Tanya Kohen: Let’s turn our focus to the startup world, an environment that practically breathes invention and its products and business models. But there is a common silent growth killer lurking in the background, the tech stack.

In the race to scale, many startups are, as the saying goes, building the plane as they fly it. They grab a tool for CRM, one more for accounting and each solving an immediate urgent problem. But over time, this patchwork of solutions becomes a tangled mess. It creates data silos, introduces security vulnerabilities and bogs down teams with manual workloads. What starts as a necessary shortcut ends up holding the entire company back from its potential. But how can founders and finance leaders apply an inventive mindset to this foundational challenge? My guest today has seen this story play out time and again and has a powerful framework for turning each chaos into strategic advantage. I'm absolutely delighted to welcome Maria Frolova to Invention Mode. What makes Maria's voice so essential is the unique bridge she builds in her work as a fractional CFO.

Before we jump into the interview, tell us a little more about your work, about what you do as a fractional CFO, about your work with startups and how you actually transition. How did you build this bridge between your investment banking career, your corporate career and the startup world?

Maria Frolova: That's a very interesting detour indeed. I spent 11 years with J.P. Morgan working in person in New York and then in Europe. And at that point, I was working mostly with large-cap companies of the scale that they're about to go public. So we do IPOs for them or buy something in the multi-billions. So I have seen finance, I worked with finance, but I didn't necessarily know what's under the hood. I would see the financial reporting notes, whatever financial models that they develop, I would rework that information.

But I never knew how this thing had happened. So that was always a bit of a point of curiosity for me. Then I transitioned to private equity and I had this pretty crazy project when we bought 11 companies, slapped their holding and the idea was to go public as soon as possible. And at that point I discovered that it takes about 270 days to produce a financial statement. It turned out that the main problem was basically finance. It was 11 disjointed finance functions with nothing in common. Not even this common chart of accounts, no KPIs, no tech stack. So we had to postpone the IPO, but we had to rebuild the entire tech stack. And that's where I took my first interim CFO role and I found out that I don't know anything about finance. As a banker, it was a bit of a disappointing thing. I thought I knew all about finance, but apparently I knew a different type of finance. That's how I got curious.

So when I moved to Silicon Valley, I wanted to be on as an operator, I wanted to build those functions and use the experience I got through this project of consolidating entities. But it was a huge discovery process for me, you arrive and this is a brand new company, and you're building a plane, which means there's nothing in finance. Where is everything? There's nothing. There's Excel, and there's usually pretty poor accounting because founders, most of them, are with technical backgrounds, engineers. So they get served the financial statement and they look at it and it wouldn't be very useful. It goes how it goes, it's typically an outsourced accounting. So I decided that what I want to do is to build the financial engine that is scalable because finance is an area where you have a view into every function of the company. That makes it pretty unique. If you do it right, you have a lot to say in terms of strategy, have a lot to say about efficiency, you have a lot to say about what you're investing money in, because you can't invest in everything at once, which a lot of startups tend to do, because your cash balance is finite and the uses are not.

This is a very rewarding part. It is very unstable in terms of risk. A lot of startups go out of business and all the efforts you put into it and your equity becomes less valuable if not zero. Having said that, it's extremely rewarding when you are laying a foundation that would take the company through the very early stages to sustainable revenue and then to scaling. The business model is becoming more complex and you hope you're evolving with it as a finance lead and your finance function evolves as well. So you kind of integrate yourself to become instrumental to the company's strategy. In terms of personal impact, I think it's a way better deal than either sitting in a large corporation doing one specific function, just because you get to do pretty much everything — GTM analytics, sometimes I used to be data head as well, it's a different discipline, but I'm the one with numbers.

In terms of what I do as a fractional CFO, it could be anything. I could come in and just build it from scratch. Because as the companies approach Seed stages and they actually get some money to manage, that's when they need a fractional CFO. A lot of founders don't realize that, but it's true.

Tanya Kohen: There's a lot to unpack here. And I like how you said — just lay the foundation. I think that's so important. It's easy to miss, and then you are building on a poor foundation, many processes can go wrong in this scenario. I can also relate to the changing perspectives. Me personally, when I first started working with WaveAccess — which is the company I'm with right now, they're software developers and IT consulting firm — as a consultant, I only had my corporate experience. That was one perspective. Then, when we started working more with startups and projects, when founders would come in with their big idea and we sit through a workshop with them, the product design from the IT standpoint, and then they start thinking about their business model” “This is what we need from the tech standpoint to support our future business model”, it's mind blowing. It changes many things for you and you expand your horizons as you dive into projects like that. I can definitely see how it's an exciting experience for sure.

