Building a Composable CDP Using Entity Resolution at Fortnum & Mason

The User

Fortnum & Mason is a heritage luxury retailer dating back to 1707, renowned for its high-end products and premium customer experiences. With flagship stores in prestigious locations at London and Hong Kong and a growing online presence, Fortnum & Mason has become synonymous with bespoke luxury and tradition. As a brand deeply rooted in delivering memorable experiences, they continually strive to innovate and personalize their offerings. Whether through exclusive in-store events, world-class dining experiences, or tailored product recommendations, Fortnum & Mason remains committed to providing customers with unparalleled value and a sense of timeless elegance.

‍

The Problem

Fortnum & Mason faced significant challenges in creating a unified view of their customer data due to fragmented and incomplete records. Fortnum & Mason's customer information resides across multiple systems, such as restaurant bookings, email sign-ups, and online or in-store transactions. The lack of standardized identifiers resulted in a missing Single Customer View. This hindered their ability to understand customer behavior holistically. Fortnum & Mason initially tried a third-party service for identity resolution. This led to non persistent customer identifiers and offered limited visibility and control over the process. Sharing the entire dataset with a third party service was not privacy compliant either.

Fortnum & Mason needed a scalable, transparent solution to consolidate their data and accurately resolve identities with a persistent unique identifier. This single customer view would then be used to manage and optimize their customer engagement strategies, plan new products and fine tune customer service.

‍

Welcome, Jonathan. Very thrilled to have you here today! I would love to hear about your background and experience as well as your role at Fortnum & Mason. So why don't you go ahead and tell us about it?

Thank you very much for having me today. I am Jon Moss, and I have been working at Fortnum’s for nearly three years now. I look after customer engagement, which covers data and insights, customer experience and customer engagement. It essentially involves looking at how customers are shopping with us and how we can deliver better experiences for them. A lot of it revolves around how we are using data, insights, and technology to deliver that. I have worked with customers for most of my career, at least for the last 15 or 20 years. So, it is really exciting to see these three different areas of experience and engagement and insight analytics coming together now.

‍

That is surely a lot of exciting work around customer data. As a business stakeholder, how do you see the role of data within the work that you are doing within Fortnum & Mason?

I think it has been becoming more and more important for us at Fortnum’s certainly over the last three years that I have been here. And I know even before that there have been a fair number of plans in place to improve how we are able to understand the data that we have got and improve the skill sets that we have got around using that data. This is to really turn it into information that our teams are able to use to deliver better customer experiences, more relevant products, and more relevant services. What we are doing at the moment is running a data transformation program that is a multi-year program. We set out some ambitions a few years ago about where we wanted to try and get to. This has helped us move towards bringing our data into the cloud, starting to build our own capability and taking ownership of our data. Now we are getting to the stage where we have got a good level of accuracy and understanding of what we have got. Now we are able to start turning that into insights and use it to personalize experiences online. We are going to start doing this in store and in our restaurants as well. Being able to connect all of these different customer experiences together is really important for us too. We are on the cusp of launching a membership program which will help to bring all of that together.

‍

Amazed to see the journey that you are taking and the kind of work that is being done. It is a groundbreaking data platform that you and the team have created. Could you please walk us through the process of how you are actually collecting the data. How does the data get accumulated? What is the life cycle of the customer data for a big company like Fortnum and Mason? And how does it all really flow?

What we have done as we have been working with you, Sonal, and your team over the last six to eight months is that we have really tried to get down to simplify what are the most important data sets that we need. We are now understanding what are the things that actually help us know who our customers are and where they are shopping with us. We have really been able to simplify the sources at which we collect data and then they feed into our single customer view. The data customers give us when they sign up to receive emails from us and give us marketing permission, the data we collect from customers that make bookings in our restaurants. Also, the information we get from customers, whether they order over the phone or online or through our in-store ordering, we are able to use that information. We bring together all the information we collected at these three different data points. We are using that to create the single customer view. That is what sort of feeds into Zingg to give us the holistic view of a customer. Then we are also able to start enriching it a little bit more with some of the other data points we might have. These points could be customer service interactions, perhaps data we might collect at till points in store, or other information that we gather through restaurants. I think we have got to quite an exciting stage now where we are able to, for the first time, understand how customers are shopping with us, whether it be online, in-store, over the phone, or in restaurants. We have never been able to do that properly before. Zingg has really helped us out in doing that, providing us with a really easy way to plug our data in, and giving us a lot of really interesting insights.

‍

We are glad to have played a part there. Can you please tell us a bit more about the challenges within the customer data that prompted you to look at entity resolution and what were the approaches you tried for solving them.

