Enjoying the podcast? Don’t miss out on future episodes! Please hit that subscribe button on Apple, Spotify, or your favorite podcast platform to stay updated with our latest content. Thank you for your support!
Small business data is very different to consumer data. When a person is born, they don’t need to go and open a bank account, obtain a line of credit or apply for some kind of permit. But a business may be three days old and want all those things. This creates a data challenge. This new business has little to no digital footprint and so the available data may be negligible. Which makes it hard and risky for banks and fintechs to work with them. Enter Middesk.
My next guest on the Fintech One-on-One is Kyle Mack, the CEO and Co-Founder of Middesk. They are focused on business identity data and they have spent a great deal of time and effort making it easier for banks and fintech to verify any business, including those that might only be a few days old. How they are able to do that makes for a very interesting discussion.
In this podcast you will learn:
- The idea that led to the founding of Middesk.
- Why solving for business identity is so difficult.
- Middesk’s three core product offerings.
- How they moved from fintechs to banks, to the mega-banks.
- The shockingly low percentage of new business bank accounts that are approved automatically at banks.
- The details of the report they send to their customers for each business.
- How they work with the business data they receive from state and local governments.
- The huge challenges of obtaining and working with this data.
- How they interact with those companies helping to create new business entities.
- Why they created their Address Risk Insights Tool and why it is important.
- How one address in Wyoming has 375,000 businesses “operating” there.
- How Middesk is working with lenders today.
- The scale they are at today.
- Why the incumbents have not created an offering like Middesk.
- Kyle’s vision for the future of Middesk.
Read a transcription of our conversation below.
FINTECH ONE-ON-ONE PODCAST NO. 515 – KYLE MACK
Kyle Mack: So if I form a company today, most likely within about a month after forming that company, I’m going to try to get a credit card, I’m going to open up a bank account, maybe I’m going to need to apply for a loan or something like this. So, there’s a very narrow window to be able to surface information on that company and build a perspective on whether that company is a trustworthy business or not. It can take six to 12 months for information to be made available. And during that six to 12-month window, that’s when these businesses are trying to open up these accounts. We felt that’s a problem that can be solved. We can go out, aggregate this information, and use it to help our customers mitigate risk.
Peter Renton: Happy New Year, everyone, and welcome to the first show of 2025. This is the Fintech One-on-One Podcast, the show for fintech enthusiasts looking to better understand the leaders shaping fintech and banking today. My name is Peter Renton, and since 2013, I’ve been conducting in-depth interviews with fintech founders and banking executives. On the show today, we have Kyle Mack, the CEO and Co-Founder of Middesk. Now, Middesk is all about small business data, and they work with fintechs and banks to verify business identity. When you are opening a new business bank account, the KYB process, the know your business process, is often a friction-filled and painful journey. Kyle explains how Middesk helps to remove that friction while, at the same time, providing detailed information on that small business. To do this, they have built a very sophisticated data-gathering operation that Kyle shares in this interview. Now, let’s get on with the show.
Welcome to the podcast, Kyle.
KM: Thanks for having me.
PR: My pleasure. So let’s kick it off by giving the listeners a little bit of context and background. You didn’t start your career in financial services. So tell us a little bit about what you’ve done and how you came to Middesk.
KM: Yeah, of course. So, prior to Middesk, probably the most interesting story, I was at a company called Checkr. And so Checkr is doing pre-employment background checks and had, at the time, built the first API for, I think, criminal screening and driving record checks. And we had this problem we called customer credentialing. So, for every new customer we onboarded onto the product, we had to put them through some due diligence process. They will get access to people’s criminal history and dates of birth, social security numbers, et cetera. This became a huge bottleneck for growing the business. It was a people-powered process. At Middesk initially, we said, let’s build the credentialing APIs. That’s where we had started. And we wanted to take everything that we learned about identity and risk of people while we were at Checkr and then apply that to looking at companies. And we thought if we could do that, we could make it easier for companies to get access to not just products like Checkr, but a bank account and loans and payment systems and all of this. So, we really came at it first and foremost from the data and API side.
PR: And so when was the aha moment or the time that you decided that this could be a real business in and of itself?