Maria, you've worked with startups from Seed to Series C. At what stage does the tech stack debt typically start to become a real problem in your opinion? What are the first warning signs like manual workarounds or data inconsistencies that founders often miss?

Maria Frolova: You got the two main ones, I think. I remember vividly I was trying to build GTM analytics for a company. It was after Series B, so I was hoping to get clean data and some consistencies. I pulled one metric from three different dashboards, and it was different each time. That's when my heart sank and I realized that problem is probably bigger than I'm seeing it. And it turned out to be absolutely true. There was no even common definition of the metric — different teams understood it completely differently and that's why it was designed differently. Sources used were different, attribution models were different. It turned into a huge data cleanup process. So when you see this happen — three places produce the same metric, the answer is different — you're in trouble, it's a big cleanup.

Another one is the manual processes. What is our scene in finance, because we're looking at everybody's efficiencies except ours. As the company grows in complexity, we take care of everybody, but not ourselves. So you find that close is actually certain tasks that take longer and longer and longer. First you had to merge two Excels, now you have to merge three Excels, then four, then you have to call the data team so they merge it for you. And rarely people address those issues practically. What I've found in my experience, and I am not very disciplined about it, but what helps is to take — at least on a quarterly basis — a nod at the various tasks and standardized processes and see which one takes the most time. If you see a dynamic that something is taking a long time and longer and longer with every quarterly close — there's probably some room for automation. You don't know what you don't know. It fixes a lot of times way easier than you might think. When I had two roles, data and finance, I was way more efficient. I would just go to my data team and say “Can we merge this couple of databases so we don't have to do it in Excel?” And the solution was typically found. It could be implementing software, but it could be an easy fix. You just don't ask and you don't get.

I think there's also another indirect sign. For example, if everybody's using Looker or some data visualization tool and everybody's complaining about the time to load, you have problems. You have data problems. You probably have 10 data tables instead of two.

So, there are indirect signs. But what I noticed is, people are so busy with critical things when you're a startup, then you just let the problem go unless it becomes unbearable. If you find the time to be a little bit more proactive, you're probably not going to get to the stage when there's a complete breakdown of everything and people are quitting because they're overwhelmed and they can't merge another spreadsheet.

Tanya Kohen: Wouldn't you say that somewhat similar problems can appear in any size and any maturity stage of a company, right? Because it's just maybe this day and age that we live in, so much data around us and so much opportunity to use more data. I've seen pretty much your finance departments deviate a little into this route of manual, especially when, for example, a new business is opening somewhere and it's just everything somewhat like a startup world. When you have to launch a new project or a new business unit and it's a go, go, go. So it has to be so quick, there is some patchwork happening and “let's just do this on a spreadsheet, we don't have a system for it right now, or we don't have time to truly integrate those two ERP systems during the M&A acquisition.” There's no time for this right now, but then it can lead to huge, huge problems.

When we had this prep session for this episode, you said “I started thinking about tech stack.” Is it really that exciting? Is it a topic that's of that much relevance to finance? And then you yourself said that it is, it definitely is very important.

Maria Frolova: I answered my own question. And I think if I had to give a recommendation to a founder, — tech stack, it needs to be stage appropriate. What we do wrong always in finance — we don't prioritize our tech stack because there's a product related, engineering related tech stack, or revenue-generating tech stack, this gets priority. We always wait until it's impossible to wait before we upgrade. But one thing to get early is the data quality, data architecture, you don't know — when you're at a very early stage — if something happens on the backend. We could spend another week optimizing it or we can just slap another one, do a little stitch and it works, but it works only that long. The examples that I was giving you — it's not a late-stage company, they just barely lifted B round, so they're fairly young, they were maybe four years in existence, and the problems took about a year to fix. So in three or four years, you accumulated enough problems that it was a project to fix it and it took a year. And until then, a lot of things were just going wrong: Looker was glitching, data was not consistent. If you had to spend time on one thing, maybe it's not a finance tech stack, but at least — give us clean data, give us something. And every fix just adds more iteration, more problems, more manual cleans. It just keeps accelerating. So I think that data sanitation is something to spend time on from a very early stage.

