Turn Fragmented Data Into a Customer Lifecycle Engine

KIM NUPP:

Thank you. I am glad to have you all here with me today. I'm Kim Nupp. I'm the director of the Customer 360 Management team at M&T Bank. It's a newly formed team since we've just implemented the Customer 360 — I got rebranded, so it's been a great success thus far. I'm here to tell you how we changed our thinking from product-centric to actually being a customer-centric organization. We'll talk a little bit about our strategic imperative, then jump into our journey and bringing the platform to life. We'll talk about our hands-on-keyboard reality, what we learned through that, with some real-life realities and advice. And then we'll talk about some of our wins and how we're using AI today.

So before we begin, I'll tell you a little bit about M&T. We're a very conservative financial organization with 170 years behind us — come this August. We were founded out of Buffalo, New York, out of the success of the Erie Canal and all the commerce that was happening at that time. Two entrepreneurs came together and built Manufacturers and Traders Trust Company. Out of that grew M&T Bank, now servicing the East Coast region from Maine to Virginia. We also have 22,000 employees in the communities we serve on the East Coast.

So let's talk about how we moved from a product-centric approach to a customer-centric approach. We're always looking at the different products we can provide to our customers — whether it's a mortgage, a home equity loan, a money market savings, a CD, or for our businesses, maybe a business line of credit, and so on. But each line of business operated independently in their own silos with their own data, not sharing that across the organization. It became difficult to really bring together one view of the customer where we could engage with them and deepen the relationship across the product mix. When we think about becoming a customer-centric organization, we like to think about understanding that customer and what products they need during their life stage and financial journey.

Do they need more credit at this time? Are they looking to buy a home, maybe buy their first car? How can we engage with those early customers that are just starting their financial journey and understand them in a way that they become a customer for life? That is important for us — to be able to understand and build those relationships as we go. And Amperity is helping us do that for the first time. We are building out customer-centricity thinking with customer segmentation. Our analytics team has been helping with that, but also thinking about how we can have a seamless journey that takes the customer through their financial experiences to not only help them in their life journey, but grow our product mix at the bank.

When we think about our single customer view, we knew when we set out on this journey with Amperity that that was the single most important thing we needed to do. We needed to take all those data silos and turn them into one unified view of our customer — whether it's a business banking customer who has a small pizza shop but also banks with us for deposits and savings. We wanted to be able to understand not only the business banking relationship but that individual business owner's personal relationship with the bank. Amperity has helped us do that. With these now-unified customer views, we can expedite our marketing, actually reach the people that need our services at the time they need them, and we're able to produce comprehensive insights.

I talked about customer segmentation — when we think about that, we're not talking about the segments or audiences in Amperity, but about how we can segment our customers through those life stages and understand the product mix they need during that time. Nathan and Ben are our analytics partners if you want to talk to them about how they've used the Amperity data to build out a customer segmentation dashboard that shows opportunities through each life stage — early, mid, and late. We want to grow that market in the early stage and keep them for life. The comprehensive insights that are coming out of Amperity are already changing the way we think about our customer-centric journey for tomorrow.

We really want to get to the point where we're meeting customers exactly at the time they're ready to buy that car or buy that house. How do we do that? Through real-time profiles. We can now know that someone is on our website looking for a car loan. And as we heard earlier today — if you don't respond within 20 or 30 minutes, they've already booked that loan with somebody else. They're ready to buy a car, and if they're not hearing back from you for a couple of days, you've lost a customer. With real-time profiles, we'll be able to gather that information, bring it into Amperity, and understand who is actually visiting our site for what purpose, and meet their need.

Finally, we want to have a context-centric experience. Even though we are in a financial relationship — we're bankers — we want to be able to meet you at the airport, meet you when you have that need. We have a lady who's celebrating — she just got that car loan from M&T Bank, so she's having a drink. We have a gentleman in the back who's looking up information on how he can get a mortgage with M&T Bank while he's waiting to board. We want to capture that point-in-time moment and be able to help them through their financial journey.

So let's talk about our journey as a bank and how we implemented Amperity. The core here is the customer data foundation — and you hear Derek talk about that a lot. When I first came to Amplify, the first conversation I had with Derek Slager was: we're not just looking for another CDP, we need something that's going to help us transform our data, bring it together, and give us a core customer data foundation — that true Customer 360. There are a lot of CDPs out there in MarTech, but what Amperity brought to the table was the ability to bring all that siloed data into one platform and give us visibility into how customers are engaging with the bank. We actually branded it internally as the Customer Data Management Platform. So we have our 360 brand and our Customer Data Management Platform that drives all of our insights and decisions.

