MulletIQ

For a retail platform with a yearly marketing budget of more than €2 million (core: Google Ads, Facebook Ads, TV, radio and print), we designed a custom MMM to help them understand the impact of their marketing spend on their sales. We based the model on their sales, marketing spend, (organic) traffic customer, customer retention, customer lifetime value and other key metrics.

Where we would normally code a MMM in Python, we build this model in Google Sheets, on request of the client. This way the model and results were accessable to everyone and easier to understand. The model helped the client to find a more optimal marketing mix and reduced the yearly marketing spend by nearly 50%, with minimal loss of sales.

We have helped one of our clients set up their own A/B testing framework on their e-commerce platform. The client was already using a feature toggle tool to launch changes to their website and was collecting clickstream data (Snowplow). We used that feature toggle tool to steer users through different versions of the page. The new data of the experiment was integrated in their existing data collection pipeline. The client was using Google BigQuery as a data warehouse, so we calculated the p-value's of the experiments with a permutation test pipeline inside a Google BigQuery pipeline (Dataform). Finally, we visualised the results in Google Looker Studio. The most important deliverable was that every element of the framework was done with tooling the client was already using, therefore they were able to maintain and improve the software themselves.

One of our clients asked us if we could find the most important elements of their e-commerce website that could be optimised. By using historic data on all the pages and elements were tracking and a Monte Carlo Markov Chain we built a software that could predict how much uplift in revenue a certain change to an element would have. We delivered the software in a Python notebook in the cloud that their Project Managers could use and the results were updated daily. This tool helped the Product Managers prioritise their backlog and make plans accordingly.

Writing content is a laboursome task. An e-commerce platform with a small content team reached out to us, because they had thousands of products online with outdated or even no descriptions and with the current team size, it would take years to write. They did have multiple sources with information about the products available, but those were in a different language or of too low quality to copy and paste. Also for SEO reasons, this was undesirable.

We built a Langchain pipeline connecting the available sources and added ChatGPT to summarise the information and generate elaborate descriptions with different sections highlighting specific attributes of the product. The final texts were delivered in several languages and automatically deployed on the website.

Every outdated or missing product description was replaced with a fresh one overnight. We also made sure the results were A/B tested and the outcome was a 30% revenue increase on products with a generated product description, compared to the outdated ones.

We have integrated ChatGPT in a WhatsApp business phone number. The WhatsApp chat is now connected with a ChatGPT Assistant, that is easily tweaked to one's specific needs in the OpenAI user interface. This solution is an ideal solution for companies looking to reduce customer service expenses.

The value of a customer is much more than just their first purchase. So marketing spend should not only be linked to this first purchase, but also to all future purchases. When looking at the CLV through time, you can measure the progression of retention efforts and conversion improvements on the platform. You can also calculate the CLV for specific customer groups and compare them to each other. This gives valuable insights on which customers are actually more valuable in the long run. For many clients we have visualised the Customer Lifetime Value and integrated it into their dashboards and CRM, resulting in better targeted marketing efforts.