KM: When we started the business, we spent a lot of time looking at the solutions in the space. And really what we found was our industry, Middesk’s industry was owned by a very small number of incumbent companies that had been in business for almost a couple hundred years at this point. They all had some of the same challenges, which ultimately got to the quality and the fidelity of the information available about businesses. They were difficult to work with and difficult to integrate with. We felt like this problem could be solved. It is, at its core, a data problem. The information on businesses exists in the world. So, we felt this was a space that was ripe for disruption. As we started building the product, we spent time with the largest banks and the largest fintech companies, and we found they all had the same challenges around business onboarding, which was about building a clear perspective on who their customer was and getting that customer through the onboarding process so they could start using it with as little friction as possible. So, the more we get into it, we see a space that’s ready to be disrupted with a growing market and a problem that we think we have a unique perspective on being able to go out and solve.
PR: What is the big challenge or the big problem for business identity today? Why is it so difficult?
KM: So, if you think about a person like you or I, we’re born, maybe like 15 or so years later, we get a credit card in our name, and eventually we buy a car or get an apartment. But in that span of time, there’s a lot of opportunity for data to be created about you as a person and for you to build this footprint of information. Businesses are very different. If I form a company today, most likely within about a month after forming that company, I’m going to try to get a credit card. I’m going to open up a bank account. Maybe I’m going to need to apply for a loan or something like that. There’s a very narrow window to be able to surface information on that company and build a perspective on if that company is a trustworthy business or not. That has a lot of ripple effects. One is from the perspective of the actual small business from the business owner. They’re trying to build a company, and they need these products like a bank account just to be able to get their business off the ground. But there’s a lot of friction for them in getting access to those products, and they have to spend their time on that instead of building their business. From the perspective of the bank, who is our customer, they deal with conversion rate problems. There are a lot of documents going back and forth that are leading to impacted conversion rates, and it degrades the customer experience as well.
So why does this happen? I think that gets to the core of your question. Our space is one that’s owned by companies that have been around for a long time, as I mentioned. Underpinning those products, there are a ton of gaps in the quality and the fidelity of the information. It can take six to 12 months for information to be made available. And during that six to 12-month window, these businesses are trying to open these accounts. We felt, again, that’s a problem that can be solved. We can go out and aggregate this information. We can use it to help our customers mitigate risk. We felt the absence of a trusted, reliable data company for business data is leading to a lot of these customer experiences and friction points.
PR: Right. Particularly given the fact that most business owners now want to open a bank account at 10 pm from their home. They don’t want to go and visit a branch to do that. So it’s even more important to have that data available. Maybe explain your core product offering and tell us about the different use cases.
KM: Middesk, at its core, is really three things. It’s a data platform that exists behind the scenes. It’s this living and breathing data platform where we’re pulling in information from all sorts of places. We’re structuring it and using it to derive insights about companies. Then on top of that, it’s an API product. So our customers can integrate that API into their risk models, into their onboarding flows. And then it’s a dashboard product. So if I have a team of risk analysts or an operations team, and they need a place to review information and make decisions, they can do that within Middesk. So that’s what Middesk is.
We help our customers broadly manage four risk types: regulatory compliance, fraud, credit, and reputational risk. So, depending on my use case, I will have a set of questions. I need to know who the beneficial owners of the company are. I need to know the industry they operate in. I want to make sure that they have a reputable online presence and that they have never filed for bankruptcy. So Middesk is sitting behind the scenes, and we’re giving them the answers to those questions in a second, two seconds. And then they’re able to use that to make automated decisions.
A real-life example of our customers is going to be, let’s call it, a top-three bank. We work with two of the top three banks today. An LLC, Kyle’s Coffee Shop, is going to open a bank account. They’ll submit some information about their business: EIN, legal business name, address, et cetera. The bank will send us that information. We’re using that information to pull together this report. They use that report to make decisions on their customers.
PR: It’s interesting you’ve got two of the top three banks, because you’re not an old company, and I imagine integrating into a bank’s flow is not a trivial task. And the fact that to get engineering time from their in-house teams to make this happen wasn’t an easy thing either. Did you start with the fintechs? What was your evolution? What’s it been like?