As we think about the customer data foundation, not only does it drive activations, but it helps us execute faster. It also allows our analytics teams to bring new strategic insights to the table on where we need to activate next. So it's a continual cycle — we keep enriching our customer data foundation. Now we're bringing in additional sentiment data, which we've never had before. We've been all about selling our products and making sure we're meeting people's financial needs, but not so much understanding how they feel about us as a bank or about the products we sell. So we're starting to move toward gathering those insights.

So let's talk about what we've done to get there. In 2023, I sat down with one of our strategic financial officers and pitched this idea that we needed a Customer 360. We needed a core place that wasn't reliant on master data management to bring identities together deterministically. We needed a probabilistic way of matching identity, so we knew that Joe the pizza guy was the same as Joe Smith that had a retail relationship with us. Once I pitched the journey and got her on board with the Customer 360, we got the financial funding and went out to RFP. We had 12 vendors in the mix in the beginning. We narrowed it down to three who we invited onsite for a demo. During the onsite demo, we gave them some of the most problematic data we had. We gave them Girl Scouts of America data where their members are all over the country, and we wanted to see if they could build a Customer 360 to bring those records together.

We gave them complicated data for Mario and Luigi the plumbers — we used a lot of fake data, but it mimicked the experiences we had trying to bring this siloed data together. Two vendors failed, one succeeded — it was Amperity, obviously. During the 45-minute lunch break, they actually built a Customer 360 with that problematic data. It was awe-inspiring. We had about 15 people in the room rating them, and they came out 40 points ahead of any of the other vendors that walked in. Very impressive, especially with our messy data. So that embarked us on the journey in 2024 to actually start building.

It was very arduous. We brought in 30 source systems — and now we actually have 70 source systems feeding our Customer 360, so we are really growing the data assets. But it all started with the Amperity ID assignment. In January 2025 we launched. Liz, our Chief Marketing Officer for our consumer and business banking divisions, is here, and she'll attest it was a rough journey that first quarter. She was very patient. We asked her to prioritize what really needed to get done and what could wait. We worked with our marketing partners to understand their priorities so we could get things out the door in the first quarter and sunset our legacy software, which renewed at the end of January. I was driven to save that million dollars no matter what. We made it through that first quarter and then we started actually seeing some lift.

We started seeing efficiencies. We started working better together. We were gathering more attributes and building more data assets. We were literally building the plane as it was landing, as they say. But all in all, it was still such a huge success that we could even launch that quickly from the time we signed the contract with 30 data sources landing in a Customer 360. What we're looking at now is really continuing to grow those data assets. We've brought in third-party data and used Stitch against it, so now we have demographic data about our customers through the stitch process. We've even brought in firmographic data — Dun & Bradstreet data — against our businesses so we can understand them better and the revenue possibility and generation. When we first embarked with Amperity, they said they only do retail for stitching — they'd never done business — but let's give it a try. So we worked with Arthur and others on the Amperity team to make sure we could stitch our customer data, which is very complex and messy.

As we think about the hands-on-keyboard reality — some of the things we did, which we'll talk about in more depth — we ended up dividing my team into value streams. I sit in the business side of the house. I'm not tech, not in the data world, although I come from that background. We divided into a data prep team, a platform enablement team that would build the data assets, an audience activation and governance team — Venkat is here in the front row if you want to learn how he's activating campaigns in the platform — and then all around this we had a data quality and governance team. I hired that team new to be part of my organization to make sure everything was vetted, documented, and that there was quality control around everything delivered, including every list that goes out the door. We're a highly regulated industry in banking, so the quality team was very important.

The Customer 360 is the foundation, but it really takes a full army of people to be successful. Beyond our Customer 360 team, the analytics team — we listened to them and saw what they needed. We would meet frequently with Nathan, Ben, and others on that team to understand the most critical and priority data they needed for insights. They've become part of the backbone as well within the Amperity use cases. And then of course marketing, sales enablement, and Venkat's team doing the actual audience creation, segmentation, build, and activations.

On the use cases — one I haven't mentioned yet is our customer experience team. We've talked about marketing and sales enablement, how we're using omnichannel journeys now, how we're triggering different events. We're building out paid media campaigns. We have our analytics and insights team creating incredible dashboards and taking insights to the business banking team that they've never seen before. We've even started using the customer data agent to show the business banking team where the future can lie with using AI to help them grow, retain, and reactivate their customers. And with customer experience, we do a lot of surveys — we send our survey audiences out to Qualtrics and get those scores back. So now we're starting to ingest NPS scores for customer sentiment. We've also looked at how we can reduce event fatigue for some of our researchers, and we've started using Amperity for those customer experience use cases as well, not just for marketing.