Tanya Kohen: Definitely. We often hear that startups are building the plane while flying it, as we mentioned a couple of times today. While that's true for the product, why is that approach so dangerous when applied to the financial and operational tech stack? We discussed data, but especially from the data security standpoint, why should a security audit be a non-negotiable first step?

Maria Frolova: There is truth to it obviously. Let me just say that the finance tech stack is also flying a plane and building it. For an early stage company, very early stage, some basic accounting and Excel is enough. You just don't have enough data to deal with, you probably have, better if you have a cash balance of 200,000, you don't necessarily want to go and buy enterprise-grade ERP if you can invest in something else. Then as you scale, you have to add FP&A tools, data visualization, vendor payments automation, a lot of things, cybersecurity, it depends on the business office, it could be an e-commerce platform.

With every stage of the company you have to have stage-appropriate finance. A dude with Excel is okay until Seed and then you start having other needs and you have to practically invest in it. In this way we are also building the plane while flying. We have a cabin, so we're building out of our cabin as the plane is flying.

Having said that, we really don't have the luxury of trial and error at all. And the role of CFO is one of the main things is risk management and not just financial risk, risks in general. So I'm of the school of thought that CEO, CTO and compliance are responsible for the entire tech stack of the whole company. So it's not just a finance tech stack, it's a centralized roadmap, it is a strategy. And there have to be procedures.

I'll give you a pretty shocking but true example. When I worked for a company and the CEO was very excited about AI, AI was fairly new, but very hot. We have a town hall and the CEO goes “AI is the future, we tool or die, and basically, we're gonna be evaluating each and every one of you on how well you use AI.” And that's mid-year. What happened was it wasn't the only company doing AI implementations. Every team started doing their own AI tech stack, every person added their personal AI tech stack. By the time we got everything under control, we all had a micro stroke, it was a legitimate heart attack. By the time we got the strategy together, approval processes, and convinced everybody that thoughtless AI implementation might not be in our best interest. We did an audit, and I can say it was probably a divine intervention, but nothing bad happened, and we didn't end up in a courtroom. But it was literally so many risks. We had like 45 risks with high capability, and it's just only in a couple of months. In addition to that, since there was no coordinated strategy, we found out that there were three or four various AI tools used by different teams doing essentially the same thing. So we basically bought the same functionality three, four times. Obviously, that was not a very smart cost decision.

I would say, startups want to be agile, but you want to be agile within reason, right? Having 45 new security risks in just two months is absolutely possible. And especially if you're dealing with anything related to either PII, money flows or anything that's vulnerable to fraud, you really can't make it a chaotic process.

So the role of CFO is not only to watch out for the finance tech stack, but have a view across the entire company. Obviously not alone, it helps to have CTO and compliance people on board as well, but I think in the AI world — where the risks are accumulating way quicker than the solutions to them — you just can't not watch out.

Tanya Kohen: I very much agree with you. I’m very much for the agile process and experimentation, but it should be in a controlled environment, right? Because what also gives me a pause is when, for example, I see companies encouraging their people to reinvent their process on their own. And you just spoke about this. It can be, to me, a good thing to come together and talk about it. Like, let's have a discussion. Let's have a few sessions, discovery sessions. How can we improve? Which tools should we use? Let's talk about it, discuss all of us together in one room. That's productive. While just leaving this up to each person to figure out for themselves, then you end up with very much unknown and that's not a consistent process at all. Especially if the person does figure out some things, even in a good way, but then they leave. So you’re left with no knowledge and no transition of the experience of an expert team.

Maria Frolova: The documentation is another thing. I mean, the data sanitation and lack of documentation — these are sins almost every startup goes through. If a product manager departs and they have five years worth of expertise, the expertise departs. You have to document.

Tanya Kohen: It's one of the largest risks. Well, let's talk about invention a little on a more positive note. Let's say invention is being done right and all is sunny and bright. But just curious about your perspective on invention. We're kind of untangling a messy stack, like a tech stack. Do you think it's only just an IT project? You mentioned that finance and CFO probably should be overseeing the whole tech stack of the whole organization, which I agree with. And that's why, again, back to corporate studying, CTOs and CIOs often report to CFOs in larger organizations as well, because it makes sense.

But as far as invention goes, where do you see the opportunity for genuine invention in this process? Is it in creatively configuring and integrating existing tools or is it by new tools? Is it a combination maybe?

