment: Connect with third party data sources and link the data with the campaign indicators.
- Aggregation: Define the best level to describe the case (from city to region and country) depending on the case needs.
- Processing: Modellized the keyword behavoiur depending on the external factors based on the correlation identified with past data.
- Visualization: Develop a dashboard used by the account managers to implement the analytical actions and display the results.
EW-Shopp solutions for Advance Data Management
In order to solve those kinds of challenges, EW-Shopp has developed a complete set of solutions covering all the challenges addressed. For data preparation, enrichment and aggregation, the combination of Grafterizer and ASIA enables JOT to be integrated with other data sources and generate the data sets needed for keyword modelling
The keyword modelling is developed based on QMINER module; this service enables the semantic clustering of the keywords. Data are then aggregated in the clusters to generate enough amount of values to generate high accurate models, which will be the basis to deliver the prediction service
Finally, visualization features are implemented based on the KNOWAGE tool, connected directly with the data basis to display the more relevant outputs and guiding the decision-making process to the account manager.
From Data Preparation to Data Visualization
For the implementation of such services, there have been required the implementation of the following actions: Historical data preparation (SINTEF), identification of the more relevant indicators (JOT), data enrichment (UNIMIB) and aggregation level based on campaign management and keyword behavior modelling (JSI) and main visuals for outputs interpretation (ENG).
The first pilot was implemented in 2018, expected to complete further development by the end of 2019. During the development, main challenges covered were data preparation and definition of the aggregation level has been the main challenges. They were solved based on domain knowledge and high skilled partners in data management and enrichment.
The implementation of the initial pilot represented an important milestone, however further rework was done for keyword modelling.
It is being developed four different services enabling a better management of the keyword/campaign management, the main goal is to achieve higher performance indicators consuming the budget in the most relevant campaigns at the right launching time.
Target market are companies publishing their content and products, who will be the main benefited entities. Also, Manuel will increase the ROI of his campaigns as he will invest the budget just in those campaigns more relevant for the society.
Boosting campaign to the next level with advanced Data Analytics
The real benefits come from the generation of more visits for our clients’ sites in those campaigns that are more relevant for their targeted audience. This service allows to invest the marketing budget more efficiently, avoiding spend money in campaigns with no relevance. There is no solution in the market offering these capabilities. The opportunity of take into consideration external factors for marketing campaign management at global scale will position JOT as one of the top players in terms of volume and performance worldwide.
“EW-Shopp provides a unique set of data analytics enabling digital marketing account managers to boost the performance of the campaigns to the next level, investing the budget in those campaigns really relevant for the society”
Fernando Perales, Head of JOT Research Lab
For additional information on the EW-Shopp approach to data processing you can also check:
- The paper presented at the 18th International Semantic Web Conference (ISWC) on the Semantically-enabled Optimization of Digital Marketing Campaigns.
- The tutorial Semantic Data Enrichment for Data Scientists presented at the Extended Semantic Web Conference (ESWC)