Subscribe
Share
Share
Embed
This is your destination for feeling empowered in building your business.
These are the real, raw stories of entrepreneurs and business owners who have built their businesses through the messy middle of $1-20 Million, hosted by serial entrepreneur Matt Tait.
Matt knows what it’s like to scale past the first million, and on this show he’ll be bringing on other serial entrepreneurs and business owners who have been there, done that (or, are currently in it) to share what’s worked, what hasn’t, and what’s next.
Benjamin Stein [00:00:00]:
Accounting is the field that is most built around the acknowledgement and processing and catching and resolving of errors. So it should be the one that should be most capable of embracing AI, because it's like you have this very natural, like person A does work, person B reviews work, person C reviews erson B's review, person A being replaced with an AI or supplanted with an AI so they can do more work. It slots so naturally into that escalation sphere that I am surprised that we haven't seen more adoption. But I think maybe also it's just like it's early days.
Matt Tait [00:00:31]:
Hi, I'm Matt Tait, Founder of Decimal and host of After the First Million podcast. And like you, I've taken the leap not just to start a business, but to scale and grow it. I've hustled from zero to that first million, and now I'm building for the next 50. I know firsthand what the messy middle looks like, and we need more than just numbers. We need strategy, we need community and we need real conversations about what growth takes. In our line of work, scaling beyond a million means going beyond the spreadsheet. This is where you learn how to grow.
Matt Tait [00:01:07]:
This is After the First Million. Welcome to this week's episode of After the First Million. I'm Matt Tait, your host, and I am really excited to have a conversation with Ben Stein, CEO and Co-Founder of Keeper. Now, for those of you that don't know Keeper. Decimal uses, our company uses Keeper for its workflow management system. And so this season, as we are talking exclusively about how to build your firm and build your business past that first million, who better to talk to than the guy that builds the software that helps you get there? So, Ben, with that, welcome to After the First Million.
Benjamin Stein [00:01:47]:
Thank you, Matt. It's a pleasure to be here.
Matt Tait [00:01:48]:
I have a lot of things that I want to talk about and you and I are totally going to end up nerding out on a couple of things. We've discussed this stuff before, but before we get started, tell me a little bit more about your background and then how we got to Keeper. What was the impetus behind it? But first, you.
Benjamin Stein [00:02:06]:
Yeah, so I used to be the CFO of a tech startup called Mira. The CFO air quotes, I think are wholly appropriate. We were like five people when I started and 20 something people when I left, so it was like much more of an internal finance hire. So I like did our bookkeeping and I ran our month end close and it was so abundantly clear to me that the tools that I had at my disposal predominantly like QuickBooks and then another tab of QuickBooks and another tab of QuickBooks was like not quite what I felt like a modern close should look like, or really just like a modern day in the life of a finance professional should look like. So that's really the impetus for Keeper. It was my co founder Kenny and I building the tools that I wish that I had. And then as we started working with other bookkeeping and accounting firms, we got this sort of really valuable perspective on how some of the problems that were like sort of shitty for me were deeply like acutely painful processes. If you have to do them dozens to hundreds of times in parallels, I think that definitely helped shape our perspective.
Benjamin Stein [00:03:02]:
And the kind of nexus of those two is really what Keeper is today, which is a month end close management solution for growing and scaling cast bookkeeping.
Matt Tait [00:03:12]:
You know, one of the things that you hit on and you hit it with your one tab of QuickBooks and another tab of QuickBooks and another tab Of QuickBooks and that multiply that by a hundred clients and all of a sudden you have 400 tabs of QuickBooks and God knows how you figure that out. One of the things that you've highlighted on is a key problem in work, but I think particularly acute in accounting is switching costs. And it sounds like removing or minimizing switching costs is kind of one of the key theses behind Keeper. Talk us through that a little bit.
Benjamin Stein [00:03:46]:
I think that's a great point. It's something that maybe we thought about like we thought about somewhat explicitly, but also I think just implicitly as you're trying to build a better, more cohesive product experience, switching costs manifests as like you don't need to lose your place and jump to a different context. It's this interesting tension between how do we give a firm a cohesive experience where they can do as much in the product as possible while still 1, being best of breed and 2 as a consequence of that not trying to overextend ourselves. So I think that's like always been the big tension. So over time we've been able to add new products and modules to Keeper. So like good examples of this are a receipt management product for pulling in and scanning documents. You can post them to ledger our new emails product, our tax suite that we just rolled out. We try to be very conscious in limiting how many extra things that we do so that every new module that we roll out reduces context switching.
Benjamin Stein [00:04:44]:
But if we just ship something that is, it's like you save on Context switching, but then you lose on not having best of breed software. So that's like the real tension. The all in one vision is something that every firm really wants, but you shouldn't have to compromise on using software just because it's more connected. So I think that's the fundamental tension as a product developer is how do you do that build order the right way. But yeah, certainly I think context switching where we can shave those corners off helps one, save a lot of time and two, just make the closing experience psychologically feel a lot less brain damage inducing. Like Inn, you work from the top of the page to the bottom page and when you're done, you're done with your closing.