Maria Frolova: I don't know if there is one answer. All of these three can work. Fortunately or unfortunately, I've been through a lot of various software implementations, either as a head of finance or head of data that included merging data bases and text, text after M&A that included implementing Salesforce, which we had to do twice just because it was not done correctly. And we'll talk about why in a second. Replacing QBO with a net suite once you grow out of QBO, which also happens. And also building tools such as a homegrown inventory management tool, because we were in the middle of “Large ERP doesn't work for us yet, but the small solutions didn't work either.” So we had to develop something on our own because otherwise it will be way too costly for us and not enough data just yet.

So I've done them all and I think it's more about how you do it. So the invention, it's probably finding a good balance. First there’s a temptation to expect the tech stack to fix all your problems. One of the reasons that the horrible implementation of 11 companies worth of IT was done well, and we actually did it — we managed to solve a lot of problems, and we condensed the time from 270 days to close to 120 in just a couple of months, and it kept getting better. It requires a little bit of afterwork as well.

We sat down and we figured out that processes, methodology, metrics, approaches, charts of accounts, everything needs to be worked out before we put any software on it. Otherwise you will be stuck with people trying to fit the information, data, processes that don’t fit into the system, and that will not serve any efficiency, or the data will just not be reliable. If that's not the metric that you're used to, you're going to show them something that you think is the metric, but you're not sure what it is. In the end, one plus two is not three, it's maybe 17. We all want to get on the same page and with AI as well, it actually opens it up for more hallucinations. If you don't have the data or your data is faulty, do you want AI to hallucinate it for you? You might make decisions on this basis. Probably not. Same thing with processes, right? AI will not design a process that works for your business. It might, if you work together for like 17 years, he knows the business as well as you do. But until then, you need to prepare before you roll anything.

Another one is buying a fancy tool that doesn't fit your processes and trying to work around it. That's what happened to that Salesforce implementation. Unfortunately, the product team had more to say about configuration of the funnel than the sales team. So when it ended up happening, the sales team couldn't work because they had to redesign all their processes. And there were also regulatory hurdles that define the shape of those processes. So we had to just redo it. And implementing Salesforce twice, it's a bit costly. So why not do it once, for example.

But the most inventive thing — and I don't have a solution — is the disconnect between expertise and understanding of business of a business user and the technical team that will be implementing it. It's almost like if you don't have a good translator, the business rarely properly formulates what they're doing, why, taking the time to explain the metrics, trying to implement for whatever reason, a problem to solve. What does it do, what exactly, just explain in business terms what it is. And then the teams who are misunderstanding what they're actually doing, build a solution that doesn't work for the user. That happened more than once. And I guess, there’re technical writers for that, the infologic companies, I'm sure there's a translator in startups. The number of disconnects, broken telephones and everybody getting really upset with each other because after a half a year of waiting I've got served something that doesn't even answer my question makes it worse, more complicated. The number of such times is definitely more than one. Let's put it this way — every other implementation runs into one of these problems.

Tanya Kohen: I love it. So it's definitely process management, and then process owners being present at every stage of the implementation, basically.

Maria Frolova: And a good translator. You need somebody who understands both sides.

Tanya Kohen: From a fractional CFO perspective, what personal traits or habits separate the founders who successfully navigate these growing pains by building the scale from those who get stuck in just-get-it-done mentality?

Maria Frolova: The biggest one is ego. I personally always walk away from somebody who just cannot be challenged. Unfortunately, you don't always know that from the start. I have a ton of respect for founders, they are my heroes. A lot of them are engineers — “design your product” is what they do. It's a beautiful product, but then you GTM, then you need to build a business that is scalable and then you also need to make sure you don't run out of money in the process. That's so many disciplines — especially for the first-time founder. This is absolutely uncharted territory. So to become a good CEO you have to realize at some point that you can't do all these things by yourself and you can't make all the decisions just on the basis of your intuition.

There's also founders (I don't know if it's ego; I call it ego, you call it whatever you want) — basically every time you challenge them, they perceive it as a personal attack, and they will do it anyway, right? So there's almost no point in challenging them. But me as a CFO, I can't not challenge. The data says what it says, not what people want it to say. So, for those founders, I probably would recommend not to get a CFO because we will just hate each other. But there is a good probability that the company will not make it through the scaling stage. I have seen people burning through 200 million in two years to reduce revenue, that happens to. That's not listening to data.