So for those of you on the Amperity platform — the Visual Segment Editor, or VSE. For those of you about to embark with Amperity, one of the most challenging things for us was moving from a platform where all the developers creating campaigns were SQL developers. We tried to get them to start using the Visual Segment Editor, which allows you to build out your campaigns by selecting various attributes in your Customer 360 across your data sets. It was quite challenging. They didn't quite trust it yet. They knew the data better, they didn't want to use a little GUI — they were coders. So it did take some time. We were 100% SQL-based when we launched in January 2025. But over the year, I've challenged Venkat and his team to use not only the VSE but also Segment AI.

We are approved for three of the AI capabilities within the platform. The most complex thing was an audience we were creating for prospective mailings every quarter — it had a lot of different attributes, a lot of different criteria on who to include, who to exclude, very specific data points. We had all those built out and in attributes. It allowed us to take those 50 different data points, put them into the Visual Segment Editor, and build that out very quickly. Now we have a reusable audience that can be tweaked if we want to fine-tune it, or just keep it running every quarter for our direct mail campaigns.

The Visual Segment Editor also allowed us to create what we now call on-demand audiences — audiences that don't need compliance approval for regulatory purposes, but are just communications. So if we want to send a communication to our platinum customers about an upcoming event, or send thought leadership to our business banking customers, we now have on-demand audiences that are 100% built through the Visual Segment Editor. No SQL coding in the background.

We've come a long way, Venkat. So let me give you some advice. The footnote here is: prioritize and focus, but be agile. I always have a well-laid-out plan, but be ready to change — at any moment we may go a different direction. But always keep the same milestones and goals in mind. We were focused on infrastructure replacement. We had an existing CDP in place, we knew we had to deliver a certain number of campaigns that were already running live, and we had to identify the must-haves in order to meet that timeline. We couldn't say, oh, it would be nice to have this data set. We went through every single source system and said, no, we'll do this later, we'll do this later. The prioritization of must-haves is critical.

Raul said we should just have one use case. I said we can't do that — we need to shut down a system, we're going to do all these use cases. He said, are you sure? I said yes. So thank you for believing in us, Raul, and taking us along the way. That core data foundation — we started with those 30 source systems we identified as critical. We brought in the raw data. We didn't feed it from a data warehouse; we brought it from the authoritative source. We identified all the PII, identified supporting data, and brought all of that into Amperity to build out the Customer 360. We actually have two grains of data in our database: the Amperity ID customer grain and also the account grain.

We need to be able to report at both levels — how many accounts a customer may have — and we may also need to communicate to them at the account level when they have specific things going on with a particular account. So we do have two grains of data there. The big one for us was the decision to do data sharing. At the time, Data Bridge wasn't quite there yet. So we set up a Snowflake secure data share, and it's been very successful — we just set it and forget it, and data continues to ingress daily. That was a big win for us. We did need to get it vetted and ensure we could do disaster recovery, so that if our Snowflake instance failed, Amperity could fail over their secure data share. We went through some testing efforts, got that approved, and that was a big win.

The most challenging thing when we first got started with Amperity was the project management aspect. Believe it or not, it wasn't the technology, it wasn't the data, it wasn't the platform — it was how we would align on the implementation plan. We operate in an agile manner: two-week sprints, we plan a quarter, and we meet those goals. Amperity was more waterfall — they wanted us to collect all of our requirements and tag all of our PII from 30 source systems. I said, we won't even get that done by November. What can we do right now? What can we get done in the next two weeks and the two weeks after that? We finally came to an agreement that we would scrap Amperity's plan and go with my plan.

We did go with the agile approach. What was fabulous is that Amperity was very flexible — they actually brought one of their engineers onto our team as an augmentation, to be our guide. They were on our Jira board and getting assigned stories to work on alongside our team. It was very successful, and I'd do it again that way. Finally, we streamlined the ownership of how we would share information back to Amperity. When we first came in, I have a team of 20 — we had data stewards going out looking for information. Amperity wanted to sit with our subject matter experts in, say, the mortgage system. The response was: no way, they can't come see our systems, it's proprietary, they need to leave the room.

So we learned that even with different team members relaying information, things would sometimes come back slightly different — like a game of telephone. We decided we would have one point of contact on my team that would take all information to Amperity, bring it back, and then disseminate it. Later it became more of a group approach, but in the beginning it was critical to have one key contact working with Amperity when it came to decisions.