Matt Tait [00:05:27]:
Well, I really like that. And you know, it's, it's funny as you kind of mention, there are two things that you kind of mentioned that stand out to me. One is I've always really liked the idea of building a business by doing more or by doing less really, really well than doing more really poorly. And Jacob and I, when we started Decimal, used to joke that what ERP actually stands for is a bunch of shitty software that does a lot.
Benjamin Stein [00:05:54]:
Actually, I was talking with another accounting firm and they just went through like an SAP integration and they were like, yeah, it was terrible. I was like, I literally don't know of one SAP integration that has ever gone well. I think I'd agree with that description.
Matt Tait [00:06:07]:
When you look at really good software, even really good services, so you can apply this to building an accounting firm is if you do something really, really well, people will buy it. If you then add something that you also do really well, the same people will buy it. Because you built trust that you did the first thing well. It doesn't mean you can't test and fail a little bit, but it also means just slowly expanding versus doing everything all at once poorly is a much better way to build a sustainable and profitable business. As well as like a bunch of people that actually like you and talk about how much they like you.
Benjamin Stein [00:06:48]:
Obviously there's a lot of natural applicability to software. Like you just roll out new products, you try and make them well integrated together. I'm curious how y'all have taken this approach in building out Decimal.
Matt Tait [00:07:00]:
So a lot of the same way we have added other services as we knew we could be really good. So we did it two different ways. One, when we acquired Spark from kpmg, we then acquired a KPMG-trained tax team. They were really, really good. So that meant, hey, guess what, we can do tax and we already had hundreds of tax clients, so adding a couple hundred more was pretty easy. We acquired our way into a skill set, and that's one way that we'll continue to do things. The other way has been we get asked a lot to do a lot of things, and we're much better at saying no than yes. But we do track how many times we say no.
Matt Tait [00:07:40]:
And when we get enough, it's like, okay, like, I actually think we could build a sustainable business here. Let's test it out. Ramp was actually that maybe three years ago where we got a lot of people asking, hey, can you guys work with Ramp? Can you use Ramp for expense management? And God knows, like, legacy software and expense management is pretty poor. And so as our kind of credit card integrations into QuickBooks. And so, like, okay, like, let's give it a shot. It was relatively new. It might have been five years ago even, but we said, let's test it, let's try it out. Let's see what it does for us.
Matt Tait [00:08:12]:
Then Ramp added bill pay, and that gave us the ability to start adding more and more. But we did it by testing, having a bunch of people that liked it, testing more and then adding it. And so we like that kind of land and expand. But we also test. And there have been dozens of things that we've tried that we also do not do because we found, number one, we weren't great at it, so it wasn't really good for us. Number two, we couldn't do it profitably, which means we weren't going to do great at it at scale. Or number three, we really just didn't end up with a lot of people. One of the greatest examples of, like, we thought we might need to do a lot of it, but we actually don't is payroll.
Matt Tait [00:08:49]:
We get asked more if we have payroll than if we do payroll. People care that we check the box, not that we actually do it.
Benjamin Stein [00:08:58]:
Interesting. So do they then, like, run the payroll internally or?
Matt Tait [00:09:03]:
They do.
Benjamin Stein [00:09:03]:
Okay.
Matt Tait [00:09:03]:
But then it's also like, if you're using a modern payroll provider, like a gusto or a rippling, you really don't have much to do. If you have an hourly worker, you're confirming the hours, which you're not going to have an outsourced provider do anyway, because they have no idea. And if they're not, if everybody's salary, you literally just auto pay because everybody's on salary, so you don't really have an actual need. And so when you unpack it, people want to know you can do it and you have thoughts on it, but not that you actually need to do it.
Benjamin Stein [00:09:33]:
Like the fact that they're. They're giving you the job to automate their back office, and it's like, yeah, okay, if I'm going to click one button that is like, going to automatically run payroll, I don't need you to click this button for me. But the fact that you know how to work with gusto and you know how to work with rippling means that you're going to know how to manage the accounting side of that and how to, like, handle the integrations if the integrations break. I would say it's like a potential buyer of accounting services. Even though I can click that one click button, knowing that you are comfortable and familiar with the systems is super appealing.
Matt Tait [00:10:01]:
Well, it actually gets me to a question I was going to ask you because, like, you have a really unique perspective because you help hundreds of firms do this stuff. And so you get to see the differences between really great firms, mediocre firms, and really bad firms. And as long as they pay you, you're fine with all of them. But when you look at really good firms, how do you kind of take the. And you've highlighted a little bit using tech to scale. What do you see are some of the differences and important parts of building towards that great firm and using technology to scale.
Benjamin Stein [00:10:37]:
Technology is like, as a software vendor, you kind of want to push tech first all the time. But I think actually the first thing you have to do that we see firms do really well is like, first figure out what you want to be great at and figure out what are the big problems in your business. And so this is like something I've always respected about Decimal. Like, I say this all the time. If I wasn't running Keeper and I were going to run an accounting firm, I would run it. I would try to just like, copy and paste run it just like Decimal. And I think the thing that is most unique about the Decimal model that I've really appreciated is, is the sort of bifurcation of work and then the prioritization of response time. As a recipient of services of any kind, it's like, so crazy how important response time is.
