Building a Sophisticated Alerting System for MandM Direct


MandM Direct

Tech stack

Google Cloud


Alerting system


AI + Machine Learning

MandM Direct are a leading online retailer that serves over three million customers across the UK and Europe. Using their existing data stack on BigQuery and Looker, the Datatonic in-house Looker team built a near real-time alerting system that monitors MandM Direct’s target KPI’s, such as the volume of order confirmations. This system allows business users to identify any issues or frictions in the customer’s online experience as these occur, and allow them to take action proactively.

Our impact

  • Near real-time reporting on deviations against target KPI’s, such as: number of adds to basket, numbers of orders and number of cart abandonments
  • Intuitive and easy-to-use interface for business users to configure their own alerts, allowing them to focus on action and not data manipulation
  • Email alerts sent automatically to operations and business teams to action quickly and remove friction for customers


The challenge

Five years ago, MandM Direct kickstarted their digital transformation journey. They retired their print catalogue business and refocused their attention on the digital customer experience. From a data perspective, this journey meant being able to collect and consolidate every customer interaction, and feed these insights seamlessly into relevant functions. 

At the time, the business lacked the technology and tools to do so.

“Five years ago, our data was siloed. We didn’t have a way to join up our disparate sources and paint a full picture of the MandM customer experience. Analysts were asked the same questions and would return different answers due to different data sources, which was incredibly frustrating.” – Ben Powis, Data Science Manager, MandM Direct

Moving forward, BigQuery was chosen as MandM Direct’s cloud data platform, consolidating their data sources and providing a single customer view. Looker was also chosen as the retailer’s BI platform for end-users, sitting on top of the data stored in BigQuery.

“Looker gives us a semantic layer between our raw, messy data and our business users. It’s here where we can set company-wide definitions of key KPIs. This means any data user across the business can focus on insights, without having to worry about using the correct definition of revenue or product type.” – Ben Powis, Data Science Manager, MandM Direct

Today, MandM Direct have established themselves as a leading online-only clothing retailer, serving over three million customers, stocking over 300 brands, collecting over 100 million website visits and sending over 400 million emails each year. 

In the highly competitive online fashion space, every hour, and even every minute over peak trading periods, is critical. Enabled with BigQuery and Looker, MandM Direct were then seeking the next level in BI analytics – real-time reporting and alerting on target KPI’s. 

This meant: (1) identifying deviations as these occur in website activity, such as the volume of abandoned carts, adds to basket and order confirmations, and (2) setting up email triggers to send directly to relevant stakeholders to take immediate action.


Our solution

As the go-to-partner in EMEA, Datatonic’s in-house Looker team have established a strong reputation for their best-in-class technical and execution expertise. Over the past five years since kickstarting our partnership with Looker, the team have led countless successful enterprise-level projects end-to-end. 

Understanding MandM Direct’s need,  our team set out to: 

  1. Technically scope the client’s requirements.
  2. Build the custom system using LookML and Scheduler. 
  3. Enable business users with an intuitive and easy-to-use interface.

1. Technically scope the client’s requirements 

The solution required setting up live performance alerts, which takes into account weighted averages or linear regressions, and gives the user the self-service ability to adjust weightings and alert timings. The system has to handle live data, calculate MandM Direct’s target KPI’s, and monitor any deviations in near real-time, such as every five minutes. 

“During the scoping phase, we had to think about what value should the client expect on their target KPI’s, and how could they adjust these values for seasonality. For example, the client would expect a different volume of order confirmations at midnight vs. midday, or over Summer vs. Christmas period. The value per KPI then needed to change throughout the day.” – Jamie Fry, Director of Data + Analytics, Datatonic

This rolling calculation for each target KPI, i.e. the weighted averages, were then built using three dimensions of time – same time yesterday, same time last week and same time last year. Business users are also able to tweak weighting sensitivities, such as placing greater importance on the same day yesterday over the same day last year. 

In addition to calculating performance based on historical data, our team incorporated a predictive element. This meant including linear regressions in the model to forecast the value of each target KPI at any given time. 

2. Build the custom system using LookML and Scheduler 

As experts 100%-dedicated to Google Cloud Platform, Datatonic were one of the best partners placed to deliver MandM Direct’s solution. The client was already enabled on Google Cloud Platform, so our team worked fast and efficiently to design the client’s custom solution completely on BigQuery and Looker. 

“BigQuery stored and processed all of MandM Direct’s live streaming data. Looker, specifically LookML, was used to design the KPIs, Time Dimensions, Weighted Averages and Regression Modelling. Looker’s Scheduler was then incorporated to send the email alerts to business users. The components were then fairly simple. The sophisticated work came in the coding that sat underneath each component.” – Jamie Fry, Director of Data + Analytics, Datatonic

3. Enable business users with an intuitive and easy-to-use interface 

The end-user of MandM Direct’s alert system are analysts and business stakeholders. So while the calculations built into LookML are complex, it was critical that the user interface had to be seamless, intuitive and simple to use.  

“The end-user can configure their own alerts by using only Looker’s Explore and Scheduler functions. In Explore, they’re able to select the metric, reference value and time range, and tweak the parameter sensitivities. In Looker, they’re also able to visualize these target KPI’s. Finally, through Scheduler, the end-user can create the time intervals and assign the email alert for specific KPI’s.” – Jamie Fry, Director of Data + Analytics, Datatonic

The alerting system has enabled business users to identify in near real-time any issues in the customer’s online experience and allow them to take action proactively.

“Being able to react quickly has been invaluable, especially in peak trading periods like Christmas, where we take thousands of orders every hour. We are very pleased to also leverage the power of Looker, so no time was spent developing additional plug-ins and other services. Instead, we were able to use our existing tools using LookML and email alerts.” – Ben Powis, Data Science Manager, MandM Direct