# Data analysis and ML modeling for digital marketing campaigns

How impactful are your digital marketing campaigns? How can you use your customer conversion metrics to best allocate your marketing budget? We teamed up with [Knightsbridge Park](https://kbpark.com/) to address those questions in the context of their real estate digital marketing operations.

To accomplish this, we analysed aggregated data from:
- [Google Analytics](https://analytics.google.com) for digital marketing campaign data
- CRM data for clients conversion results

And used the following stack of tools:
- [Jupyter](https://jupyter.org) as the development environment
- [Pandas](https://pandas.pydata.org) for data analysis and manipulation
- [H2O.ai](https://h2o.ai) for practical and powerful machine learning model training and importance analysis
- [Plotly](https://plotly.com/python) and [Seaborn](https://seaborn.pydata.org/) for interactive, clean and insightful data visualization

![solution](/assets/portfolio/marketing_analysis.png)
*Combining rich data sources with powerful analysis tools to produce valuable business insights*

Our Exploratory Data Analysis (EDA), warranted by clean and informative visualization, revealed notable patterns in the interactions between marketing spending and customers conversion metrics. We were able to uncover meaningful sasonalities and to detect rises and drops of cost efficiency.

A machine learning model-based approach corroborated the EDA findings, describing in further detail latent relationships between variables. Out of many model architectures tested, the winner XGBoost allowed us to use SHAP for feature importance estimation. By using a model like XGBoost we can run future what-if scenarios, fine-tune budget allocation and test out different scenarios to determine the potential outcomes. This can ultimately lead to better conversion rates and a higher return on investment for the marketing campaign.

![shap](/assets/portfolio/marketing_analysis_shap.jpeg)
*SHAP analysis is an effective tool to assess features importance*

In summary, the combination of rich data sources, professional EDA and advanced machine learning helped us produce objective insights for direct business impact. The patterns and relationships identified can inform budget decisions, helping to maximize the efficiency and effectiveness of marketing spend.



 

---

Do you work with digital marketing and want to explore the potential of your analytics data and create true business impact? [Get in touch with us](/#contact)!

How impactful are your digital marketing campaigns? How can you use your customer conversion metrics to best allocate your marketing budget? We teamed up with <a href="https://kbpark.com/">Knightsbridge Park</a> to address those questions in the context of their real estate digital marketing operations.
To accomplish this, we analysed aggregated data from:
<ul>
<li><a href="https://analytics.google.com">Google Analytics</a> for digital marketing campaign data</li>
<li>CRM data for clients conversion results</li>
</ul>
And used the following stack of tools:
<ul>
<li><a href="https://jupyter.org">Jupyter</a> as the development environment</li>
<li><a href="https://pandas.pydata.org">Pandas</a> for data analysis and manipulation</li>
<li><a href="https://h2o.ai">H2O.ai</a> for practical and powerful machine learning model training and importance analysis</li>
<li><a href="https://plotly.com/python">Plotly</a> and <a href="https://seaborn.pydata.org/">Seaborn</a> for interactive, clean and insightful data visualization</li>
</ul>
<img src="/assets/portfolio/marketing_analysis.png" alt="solution">
Combining rich data sources with powerful analysis tools to produce valuable business insights
Our Exploratory Data Analysis (EDA), warranted by clean and informative visualization, revealed notable patterns in the interactions between marketing spending and customers conversion metrics. We were able to uncover meaningful sasonalities and to detect rises and drops of cost efficiency.
A machine learning model-based approach corroborated the EDA findings, describing in further detail latent relationships between variables. Out of many model architectures tested, the winner XGBoost allowed us to use SHAP for feature importance estimation. By using a model like XGBoost we can run future what-if scenarios, fine-tune budget allocation and test out different scenarios to determine the potential outcomes. This can ultimately lead to better conversion rates and a higher return on investment for the marketing campaign.
<img src="/assets/portfolio/marketing_analysis_shap.jpeg" alt="shap">
SHAP analysis is an effective tool to assess features importance
In summary, the combination of rich data sources, professional EDA and advanced machine learning helped us produce objective insights for direct business impact. The patterns and relationships identified can inform budget decisions, helping to maximize the efficiency and effectiveness of marketing spend.
<hr>
Do you work with digital marketing and want to explore the potential of your analytics data and create true business impact? <a href="/#contact">Get in touch with us</a>!