Benjamin Stein [00:11:20]:
And it doesn't mean that, like, I need, like, if I ask for my financials, I need them like, I need the financials turned around in two minutes. But it does mean, like, if I ask for something and then two minutes later, someone responds like, hey, I'm on it. Like, we'll get this back to you. In 24 hours or 48 hours or even like two weeks, just like, knowing that someone is that responsive totally changes the way I perceive the value of the service. And so that, I think, is the first thing. It's like, okay, we're going to prioritize response time. It's like, okay, great. What tools will help us do that? Starting from there, from that, like, here's the things we want to do and want to be great at.
Benjamin Stein [00:11:55]:
And then figuring out like, okay, here's the tech and here's how the tech might fit is like, honestly more important.
Matt Tait [00:12:01]:
Think about the outcome. I got in trouble this week, you know, me and my co founder and team. And so you know that I get in trouble a lot. This week it was because I posted on LinkedIn that Whenever I get a client call, and at this point, about a thousand clients, I let it go to voicemail every time. What I said, when you unpack, it was, I will let it go to voicemail. I will immediately listen to the voicemail and text back, hey, sorry, I'm in a meeting, which, realistically I usually am, but I'm going to circle with my team, check in on what's going on, and I'll get back to you asap. Because what I do better at, I'm an introvert. So I actually do better when I have a chance to process.
Matt Tait [00:12:41]:
So I listen. It's almost always an escalation. Like, if somebody's calling me out of the blue, it's almost a 95% escalation chance, and I want to hear it. I then want to see what's really going on. I ask my team, hey, what's going on? What are we doing? What's the plan? How can I respond that way when I actually do talk to the client? I have, here's what we saw on our side, here's what you saw on your side. Here's what we're doing about it. Here's the plan. I'm able to give them that full spectrum, but to your point, I responded immediately with a text.
Matt Tait [00:13:14]:
I got what you're looking for. I'm researching it. It's very important. I'm going to come back with a full plan for what's next. It blew a lot of people's minds because they're like, wait, you guys are all about responsiveness and client service. And I think we do do a great job. But you don't ever pick up the phone when a client calls. And I'm like, no, because you know what that call does? They're venting, they're angry.
Matt Tait [00:13:37]:
If I'm in a bad mood, maybe I get snippy and I don't mean to be, but who knows? Or I have to listen to them, I have to unpack it. Then I have to do a call. Or I can let them go to voicemail. Quite frankly, they can yell as much as they want at my voicemail and all I have to do is turn the volume down. I then see what's really going on. And when I do have a conversation with them, I'm prepared, I have a plan. But in the meantime, I responded quickly. And so like, that's how I got in trouble.
Matt Tait [00:14:03]:
But I thought it was actually worthwhile. And it's also exactly what I do. How do you deal differently with those scenarios and situations?
Benjamin Stein [00:14:10]:
I think that's very thoughtful. I am not quite so thoughtful. So I just pick up the phone. So what I do that I think more than anything else helps me be responsive to our customers is I have a filter in my inbox that lets me see all emails that come in from my customers. And that way I can just like respond to those first. So like random emails I get from vendors or app notifications or all the million things that are in my inbox that doesn't clutter up me being responsive to my customers. And so I actually have like a literal sticky note on my desk. It's a to do loop.
Benjamin Stein [00:14:38]:
It's like, first, am I blocking anyone internally on the team? Second, like, I check my top priorities for the company and make sure I'm executing on those. Third, am I being responsive to my customers? Fourth, like the random other things I have to do today and then 5th is like the lower priority stuff. But so I'm always kind of running through that, that loop, making like the response to customers like one of the actual action items on my to do list. It's been super helpful in terms of making sure that everyone who emails me gets a response for you and maybe escalations for me. A lot of times it's like bugs a lot of manual or automated emails. Like, I send a lot of emails. Like every Thursday when we ship new product updates. Like, those emails come for me and so people will like respond with a bug.
Benjamin Stein [00:15:17]:
And sometimes if I like, know what the cause of the bug is, I'll like write back with the solution. But usually what I'll do is just loop in our support team and then our support team responds like one minute after that and then they get their issue handled. And so that's actually faster from time of initial bug report to resolution. Like, if I didn't respond to this until I had time to respond to it, and then I had to, like, do the debugging, and this is going to take 30 minutes. I got to slot 30 minutes in somewhere. So I think actually being responsive and then delegating ends up being better quality service than if I had, like, tried to make sure that their debug was, like, done by hand by the CEO for some reason.
Matt Tait [00:15:52]:
Well, and it's also. When you're trying to grow a company, your goal should be working yourself out of a job. In order to do that, you have to empower people, which means you have to give them work and delegate. And that's just the right thing to do. You can't do everything. And I think that's a tough thing that a lot of people that start a company, and your first one in particular, you. You really think for too long I have to do everything, I have to know everything that happens, I have to see everything, and it breaks you at some point.
