<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=306561&amp;fmt=gif">
Skip to content

Setting The Groundwork for Customer Personalisation

Latitude Financial Services Head of Data Platforms & Engineering Fahad Najeeb talks customer personalisation systems

One of the opportunities data and AI is presenting is the ability to personalise systems. Netflix and Amazon are both known for their personalised recommendations, content suggestions that are determined by an algorithm that tracks our viewing habits.

While on occasion we’ve probably all had bizarre suggestions and wondered how on earth Netflix thought that would be something we wanted to watch, these systems play a central role in maintaining engagement in digital viewing platforms.

Earlier this year, Fahad Najeeb, Head of Data Platforms and Engineering at Latitude Financial Services, joined Corinium Global Intelligence at the Chief Data and Analytics Officer event in Melbourne. During his session, Najeeb explored how these AI-powered personalisation systems can open up markets and drive new business growth.

We had so many questions come in during his session that Najeeb didn’t have time to answer them all. However, Business of Data caught up with him recently to explore the personalisation journey in a bit more detail and discuss some of the questions that came in.

Building Confidence in Data

When undertaking any new data or AI project, an initial challenge is always securing the business' buy-in. This is a common problem and part of why many companies are seeking to build a data-driven culture, where teams at all levels trust and aim to better utilise data and other tools available to them.

On this topic, Najeeb argues that the key is selecting a use case that is close to the heart of the business, and then using technology to map to the challenges seen. From this, a story can be articulated and the ROI opportunities demonstrated.

“We have found data ecosystem and martech platforms complement each other quite well where you can ingest and move data at a fast pace, then utilise it for marketing comms, then measure the success of campaigns. You have a faster turnaround, and you can attribute it to effort spent,” Najeeb says.


The ability to deliver ROI is one of the main challenges for AI projects, and so far, statistics suggest that a project will not deliver. As such, new measurements and an understanding of ROI are required to understand the successes and the value these projects deliver.

For his project, Najeeb discussed the use of control and test groups.

“To measure success you need to have two groups, one receives the personalised offer and the other who doesn’t,” he says. “Then you compare and measure the uptake of both groups and, if your personalisation project is successful, you should see an increased uptake in the test group.”

Data Collection

In order to deliver a personalised service, one thing is needed: data. Although this may sound simple, preparing and utilising data for business value is a challenge for many businesses.

To create a 360-degree view of the customer, there are two options, build or buy capabilities.

According to Najeeb, making this decision depends on the appetite to spend and what is true for the organisation.

“You will need to do some build effort and engineering no matter what product you end up buying be it an MDM solution for transactional purposes, an end-to-end modern data lake ecosystem for analytical/ML purposes, and/or a customer data platform for enrichment/segmentations. Ideally, it will be a mix as one can’t go about writing any of these from scratch,” he says.

“The key is not to reinvent the wheel. Conduct a market scan, map it with your ground realities and decide on build vs buy vs build a little and buy the rest.”

Data Building

When it comes to the data build, the more you know about the customer, the better the personalisation can be. There are challenges in building a rich enough customer profile though. To combat this, Najeeb recommends a combination of online event data, complemented with offline transactional data.

“This provides the best breeding ground for an enriched profile, allowing you to further enrich this with your organisation’s operational metrics and use them for active decisioning for personalisation,” he says.

In terms of the practicalities of building a data set for providing personalised services, Customer Master ID is key, and Najeeb recommends starting with the practice of a master identifier early on, as having one master identifier makes your landscape simpler and easier to use.

Although he concedes that you can achieve some wins without it, albeit trying to harmonise customer data with multiple systems increases the operational and engineering effort required.

Having a Customer Master ID, however, doesn’t mean that you always need a single version of the truth.

“You may want to use multiple versions of truth for aggressive data strategy, think of marketing use cases that may rely on domain-specific multiple definitions versus a defensive data strategy, think of regulatory use cases,” Najeeb says, “HBR came out with a well-thought position on multiple vs single source of truth.”

When it comes to data definitions, Najeeb believes that multiple definitions early on may help organisations get their data product to production faster.

“Multiple definitions enable frictionless adoption of the data modernisation journey. Later, you can homogenise using an overlaid Enterprise Data Model. You may also grade your data and tag it for appropriate criteria such as trust, quality and so on,” he says.

Personalisation Challenges

Another question put to Najeeb was around the challenge of identifying and then stitching data from users who are not actually logged into the service together with other data sets.

“Leading CDP platforms provide this out of the box, they use deterministic and non-deterministic algorithms and are smart enough to tie / stitch anonymous / non-registered users with their registered versions,” Najeeb says.

Of course, getting data-ready is the first step. Next is using the data and creating automated tools that personalise the offering for a customer. In order to automate image and comms for these personalised services, Najeeb says they must be organised and centralised.

“Your digital assets, such as images, would usually be in a file server setting as bare minimum. Or you will ideally have a digital asset manager and/or a CMS which you can integrate with your personalisation services, think of a custom written microservice,” he says.

“That microservice will hook up with CMS/DAM on the run to fetch the appropriately tagged digital asset that will have to go with the relevant offer that you are about to present to a cohort / customer.”