Data Science is used in various spheres of digital marketing : Search Engine optimization, customer profiling and targeted marketing, market research, customer churn prediction, predicting customer lifetime value, etc. Let’s dive deeper into this in this article!
- Marketing analytics: How effective your marketing strategies are in terms of reach, ROI, etc is the main thing that marketing analytics deals with. There are many marketing analytics tools like Adobe Analytics, Marketo, Hubspot, Alteryx platform, etc. They provide features like Customer analytics and segmentation, forecasting, optimization and predictive analytics, etc — all of which require data science and analytics to some extent.
- Targeted marketing and recommendation systems: Concentrated campaigns that look to widen a specific customer base on the basis of age, class, etc. For example “The Social Dilemma” — a Netflix original about how social media is designed to be addictive talks a lot about targeted Marketing.
“The Great Hack” — another Netflix original told us how mass harvesting of user data from Facebook combined with microtargeting led to one of the biggest scandals in the US — the Cambridge Analytica Data Scam. This company’s work influenced the results of the 2016 US Presidential election and Brexit and many more political scenarios. It was a political data analysis and consulting firm and the CEO claimed to have “5000” data points on every American voter. Imagine that- a company knowing 5000 things about you. I don’t even know 500 things about myself!
Remember that time you thought of buying that gadget or that dress on amazon and took a look at it and then the ad for that very gadget/dress ended up chasing you everywhere? That is targeted marketing enabled by data science. So are the recommendations provided on Netflix, Amazon, Spotify etc. So many of the shows I’ve found and loved have been recommended to me by Netflix. The recommendation algorithm is the main reason people love Spotify because it feels like someone spent a large amount of time to get to know you and then curated a playlist just for you. Amazon’s “People who bought this also bought” feature simulates a real-life aisle in a retail store.
Looking for datasets to get into building recommendation engines? Click here
![](https://i0.wp.com/supervisedlearning.com/wp-content/uploads/2020/11/sdds-1.png?resize=581%2C383&ssl=1)
The point of marketing is to influence individuals to buy/do something. Knowing these individuals better and putting them into different buckets and then tailoring marketing activities to these groups/buckets is customer segmentation and targeted marketing. The more you think about that statement, the more you realize this is exactly what the sorting hat did with the folks at Hogwarts. With greater amounts of data available and more complex algorithms, micro-targeting has emerged as a new and improved way to understand and target customers. It was even used in the 2019 Lok Sabha Elections so as to calibrate the political ads and messages to the target voter base.
Market basket analysis is done especially by large retailers. The purpose of this analysis is to find out what products are bought together by customers. For eg: people who buy bread might also buy eggs, butter, and jelly. Using the point of sales data of customers can provide enough data for doing market basket analysis. Based on the results of this analysis, action can be taken to nudge the consumer to buy all the related products- for eg: placing these items together in the same aisle. Here’s a dataset if you want to try your hand in market basket analysis.
3. Predicting customer lifetime value and customer churn: After segmenting customers ( based on their frequency, revenue, demography, characteristics etc ), one has to predict the customer’s life time value ( CLV ). The CLV metric shows how valuable a customer could be to the company over a limited span of time. The higher the value, the more attention and pampering ( discounts and other such attractive offers) is worth it. The next important metric to predict is customer churn. Retaining customers is one of the main concerns of a company because more often than not, the retention of existing customers is cheaper than acquiring new ones. Predicting churn will throw light on who is most likely to leave and then preventive measures can be taken by the marketing department to make sure they stay. Churn prediction is a binary classification problem and hereit has been done using XGBoost.
![](https://i0.wp.com/supervisedlearning.com/wp-content/uploads/2020/11/0DDoqzU1LYz-Ucf9L.jpg?resize=525%2C350&ssl=1)
Predictive analytics can be used in prioritizing leads for full-funnel marketing by predicting if and when they will flow through the different stages in the funnel ( especially employed in B2B sector). This is where Lead Scoring comes in. Based on the characteristics ( behavioural, social media, demographic, purchase history etc) of the lead, scores are given to each customer ( the higher the score, the more they are likely to buy). Thus, people from the sales department can be appropriately deployed to each of these leads, thus optimizing ROI.
In here, random forest has been used for Lead Scoring.
Click here to find 10 datasets in Marketing and Advertising to test out your ninja data science skills and have fun!