Benjamin Stein [00:16:22]:
I'm curious where on your journey, like, have been those. Those big moments where you reached that tipping point and needed to delegate big parts of your work. And I'm curious how you've thought about building the team structure to make that. Make that something that's functional for y'all.
Matt Tait [00:16:35]:
So I got lucky. That Decimal is not my first company. This is the. The third company I've run. The first one did well. The second one failed entirely because I did all of these things wrong. So when we started Decimal, we literally started a company that the two co founders were completely incapable of doing the work. Client one, the first thing we had to do was convert them from desktop to QuickBooks Online.
Matt Tait [00:17:00]:
And Jacob was very good at that part. But once we got them live on qbo, it was like, holy crap, we need to hire an accountant to actually do the accounting work. So we built it in from the get go. The way that we looked at it was we always broke work into buckets. And when it was the two of us, it was two buckets. And then we hired a third person and it was three buckets and then four. And every time we lessened work all around. And so we really started to look at what buckets could we give away.
Matt Tait [00:17:33]:
Something I do on a quarterly basis is I'll do an audit of my time for the past quarter and I'll look at all the meetings I'm in, all the work I did, all the output I accomplished, and I look and I think what shouldn't I have done? What do I not like doing and what do I not need to do? And then I look on the other side. What did I love doing, what was I really good at doing, and how can I do more of it? And that time audit helps me in the next quarter figure out what I need to delegate and what I need to keep doing more of. I have team leads that now do this. Like, we encourage that all the way down the organization of doing audits so that you can push your work down. And that's really helped through the life cycle of Decimal to do that more and more. And we're always late. I mean, you're doing it in arrears. So I'm always a quarter behind in terms of getting it right.
Matt Tait [00:18:26]:
But how do you guys do it? How have you kind of gotten through that with Keeper?
Benjamin Stein [00:18:30]:
Yeah. I think this is one of the places where I'm like, actually pretty grateful. There's like, a lot of best practices. Like, I think all of our innovation happens in, like, what products are we shipping for our customers? And none of our innovation happens in the like, org structure or anything like that. Like, we have an extremely conventional org structure. Like, if you've ever seen what a SaaS org chart looks like, it's going to look the exact same from business to business. And like, even the ratios of like, total CSMs per company size or like engineering staff for company size, like, all very similar. We have a product or EPD engineering product design.
Benjamin Stein [00:19:03]:
And like, Kenny is the cto, so he kind of takes over a lot of those responsibilities. Product managers who own different service areas of the product. We've got a sales manager who manages our account executives and our SDR manager who manages our SDRs. And so it's like, all very conventional. We really don't have to reinvent the wheel. So I'm like, pretty grateful. There's a lot of good playbooks out there. I feel like that's actually something that's, that's like maybe missing in the accounting space. I'm actually curious, what do y'all do? Do you do pods or functions in terms of your different service lines?
Matt Tait [00:19:36]:
We do all of it. I'll actually design it out because it's. It's one of the things that we're focusing more and more on is making this transparent. I actually think there's another underlying issue that you didn't hit on, which is there isn't really that great of continuity between definitions when people mention bookkeeping. There's a different definition between A lot of firms and companies and there's a different definition between the buyer and the seller. Controller services, CFO services. There's too much ambiguity, which also means that there's problems in org structuring behind that ambiguity. What we've done at Decimal is we have teams.
Matt Tait [00:20:18]:
So each team is us staff. They handle client relationships and quality management. We don't want them to do the work. So they are in teams. Which also means that, hey, if you're a client, Ben shoots an email over the whole team, gets it, maybe one specific accounting manager answers it most of the time, but anybody can hop in and help. And so it just means that everybody's allowed to work together, which also breaks people apart from the islands that they get to in remote work. But also just accounting in general feels like an isolating industry sometimes. And so we've created that structure.
Matt Tait [00:20:53]:
But then on the Philippines team we also have what we call delivery support. They do like projects, onboarding, stuff like that. But behind them we have queues and all of our transaction classification, all of our bill pay soon payroll and invoicing tax is already done in a queue, which allows for work to just be accomplished in an efficient manner and then pushed for quality to the next layer. So we've really broken apart relationship from work. And then technology is an underpinning foundational infrastructure as well as what buttresses all the above groups together. So there's a way more complicated answer than you were expecting. But does that make sense?
Benjamin Stein [00:21:35]:
No, I think it makes sense. I'm curious, like was there a point in scale where you're like, this makes sense now, like is there a minimum viable scale that you need to have that type of an org chart?
Matt Tait [00:21:46]:
I think as you scale you need to look at things differently. I think you should always try to break up work from relationships because the cost of a US staff person doing work is more expensive than your client is really going to pay. And like right now at Decimal, our newer clients are averaging $3,500 a month. They're paying us for this work. So this isn't cheap. But at the same time what we find is breaking that cost structure. So if I'm Ben and Ben says, hey, I'm done with Keeper, I'm ready to just run the exciting world of an accounting firm. So you start your firm today.