KM: Yeah, great questions. Our earliest customers, as you can imagine, were the fintech companies. They’re the folks who are going to be the earliest adopters, right? And they were really excited and willing to work with us on iterating on the product. So our early customers were companies like Plaid and Brex and these types of businesses. And then over time, though, we started having conversations with the banks. And what we found was, yes, slow-moving. Yes, hard to get engineering resources. But the scale that the banks were operating at was so significant that the inefficiencies around business onboarding were a real pain point for them. It was not uncommon for a bank to approve only 10 or 15 % of the businesses that apply for a bank account automatically.
PR: Wow. That’s terrible.
KM: It’s really inefficient, especially when you compare that against the efficiency of opening a consumer account, like for you or I, a bank account where you could be, you know, in the eighties or nineties or, something like this. The performance was really, really rough. So we started with the fintechs. We spent a lot of time with the banks. We work a lot with the digital identity and product teams at these banks. I think one thing that we’ve done at the company that’s allowed us to be able to make significant headway with some of these more traditional companies ultimately gets back to our strategy, which is to own the supply chain of data, to work with all the different government agencies to get that information on file, and keep it up to date. The reason that that’s important is it allows us to have a very clear explainability of where information comes from and how we control its quality and accuracy. Then, those banking customers were able to pinpoint very specific parts of their onboarding process where they were able to plug Middesk in. They didn’t have to rip and replace the front end, kind of the UI of their experience. They didn’t have to fundamentally rethink their identity programs. They were able to simply plug our APIs in to be able to enhance their existing stack of vendors and their existing system and see immediate performance benefits.
PR: Right. So, what are you returning to the bank? Particularly in the onboarding process, a new business, are you returning a score, or different data points? What do you send back to the banks or the fintechs?
KM: So we typically first will receive some information. We might receive the business’s tax ID, the company’s name, address, and a fairly simple set of inputs. We use that to assemble a report that is made up of all the information that we have on file about that company, and we return that report to them. The report is a combination of a few things. It is a lot of raw data, and they can use that data to do things like pre-fill form applications or just to enrich their understanding of the people associated with the company or kind of who this business is. It’s a set of insights, and those insights are sort of a set of business rules, and the rules will map back against their policies around compliance, fraud, creditworthiness, and reputational risk. I might say I need to verify certain pieces of information about a company. I need to make sure that the company doesn’t have any active lawsuits, you know, liens being held against the business. We provide them with a set of insights, and then they can use those insights to decide whether they want to approve the account, reject the account, or request additional documentation from that business to prove certain things about themselves.
PR: So is this designed to be like an automated workflow without human intervention or is it designed more to have someone come back and say, “Oh, I want to read this report.” What does it look like?
KM: Yeah. So the best customers for us probably care about three things. They really care about the customer experience and try to make that as frictionless and seamless as possible. They’re probably trying to open accounts at scale. I’m talking on the order of thousands or tens of thousands of accounts in a given month. Obviously, we have customers serving a smaller volume, but we’re really trying to open things at scale. And they are customers who need to mitigate some form of risk along the way. So, to your point, for sure it is businesses who are trying to make some automated decisioning, and they want to be able to get people out of the process, and they’re able to have those teams instead of focusing on, let’s say the middle of the bell curve, the majority of businesses, they’re able to now focus on the edge cases and the businesses who potentially pose the greatest risk, where today there’s just been a lot of time spent on all of the businesses that are applying for accounts. Automate what’s obviously good, and focus on the things that really require more attention.
PR: So the report comes back from Middesk, and everything is in line, then they just keep going on with the onboarding process. If something looks a little strange, then you kick it out to a human. Gotcha. Okay. I want to talk about the data because you are working with government data, the data that’s sort of, at secretaries of state. And, I imagine that that’s not easy to get, and it’s not easy to work with. Tell us a little bit about that process of how you’re working with them.