Well, a few of the main challenges that we had were using D365 as a starting point, it is where all our customer data resided. We had multiple records for the same customer just in terms of the way that orders were created. That is a really big challenge for us to try and get over to actually understand how customers are shopping with us when every single record that we have got looks like a different individual. That is one thing that we recognized. The kind of identity resolution we were doing was being managed by a third party beforehand. We did not really have a lot of visibility around how that identity resolution was being completed, and also, the information feeding into that was incomplete. So these three areas probably: being able to feed all of the relevant data sets we want into identity resolution; being able to get some confidence about accurately matching data together where it is the same customer rather than looking like a number of individual customers; and building the capability within our team as well to be able to have a platform that we can own and operate and understand. One of the things is also having the flexibility to be able to quickly look at what's happening with the data, being able to adjust the settings around how we want to group customers together, and understanding the nuances and things that are going on. The part around the visibility and the understanding of the rules started to give us a lot more confidence in how we are using our data because now there is more transparency about what is in it.

‍

You started using the community version of Zingg and then eventually moved to Zingg Enterprise. Can you walk me down through the reasons that led you to opt for the Zingg Enterprise version?

Yeah, of course. So, as part of bringing the single customer view in house and starting to take more ownership of our data, we wanted to explore different options that were available to us in terms of how to build out a single customer view, which was really where we wanted to get to. Zingg was one of those options open to us. Having the open source, it was quite an easy way to understand how it worked, and give our technical team confidence in the tool. It was just that level of understanding really of how everything would come together in Databricks and then work in the tool. For us, that was phase one of building our confidence in the tool, and understanding how it worked. It gave us some quite quick outputs actually and built a level of confidence with the business, like this is what we are going to be able to see. Once we were happy with the tool and the results we had seen, we realized that there are additional features that the Zingg Enterprise version offers us. It was quite easy for us to build a business case around moving to the enterprise version, given the benefits that we would have around the time that the model would take to run, and some of the additional features that we get as a result of having the enterprise model. These were the two phases: Understanding, testing, learning, and validating the results; and then going ahead and building the business case once we were happy and moving forward.

‍

I remember a lot of conversations between us regarding the Zingg Incremental Flow and the Zingg ID. I am glad it has been working so well for you. In terms of deployment, what were the timelines from start to finish, starting with you discovering Zingg and ending with getting to production?

I am going a bit from memory here. I was actually talking to a colleague of mine about this earlier today. We were sort of reflecting on how quickly, realistically, we got from the proof of concept to actually having the solution live. I think it was probably about two or three months in total going from the point at which we had started the proof of concept to us having the final outputs flowing into the different tools we are using around CRM. A lot of that was down to the simplicity of the tool, how easy it was for us to understand the outputs from that. Relatively speaking, as a layman coming into it, it was pretty easy for me to understand how the model was working, what it was doing, and what the rules were. That level of simplicity is what probably helped speed things up and quickly gave us confidence. Also, the level of involvement from your side really helped us to get from the proof of concept to the stage where we use the Zingg Enterprise version.

‍

Thank you so much for that. In business and technical teams, a lot of companies would talk about single customer view, but you are probably one of the few teams who has gone out and done that. You should be very proud of the fact that you have achieved a single customer view, and that you have invested time and effort into that successfuly. My question now is, what is next?

We have two or three things now that we really want to work on. The first one of those is enriching the single customer view we have got with some of the additional data points that I mentioned earlier. We are collecting information that we think might be useful to help give our customers a better experience. It might come through customer service, it might come from their online experience, maybe through Adobe Analytics or through Adobe Target. We want to bring relevant information back into our data set to understand how we can better serve our customers. Hopefully in turn, that then means we can deliver better outcomes to more of our customers. The next big thing will be feeding in the information that we started to get from memberships as well, being able to identify and flag those members in the single customer view. One of the big challenges that I have also heard from lots of other retailers as well is how do you bring together information about customers who are shopping in store where they might normally be unknown. That is still probably one of the big things we have got to work out this year. Without having something like a club card or a Nectar card, what can we do to try and encourage customers to share information with us when they are shopping in store and how can we use that to give them a better experience. So, one is enriching our single customer view, and then the next thing is really about putting it to work. We are working now on understanding how we could build and deploy a customer data platform. The single customer view is going to be a really big part of that in terms of being our single source of truth, and feeding a lot of information is going to help us determine the next best actions and experiences across the different channels. Although that only sounds like a couple of things, it is probably quite a lot to be getting on with over the next 6 to 12 months. All the while, now that we have got a better view of our customers, we are able to provide more insights across the business, around who is shopping with us, what are they buying, what sort of products should we be thinking about, and how can we improve experiences across different parts of our business. Hopefully this will also in turn help us with our growth going forward.

‍

Absolutely. I do hope so. Surely with the effort that has being put in by you and the team, there is going to be a lot more action happening and a lot more results too. As we close, is there anything else you would like to add, Jon?

I would like to say that it has been a really great experience working with you and your team in terms of not only the level of support you've given, but also the speed of the support you have given us, and some of the things that we have identified as a result of now having the ownership of our data. We have realized that there were some discrepancies in some of the ways that our systems were talking to each other, which we have been able to rectify. A lot of this has been picked up as a result of looking at the data coming out of Zingg. So, it has been a really good experience for us, and it has been great to work with you and your team as well.

‍

Thank you so much for that. We are very thrilled to be working with your team and it has truly been a collaboration in that sense. It has definitely been fun together. Thank you so much Jon.