Matt Tait [00:22:19]:
Your first hire should be a delivery support person in the Philippines and you should be the relationship. You should manage the quality of their work, but you should push and train them to do all the work. And then when you get to maybe 100 clients and you've hired your first two accounting managers and you have three or four delivery support people, so you've got some redundancy. Then you can start looking at your next point of scale. And your next point of scale could be queues, which is pulling more work down to a cheaper, easier way of doing it. Which also queues is where you start to see AI and automation, some rpa, being able to take over some of that work. Because the goal is you actually want technology to be able to eat up as much as you can, but you need the work to be bucketed into interchangeable parts in order for that to happen. Because at least not for a long time, you're not going to have AI or an AI platform able to take over this whole bucket.
Matt Tait [00:23:17]:
They're going to be able to take over pieces. Some people claim 100% accuracy, 90% accuracy. I don't believe it just yet, but I do believe that you can start interchanging the pieces and that helps the cost structure.
Benjamin Stein [00:23:29]:
As long as you're shy of 100%, you need an accountant. It's something, I think we see a lot of people who are like, who haven't spent a lot of time doing accounting or in the accounting space that think that this is just going to be eaten completely by AI. It's worth having the intellectual humility, from my perspective, that they may be right. But also, like, I see the work, I see what the tech can do. We try to make the tech do the work. And I think there's a lot of potential. Like, I do our books, my own personal clothes take somewhere on the order of four, four and a half hours. I think by the end of the year it will be closer to one, one and a half.
Benjamin Stein [00:24:02]:
And a lot of the things that I am currently doing today, it's like manual button clicking will get pretty much eaten by, like, a combination of AI and deterministic software that's automating what was previously like spreadsheet copy and paste. But that still leaves you with like an hour to an hour and a half of, like, oversight, supervision, correcting, tuning the systems. And it's like, okay, do I want that being done by, like, me, the business owner, or do I want that being done by, like, a professional accountant? You still have that functional desire where it's like, I want this done. Like, I want an accountant to be accountable to delivering this thing right? And if they get more efficient, great. Like, that's awesome. But if I'm a business owner, I don't want to start suddenly spending an hour, two hours more of my day managing an accounting system.
Matt Tait [00:24:47]:
You can spend an hour like eventually you. Ben, who has an accounting background, who deeply understands the software that he's using, can get his close to an hour. Me, who understands fundamentally what accounting is, but can't count and do any of the work. It would take me six to eight hours to do it. A thousand times worse than you ever could. Or I can pay somebody to do it. The other thing that you want out of this, and, and here is where I think the interchange of technology and AI is, is really putting accounting at a massive improvement point is what I really want as a business owner. I want that accounting person I'm paying to also give me advice.
Matt Tait [00:25:29]:
Hey, what did you notice? What did you see? What can I do? How can I improve? That's what I'm actually looking for. And that's where when you take that five hours of work, as right now you're at six hours, you're going to one that's five hours of change. Take that and apply it to advising and all of a sudden everybody is happier and you're making more money. I think that's where we have to get to.
Benjamin Stein [00:25:56]:
I agree. I mean, I think it's funny the way AI kind of, it's the same as every other type of automation in that respect. It's like, okay, how do we, like save time from the drudgery so that we can spend more time doing the high value, high ROI work?
Matt Tait [00:26:10]:
Yeah, I mean, computer software brought it off of paper, the cloud brought it off of a server. AI is bringing the tasks up.
Benjamin Stein [00:26:18]:
When you're talking to AI developers, they talk about human in the loop. The AI is doing work and then the human is like checking the AI's work. This is literally just another way of describing software. You could describe tax filing software as human in a loop, right? Your Android members in. And then the software is doing the human in the calculation and then you as the human in loop are doing the checking. But as long as you have humans checking the AI's work, it looks a whole heck of a lot like software, like powerful software, but software nonetheless.
Matt Tait [00:26:45]:
Well, and I think the other thing too is, look, human beings are fallible. They make mistakes. AI isn't 100% right 100% of the time. Won't be for a while and maybe ever. Because as long as the data that gets there is inputted by human beings, there are going to be mistakes and they're going to be weird mistakes which are going to require things that the AI has never seen, which is going to require a human being to pop in and put some discernment on there for at least a measurable amount of time. So if you need that, you understand that if people are fallible and AI is fallible, you can actually see improvement by implementing AI. Just understand that it's not going to be perfect. I see so many people saying, well, hey, I'm going to wait till it gets to perfection.
Matt Tait [00:27:31]:
Okay, why? If it's cheaper, if it's equally as good, and if you just have to have somebody check it, why aren't you putting it in today? It's something that I'm seeing more and more that I just. I don't understand in accountants is why not take the opportunity versus then fix the risk versus fix all the risk, then take the opportunity.