KM: So let’s think about the history or timeline of starting a company. Let’s say I’m opening a restaurant. I’m probably going to set up an EIN with the IRS, my employer identification number. Then, I’m going to form probably an LLC. I live in New York. Let’s take New York. I’m going to form an LLC in the state of New York. Now I have that entity. Then I’m probably going to go and apply for a food and beverage permit or a liquor license or something like this. I’m going to create a website for my restaurant; that restaurant is then going to be listed on something like Resy or OpenTable. So there’s this footprint of the business. Our goal is to assemble all of that data. It’s a combination of primary source government data, alternative data, things coming from the internet, and then data that we aggregate through the usage of our own products. The foundational thing we do is the government data, which you’re talking about. Over the first few years of the business, our goal was really going and setting up relationships with state and local government agencies to be able to acquire on a regular basis all the information that they have on file about a business. So if I form my LLC in New York today, we need that information, preferably today, if not today, hopefully tomorrow. And so the way to do that is to go and set these relationships up so we can get all the new businesses and we can refresh everything. So we have hundreds of millions of records on companies across all sorts of government agencies. And then we’re building all the relationships between those records and trying to keep it up to date on an ongoing basis.
PR: I imagine there’s not one government agency that you’re working with today that has an API connectivity you can just pull. Is that correct?
KM: Yeah. I mean, the beginnings of doing this was CDs getting sent through the mail, was us buying hard drives and sending the state agencies hard drives so they can provide us the file and give it back to us. And, you know, we’ve been able to mature that over time, but it is the epitome of an unsexy picks and shovels business about going and getting these relationships set up and then trying to scale that over hundreds or thousands of different government agencies over time.
PR: And even like a secretary of state, right? Because every state, there are 50 of them, has a secretary of state, and there’s no uniformity, right? There’s no national standard. This is how a secretary of state needs to process their data. Right? So you’ve got 50, I imagine 50 plus different formats coming into Middesk.
KM: Fifty just from the states and hundreds when you look at other state agencies and local county agencies and someday thousands, as we continue to grow. That, at its core, is the technical problem to solve for Middesk. You have a name and an address of a business, and we need to go out and look across hundreds of millions of records and find just the records that belong to that company, where there are no unique identifiers across any of it. Behind that are many data pipelines running, each with its own file structures and formats and ways of getting the information and different ways of keeping it up to date. At its core, that is the data platform we’ve been focused on building.
PR: So you can’t even rely on the EIN like you would for a consumer social security number to be in every single database?
KM: No. Actually, only one or two states in the country make the EIN available in the state offices. Even the IRS and the Secretary of State have no connective tissue between them. You actually have no unique identifiers. You’re dealing with names of people, addresses, and names of companies.
PR: So do you work with IRS data as well?
KM: We have a service that hits the IRS directly. They have some services available. And they also make publicly available data sets around certain tax filings and tax forms that we work with. We really are trying to hit federal, state, local, and either county or city-level authoritative government data and be able to pull that all together into a single place.
PR: What about the companies these days like Zen Business and LegalZoom that really make it easy for you to set up your company? You just do a quick online search, and they come up right away. Are you working directly with those companies or not?
KM: We don’t work directly with those companies. It’s a great question. They are a part of the puzzle, the puzzle that exists around business identity. To your point, there are these formation services that have made it really easy for founders, whether here in the US or internationally, to form a company, register it with different government agencies, and then get things like a virtual address and a bank account and some things like this, depending on the company and depending on the state, they are the ones filing that paperwork. They also might be the ones representing the founders of that company. In some cases, they actually make it very difficult to know who is associated with a company or where a company is actually operating because they can operate as what’s called a registered agent. This company can represent themselves as somebody who’s associated with the business. They actually mask the end owner of the business data, which for a bank is a huge problem because one of the banking regulatory compliance requirements is to understand who the owners and officers of a company are. So, while we don’t work with them directly, we certainly have indirect exposure to them because they’re a part of this messy world of forming and registering businesses with the government.
PR: Right. And particularly if you’re opening up a bank account, the bank wants to know details about the company, but they also want to know details about the individuals opening it. So you’ve got to have a whole bunch of consumer data as well, right?