Some practical advice: first, prioritize. Prioritize your use cases and know what your acceptance criteria for success is. When is it done? When have you completed the task you set out to do? Set that early. And be agile — and I'm not just talking about agile as a framework. I mean individually be ready to change. Keep your goals in mind, but you will need to adapt as you're learning the platform. It was something very new for our team. I told the team many times: we're going to pivot. Wait — we're going to pivot again. We were the best tango dancers after six months — it was unbelievable. But through that flexibility we were able to accomplish our goals, and the platform allows for that.

When you meet a blocker — and it's usually not an Amperity blocker, it's going to be on your own team or with your subject matter experts — be ready to move on to the next thing until that's resolved. Also adopt a way of working that aligns with your team but allows for continuous improvement, and document everything. Any time we made a decision, we had a decision log. Any time we ran into an issue, we had an issue log. Any time something needed research, we had a research log. All three of those are still alive today. Sometimes those issues aren't critical; sometimes they get raised to the top and we put a team on it and resolve it immediately. But document everything, because you won't remember four months from now why you decided to do something, or why you left out a piece of data because of data cleanliness or quality issues.

That's my biggest piece of advice. My team documents everything — they know they can't get anything off their Jira board unless it's documented. But it's really helped us share knowledge with all the team members using the product and understand why we arrived at certain conclusions. And the final point: avoid the perfection trap. Do you really understand what's required of you from a data governance perspective when you get started — from a risk, compliance, and privacy perspective? Your opt-outs, your calling rules, your regulations. Do you have everything you need to identify a cell phone number? Make sure you know all those things upfront and engage early with the teams that manage those processes.

So what we know versus what we need to know: we have now identified our 5.4 million customers. They all have Amperity IDs, including businesses and consumer individuals. We also have 100 million prospects loaded into the system and 14 million Dun & Bradstreet prospects loaded in. So we have over 133 million Amperity IDs being stitched on a daily basis. All of that is tied to our account data, which is critical in giving us that first level of understanding — our product mix and how we're engaging our customers. But where we want to go next is that product penetration, the cross-sell capability, being able to understand where they are in real time. Those people at the airport — that's where we want to go. And I think Amperity, with our innovative solutions, is the best partner to have in getting us there.

On AI — one thing about the bank is that we do have a comprehensive program. Our Chief Data Officer was very upfront that we need to understand AI and use it responsibly. He set some best practices in place and has even set up competitions. There's been a Co-pilot competition with a race car format — encouraging the best use cases, which are showcased on a regular basis. There's an AI learning hub where you can sign up, take classes, and participate. There's also an AI Center of Excellence. You do need to register all of your AI use and it goes through the third-party risk management process. We did that early with Amperity, and re-engaged again as the new AI capabilities came out — they just went through another approval process through third-party risk management. All of those things go into allowing us to use AI responsibly. It does take a little bit of time to get through the model risk management process — approximately four months. But my encouragement would be: if you stay on it and make sure you're driving home the use cases it's required for, you'll be very successful.

So this last section I want to talk about is a chain of positivity — some of the small, measurable impacts we've seen thus far. There's 133% first-time right contact. That sounds like a very obscure KPI, so let me explain. We have a call center that gets complaints, and our complaint system isn't great — it doesn't identify who's calling, they just tell you who they are, and it only captures maybe an account number. We wanted to do some surveys and reach out to customers that had a particular incident on their mobile phones — maybe they couldn't perform a particular action. We were given just random data, whether it was a credit card account number or whatever it might be, and were asked to find out who these people are.

When we did this, we used the Customer 360 and were able to match 133% more than they'd ever matched before. They called the people and it was the right contact the first time. It was a huge win for us because it was the first time we used our data outside of marketing, and allowed other people in the organization to see the benefit of the Customer 360. The 36% lift in Google match rates — we just validated that last week. We're also seeing 40% more campaigns activated in 2025, our first launch year, than we had in 2024. That's huge — it kept our campaign team busy. And we saw a 4,900 man-hours reduction — essentially a productivity gain equivalent to two full-time people.

For our long-term vision — honestly, we want to do everything that was talked about this morning. I'm very excited about the possibilities with Amperity and all that can bring. But most importantly, we want to engage with our customers where they are. We want to understand their journey, understand their needs, and educate them if need be. We've been working with the customer data agent — we're on private preview for that already — and it's been very impressive. I've been working with Liz and Carly on strategies around how we can use data to bring insights around these different life segments. It's been a really rewarding experience, seeing what's coming out of our data — that core data foundation — and the insights and strategies it can bring forward. So that's all I have for today.