Benjamin Stein [00:27:54]:
I think it's ironic because accounting is the field that is most built around the acknowledgment and processing and catching and resolving of errors. So it should be the one that should be most capable of embracing AI because it's like you have this very natural, like person A does work, person B reviews work, person C reviews person B's review depending on level of risk associated with that process. So because you have this natural tiering, person A being replaced with an AI or supplanted with an AI so they can do more work, it slots so naturally into that escalation tier that I am surprised that we haven't seen more adoption. But I think maybe also it's just like it's early days. I'm curious, what are the places where I guess y'all have already found great success implementing AI in your org and then also where you're excited about AI going over the next couple of years.
Matt Tait [00:28:43]:
I see AI is doing two things and. And we've already started on both. I see the biggest impacts being eating from the bottom up and the top down. Bottom up is there is a lot of manual repetitive work that just goes into basic accounting operations. C, bill, input bill, process, bill, pay bill. There's a lot that you can. And we've started implementing AI and even building towards it a little bit internally. But then there's also the bookkeeping and our partnership with Puzzle was a big jump forward in this.
Matt Tait [00:29:15]:
And we're constantly evaluating new and new ways to maybe layer on top of legacy software like QuickBooks. Because if we can take what our queues are doing and apply some AI to those buckets, that starts to remove costs from us. And it doesn't mean that we're replacing people. It means that our people can do 10x more. And so that's how we look at it from the building, from the bottom up, the top down, which I think is also more fascinating is everybody talks about advisory work and I think advisory is really important. It's what we need to get to, it's what we need to price to, it's what businesses want. The problem is is we have a lot of people that aren't trained on how to do it and you need to lead them. Well, AI is actually a really good leader in that because you can start to apply it to some financial closes to variance analysis in reporting, to some custom reporting.
Matt Tait [00:30:08]:
Hey, what do you notice? What do you see? And using that to then script out a client meeting that is, hey, here's what we're seeing, here's what you need to notice, here's what we would do about it, here's what other clients. So you can kind of spoon feed and lead people to advisory by applying good technical AI on top of the kind of reporting process in the month end meeting. I also think that you can analyze scripts from meetings to really help tell sentiment analysis. So pre escalate, churn, get ahead of poor calls or bad client experiences or maybe a client's not happy. Those are some areas that we're doing it right now. When you look at your product and how there are a million ways I could see you implementing AI into your product to make work easier to pull out analyses, advisory, sentiment, what are you guys doing and how do you look at implementing AI into Keeper?
Benjamin Stein [00:31:02]:
So the lowest hanging fruit sort of two things that I expect will ship in the next couple of mid Q2 to late Q2. First one is AI bank feeds. This I think is like fairly obvious in people's minds but like how we take the transaction feeds and make sure that they're classified correctly. I think there's sort of the model we've seen some software providers take that I think is like the wrong model is okay, let's automate 100% of things and do them 90 to 95% correctly. And I think the big problem with this is you end up with a system where like catching and correcting and figuring out why a system made the error ends up taking more time than you saved implementing this AI solution in the first place. So where we want to start with our AI bank feeds is let's do 60 to 70% of the transactions 100% correctly and then build from there. So go from 60 to 70 to 80 to 85 to 90% and make sure that when we do process something it's 100% correct. So I think that's really the core focus.
Benjamin Stein [00:32:01]:
The way we think about rolling out AI writ large is again like this kind of human in the loop process, but really like powerful software that's just doing more of the work for you. It's taking away the clicks you had to previously make. AI bank sheets is one, accruals automation is another. And and then as we ship AI bank feeds we kind of get bank statements for free so we can get auto reconciliations. And so I think those kind of are the three big ones in close. So bank feeds, accruals and auto reconciliations. Once we have statements that will take a huge breadth of the closing work off of the plate of accountants. The one other tool we have it live today and we are in the process of updating our architecture, which previously used 4.0 to using O3 mini is AI journal entries.
Benjamin Stein [00:32:45]:
So this is basically taking unstructured data and turning it into structured journal entries. In my use case this is like I get an export of a CSV out of stripe every month and I need to copy and paste data for 30 minutes from one screen to another. So I expect AI journal entries to be great at. I can just give it the spreadsheet and tell it what I want it to do and it will be able to post the journal entry. Those kind of four big things are what I expect will take something like 50 to 50, 70% of the like coding closing work off of my plate for my own personal close. And I think there's some interesting things we can do around practice management. So sort of like proactively, preemptively answering emails or like drawing drafts that you can then like click send on. But I think really we're excited about the low hanging fruit in close today.
Matt Tait [00:33:29]:
We use it in emails all our go to market marketing. I mean it's, it's the first draft of almost everything. What I think is interesting that you said and when we talk about work we talk about breaking it into smaller and smaller buckets. If we can get to be a hundred percent correct 60% of the time and to know the difference between that and when it's not, that's Perfect, that removes 60% of the work and then all we have to do is look at the 40. That to me is a much better thing than when people say we're going to automate all of it and you have to go check 100% because you're not sure where everything's wrong. Also, you know what? Human beings are really terrible at knowing when they're wrong and telling you. So that to me is one massive improvement with AI is if the AI tools that we use can tell us when they're uncertain, that's perfect. Do the things they're certain and then tell us where they're uncertain.