KM: We have information on the people associated with companies. It’s pretty shallow. It’s usually names and titles. But we’re working on enriching that and building that out over time as well. So, you know, what we want to understand is, it’s a three-sided problem. Who is the company? Who is the person who’s creating the account on behalf of the business? And then probably the hardest leg of this problem is, does the person who’s creating the account actually have the authority to do so? So it makes sense for Kyle to create an account for Middesk. But it wouldn’t make sense if Peter created a bank account for Middesk. Our goal is to focus on making sure that the company is real and then be able to, as accurately as we can, find that relationship and determine that a person has the necessary authority to create a given account for a business.
PR: Right. I want to talk about addresses. Everyone has an address. Every business must have an address. And you recently, we’re recording this in mid-December, and you recently launched this Address Risk Insights Tool. Tell us about that and why it was important.
KM: Yeah. So if I’m a bank or a fintech company, my core set of compliance requirements is to understand the name, the address and the beneficial owners, the people associated with the company. As it relates to the address, I really need to understand where a company is operating. The reason that that’s important is I need to make sure that the business I’m opening an account for is not operating in China, Russia, or Iran. I have a set of sanctions lists that I need to run against, and cartel lists, right? So I need to understand truly who this company is and who’s associated with the business. Understanding where a company actually operates is really difficult. One of the biggest reasons it’s difficult is because when somebody goes to form a company, the state governments require, and some states are more challenging than others, require a very shallow amount of information to create that account. Wyoming is probably one of the worst offenders. Wyoming has tried to become for LLCs what Delaware is to corporations. They allow somebody to create a company in the state of Wyoming without actually having to provide any information on where the company actually operates. They can just represent a company’s address using their registered agent’s address. It could be LegalZoom or ZenBusiness or something like that. And so we built a whole set of insights to help our customers understand if the addresses that they have on file truly belong to the company and represent where they operate or actually might be the address of a lawyer, an accountant, or a third-party registered agent. To give you a sense of how prevalent this problem is, in the state of Wyoming, there’s a notorious address, it’s 30 North Gould Street. If you Google the address, you’ll find articles about the businesses that have defrauded consumers that are apparently operating from this address. They’re not really operating from the address. It’s the address of their registered agent. We know that today, there’s over 375,000 companies that have formed from this address. And if you were to look at it, it’s crazy.
PR: Wow! What is it? Is it a lawyer’s office?
KM: If you were to look at this address, it is a small brick building that is actually the office address of one of the larger registered agent companies. But that address is used to open up bank accounts, to onboard onto payments rails, and all these other things. Our goal with Address Risk Insights is to bring some transparency and clarity to that type of information.
PR: Right. It’s perfectly legal for a company to set up their business at 30 North Gould Street in Sheridan, Wyoming, but there also are a whole bunch of fraudulent companies there. So this address in Sheridan, Wyoming, makes it more risky, but what you’re trying to do is get beyond the registered address, right, and find out where the actual address of the people that they’re operating from. Is that right?
KM: Well, to your point, it’s legal for the entrepreneur in the small business to say that that’s where they operate. But actually, where it is more challenging is for the bank or the fintech company. They have a legal obligation to actually understand where the business is operating from. And they need to know if the addresses they have on file truly represent where that company is operating or not. And if not, then they need to put the necessary policies and friction in place to understand where that company is operating. Because in the absence of doing that, they don’t really understand the operating nature of the business and where their footprint is. And so, from a regulatory standpoint, it poses a lot of risk to them in begging the question, “Do you really know who your customers are if you don’t actually even know if they operate in the US or outside of the country?”
PR: Right, right. Okay. I want to switch gears and turn back to credit because we haven’t talked about that yet. And I’m interested; you mentioned that as one of the use cases. Are you providing the lenders just the KYB process, or is there anything else? What are you doing with lenders?
KM: If I go back to the four main types of risk: compliance, fraud, credit, and reputational risk, our goal is to build products that underpin all of them. Compliance is our bread and butter today. On the creditworthiness side, we’ve started to dip our toe here. So we can help a lender understand if the company has ever filed for bankruptcy. We can help them monitor to see if they file for bankruptcy for their portfolio. We can help them see if there are any lawsuits or judgments for that business. If so, how much money the business might actually owe to another counterparty in a lawsuit. We can help them see if any liens are being held against the company, where they might have some creditors they owe for certain things. We can help a lender file a lien on a business. If they decide to underwrite the company, we, Middesk, can help them secure that loan by filing a lien on that business, and we’re sort of the API behind the scenes. So it is both the data to power some of their modeling and their insights and then increasingly step into parts of the actual origination workflow. But again, as the API layer that sits underneath the scenes.