Tanya Kohen: Agreed. What is one practical inventive step a startup leader can take this week to start addressing their tech stack challenges without needing a huge budget or a complete overhaul? What's your number one advice?

Maria Frolova: Where is that magic wand that will fix everything and spend your money? I think it's not going to be magic. That was just an example that I really liked in one of the companies I worked with… It's a good time to do it, we're going into planning processes if somebody hasn't started yet. So what one of the CEOs liked doing, and he would get all his CXOs to write three inefficiencies or bottlenecks that are major epic pain for their teams. And then, when the executive CXOs meetings are going, they would discuss. Some of them were not necessarily anything related to tech stack. It would be just a bad process or some understaffing. But inevitably among those three top pains of every CXO there would be something that somebody could find a solution like “Hey, you can hope on our system, you will get that data while you're 17 hours connecting 75 different spreadsheets and running whatever Python and whatever analysis, I can give you this data.” So a lot of times somebody else would be “OK, I have the same problem.” At that point, it becomes clear that you need to improve your tech stack and get something that's used by several teams and everybody's happy. Right?

This kind of exercises they're not magic but you would be surprised how many silos the smallest corporations get. There are silos even in a company that's under 20 employees. I don't know how that happens but it does.

Tanya Kohen: You gave a brilliant framework for things like that. There were a few times when you mentioned: think about your workflow, think about your process, think about what you’re spending most time on and if there is something that you’re spending too much time on, which should just take you minutes, but it takes you hours. That’s a red flag. Think why it’s happening. I’m with you on that because I think every person in any organization, especially in a smaller staffed organization like a startup — where there is this drive very often and this mission and the founder's personality, which is all great. It can be a really exciting time, but also it's dangerous to forget about efficiencies in the process because as you said, the money is not there infinitely. You have to move, you have a runway, and so you have to move efficiently.

Final three questions for you, Maria. What does an invention mindset mean to you?

Maria Frolova: It means the opposite of absolutely despised and hated by me “We do it because we always do it this way.” And that's how everything that you described with people just putting up with manual processes, having to do the same stupid thing every month that takes 17 hours, and nobody's doing anything about that. From finance people, it's a curse, right? We're always like “Monthly close, I hate this, it will take 24 hours of my time.” And we keep doing it. And very rarely you raise your hand and say “Hey, why am I doing this? Can I just call someone and see if we can automate it?” That also kills the progress. You don't look for a solution to the problem just because we’ve always done this. Who said you did it right the first time? Why do you keep doing it the same way? Or it was right back then, but now three years later, is it still the right thing? It's not a constitution, you don't have to stick to it.

Tanya Kohen: What do you think is missing in today's tech stack for finance and ops?

Maria Frolova: There might be various things missing. One tool that I would personally like…It's ironic that the CFO is watching out for everybody's efficiency. They don't know what their own people do, they are busy. I'm sure you couldn't unlock more efficiency if you had a better view in their day-to-day lives. And finance people and operation people are troopers. They don't have a mentality of revenue earning guys, they're sitting there grinding. So I wish there was a better tool that would give visibility into what everybody is busy with and which tasks are consuming the most time. It's difficult with ad hoc’s, but with standardized tasks, you definitely could find efficiency if somebody just saw what they are. I think that one is missing, just the efficiency of the finance team too. That would be awesome. Another thing that's missing, it's philosophical, it's stage appropriateness. We always are behind, we always are working on something that should have been replaced two years ago. We're still stitching our quick books manually, if you have 50 legal entities, we always put ourselves last. Which is ironic because we control the budget, but we're kind of always shy from keeping up the stage-appropriate tech stack.

Tanya Kohen: If you could improve one thing in a business, not necessarily tech related, what would it be?

Maria Frolova: Communication. In big or small companies, things break down, right? You don't bring in the right people at the right time. When I was leading data, that was one of the pitfalls. People would build something and they'd be like “Hey, we need a dashboard.” There is no data infrastructure that you built as you were building that thing. Well, if you called me a month ago, I could deliver it today. But you called me now, so it will have to be six months now. This is completely unnecessary. And unfortunately, it's very rare that people actually communicate properly and in a timely manner. That would be my one wish, especially if you're 20 people. There's no explanation for that.

Tanya Kohen: No excuse for not talking to each other, agreed. A huge thank you to you for this conversation. Truly insightful perspective for our listeners.