Matt Tait [00:34:20]:
It starts chipping away at work. Nowadays we're still at this early stage and this is where I, I caution people on implementing too much is the costs haven't gone down yet. And so if it's not doing enough, it's actually just increasing costs and decreasing margin. And so you have to kind of walk that balance. It's like I always joke with my kids, like when a new TV comes out, we wait for year two. I'd rather buy the $200 great TV a year later than the fifteen hundred dollar TV the day it comes out. And so I think that's a little bit of the stage I also think we're sitting at.
Benjamin Stein [00:34:53]:
It's going to take time to make these systems good. And there is a tension between wanting to be the first one and wanting to set up these systems. Some piece of it is it will actually save you time. Some piece of it is it will reduce the acute burdens of hiring problems because you just won't need to hire as many people. I think another piece is just aesthetically the right way you want your business to run is AI doing the low value transaction at work. I think that's actually where we got a lot of our early customers is like now we can tell you like, we'll save like 20 to 30% of your month end close. And we'll help you get responses back from clients like more than twice as fast if you get them at all in your old model. But we didn't really have any of those data points to go off of. So it was really like our early customers were like kind of just looking at this is esthetically the way that we think an accounting firm should be run. It felt better, right? It's hard to really tell that story. You just kind of have to show it well.
Matt Tait [00:35:47]:
And I think what you've hit on too is people make emotional decisions almost always before they make a rational decision. Like you can't untether those two things. And I think it's so cool how much you guys have learned from your customers. I'm interested as you've worked with a lot of accounting firms and you've Worked with firms of all sizes. You asked me about life cycles and how I would build the life cycle of a firm. Talk to me about what you've seen in terms of what are the life cycles of a firm that you've worked with.
Benjamin Stein [00:36:16]:
So the sort of very, very early business stage is like, find any customer who will work with me. People also try to pick a niche so they like, look at their current client base. They say, okay, who are my best clients? Who are the clients I like working with most? Who are the clients that like working with me most? And then they start to sort of shift their. At the very least, their. Their focus. But a lot of times their marketing focus on, let's just go after this. This one niche in particular. And then from there they can build to a pretty decent size.
Benjamin Stein [00:36:46]:
At some point you have to start hiring a manager, like to manage the cast team. And then at some point you have to have team managers or pods and then you have a sort of director who's running the entire cast function. Then at some point you have to make the decision, am I going to go with a pod structure? Am I going to go with a more horizontal, functional structure? So I think that's kind of the high level or of what we see. But I think the sort of interesting question is, do you generalize or do you specialize? Ultimately, what we see is like super large firms, they'll have verticals, but they won't have a single niche. And I think that's kind of the best way. Best of both worlds. But it's an interesting tension because when you're smaller, having niche is super important. It helps your marketing stand out.
Benjamin Stein [00:37:26]:
It helps you become a lot more efficient. I'm actually curious from Yalls perspective, like how you've wrestled with that from a. Either like setting up verticals, Will you take any client or their clients that you turn away because they don't fit a vertical? Or are you large enough now where you kind of done every industry? I'm curious how you think about that.
Matt Tait [00:37:43]:
Yeah, so we're big enough now that we work with pretty much anybody of a certain size and scope. But we do have industry verticals that we focus on, and those are the ones that we seek out. We talk about the difference. I'm a baseball fan, so the difference between what we pitch to and what we catch. So we pitch to the industry verticals that we like to focus on and that we're best with and that like us the most. But we'll catch anybody that's willing to Pay our fees, do it the way we want to do and you know, fits generally within our tech stack. That's another big thing that we have is we want them to fit into our tech stack. So we use QuickBooks or Puzzle for the ledger Xero as well bill pay softwares, we have a few expense softwares.
Matt Tait [00:38:24]:
We want everybody to be able to work within those because that's where we see a lot of decreased margin. We talked about switching costs earlier. We want our team to have as much of a minimal switching costs on a daily basis as possible. And so there are a lot of software products that we just won't help. Now somebody may come in and they use paylocity for payroll. It's not one of our preferred payroll providers. We'll do the payroll entry and bookkeeping. We're just not going to touch it.
Matt Tait [00:38:50]:
So there are ways that we can do it and we'll cost our way around it. We started out before in January 2020, outbound was still working well and so we were able to grow using the kind of tech playbook to growth. And that's where we spent our careers. It's where we grew up in. And it worked really well. Then the pandemic and post pandemic and AI have made it so that pure outbound has become a lot less effective I think because probably like you, the sheer volume of things that hit my inbox, my phone, my LinkedIn profile on a daily basis means that quite frankly, I don't see any of it. You could send me the world's greatest outbound email. I'm not reading it because I got 300 other ones today, but we should have niched down sooner.