PR: Right. Okay. Can you give us a sense of the scale you guys are at today? When did you sign your first client, and how big is the team, that sort of thing?
KM: Yeah, of course. Our six-year anniversary party was last week, and so about six years in business. We started the company in January 2019 and went through Y Combinator; our first customers were companies like Plaid and Checkr, actually were the first two. So if you fast forward to today, we support about 500 customers today, including two of the top banks in the country, as well as many of the big fintechs you would think of: CashApp, Shopify, Ramp, etc. The team’s about 90 people and growing. And today, we’re helping our customers evaluate and onboard millions of end businesses every year. And again, that number is growing as well.
PR: Do you charge by the API call, or what’s your business model?
KM: The business model is simple. We charge on a per-request basis, per API call. For each business that’s onboarded, depending on the types of insights that we’re helping one of our customers look at, the charge is shown on a per-request basis.
PR: Right. Okay. So, you mentioned some of these really established companies. I know you didn’t mention any names that I am going to. There’s LexisNexis and Dun & Bradstreet, both really well established, providing great services for, in some cases, like you said, centuries. Why haven’t they done what you have done, do you think?
KM: I think, to your point, this is a space owned by the big incumbents, and there’s a small number of them. There are things that they are actually really good at, things that we today are not able to win on. They will have a far greater breadth of data than we have because they’ve been aggregating it for 100-plus years. They are going to have better brand awareness than we have, at least today. But there are some things that we do have the ability to do. We can move much faster from an innovation and execution standpoint. When we see new fraud patterns emerging in the market, we can move much faster to adapt our product to help with those things. We have extreme depth and focus around a much smaller number of business problems, so we’ve been able to go super deep on them. We can focus at the experience layer, not just the dashboard for risk teams, but also for developers, and really build an API-first developer product. And so we think all of that combined with our focus on owning the data supply chains from start to finish really leads to us building a lot of trust with our customers. And that, for us, is really the way that we win. And it’s all about building this business from first principles on a modern data infrastructure that can really scale and focusing on that supply chain strategy so we can control things from start to finish and do it in a way that really can be easy to integrate and easy to scale.
PR: Okay, so last question then, what is your long-term vision here? Are you looking to be the Dun & Bradstreet of the 21st century?
KM: I think that’s a fair comparison. Our goal is we want to be a global business identity company. We want to look at every B2B use case, not just in financial services, but every single B2B use case that exists. We want to understand what are the questions that somebody needs to answer to be able to onboard and transact with their customers. And then we want to build a set of infrastructure products that help them do that. And we think if we can do that, it’ll make it much easier for businesses to work together. It’ll lead to a better customer experience. And we think it’s a good thing for the world. It’s a good thing for the small business owner. It’s a good thing for our customers. So that’s really what we’re trying to do, but it’s certainly to go after that problem globally and really be that de facto business identity standard.
PR: Well, best of luck to you. It’s a great product you’ve built, and you know, America continues to have more and more small businesses get started on a daily basis, so there’s more and more need for a company like yours. Kyle, it was great to chat with you. Thanks so much for coming on the show.
KM: Thank you for having me.
PR: As someone who has founded five companies, I love any product that will help make the new entrepreneur’s journey more efficient. Because when you start a business, you’re excited to get moving on your new business. The last thing you want to do is take two hours out of your day to sit in a bank branch while opening a bank account or start opening a bank account online and literally be told you need to come into the branch to finish the process. That is something that should have gone the way of the VHS tape. It is sad to say that it is still the norm at many, dare I say most, banks today. But not for long. The technology that Middesk has created will eventually make this long and painful offline process a thing of the past. Anyway, that’s it for today’s show. If you enjoy these episodes, please go ahead and leave a review on the podcast platform of your choice. And thanks so much for listening.