Matt Tait [00:39:37]:
And that to me is one of the things that I think Decimal could have done quicker, that would have even accelerated the growth beyond what we've done. I think what people should do is like you said, create a niche, start, do whatever you can and then stack niches. And I think stacking niches is the way that I would do it. Rather than us, we just grew really big. Then we looked and said holy crap, we actually have some verticals that we do well, but we still have all of it. So we came at it backwards and I would do it very differently today.
Benjamin Stein [00:40:08]:
For your verticals now do you have like a verticals lead? To what extent is that, is that compatible with versus like a little bit oil and water on the rest of the model that you've describing. You've described in terms of the like shared services team versus the individual teams that are Client facing.
Matt Tait [00:40:26]:
For me, the niching and verticalizing is an internal and an external thing. As you grow, your team gets better and more efficient as you verticalize, your sales gets better and more efficient as you verticalize. So to me it's an inside and an outside thing and you need to do it together. How have you guys done it?
Benjamin Stein [00:40:48]:
For us, the sort of primary thing that changes how we work with that customer is just size. I think this is one that we've had to go through a bunch of iterations on. But our customer success team is a sort of mix of dedicated customer success managers. Like an individual customer will have a dedicated relationship with one person at the company. Then there is, we have our implementation specialist team who will work with smaller customers during the beginning of their experience with Keeper to make sure that they get the product set up correctly. They're using it the way they want to use it. They're taking advantage of, of the features that we have available to them. And then there is the, the sort of reactive side for self service customers.
Benjamin Stein [00:41:26]:
We've tried to make good resources and training available to anyone who wants to access it and then also like anyone who can write into our customer support. That's kind of the three different tiers for us. It definitely has, has been a lot of iteration and even within that just like making sure that we set up our, our like first 90 day process really well so that businesses who decide they want to try Keeper actually have a good experience in. Keeper has been where we spend a lot of our time.
Matt Tait [00:41:50]:
I love how you've put together multiple ways for people to learn and to understand how to use your software. That's something that we talk a lot about is some people like me do much better at listening and hearing to learn, whereas other people like Jacob, my co founder does much better reading. But you have to understand that different people do things differently.
Benjamin Stein [00:42:14]:
I completely agree. I think that's it's one of the ones where you will never regret doing more. Like you'll never regret having written more support articles and more recorded more videos. I think this is one of the ones especially because a video is just so much higher fidelity for learning how to do something. The big hurdle of like you want everything that you put out to look polished and professional. It ends up meaning that you produce a lot less stuff which means your support is a lot less robust. So you'll see all sorts of just like random short videos that are in our help center that are just like here's how to do this thing that may not look like the most professional thing in the world, but it does the job and it means that anyone who wants to learn how to use the product can learn in the way that's best for them.
Matt Tait [00:42:55]:
I really like that. I think so many people focus too much on perfection. Whereas just like get stuff done and like there are certain things that you want to be very polished and perfect on, but there are others that you just need output. And that I think is so important. When we look at client relationships, we look at service offerings, support, internal, like people just want help and when they need help, if they can find it, that's the most important thing.
Benjamin Stein [00:43:25]:
I couldn't agree more. We orient our entire company around this like it's our single core operating principle. We call it move fast or die. We try and implement this in software by making sure that like we're shipping software weekly, we're reducing the scope of what we ship so that we're shipping new stuff that our customers can try and that our customers can lead in terms of how they want the product to continue to evolve. And then in success and support and customer facing roles, it's like just respond to the email, like get the thing out the door rather than trying to make it the most perfect email in the world that won't get any more attention.
Matt Tait [00:43:57]:
I really appreciate your time today, Ben. I've loved how you guys have built things and you know, there are two core parts of Keeper that were game changers as we made the decision.
Matt Tait [00:44:08]:
Three. Number one was team. Like finding a great partner and having great partnerships with tech companies I think is so important when you're a service provider. People that you can trust that understand you and are willing to listen, but also willing to push back because you've pushed back as much on us as we've pushed on you. Maybe not as loudly, but you've pushed back on us. But the other thing was you looked at one of the core problems in switching costs and you integrated directly with the ledger. And that's a problem with workflow management systems in all scope of work and all types of companies is when it creates an extra system of work and it creates switching costs, that becomes a problem. And the final thing is also taking that premise to the client and making it so that you have the questions on top of a transaction, making it easy for them to do the work that clients have to do. And I think that's so important.
Matt Tait [00:45:01]:
So I really appreciate you, I appreciate Keeper. I appreciate your time today. I also think it's been fascinating for the people that Listen, to hear a tech CEO say that tech wasn't the most important thing and it is a secondary decision. And I completely agree with that.
Benjamin Stein [00:45:19]:
Yeah. Awesome. Well, Matt, thank you so much for having me on. It's truly always a pleasure. Like I said, I respect the hell out of the business that y'all built at Decimal.
Matt Tait [00:45:29]:
All right, I am excited to go over all that I learned and the takeaways that I have from my interview with Ben Stein for After the First Million as you all know, growth doesn't stop at the first milestone or the first million. It's about what comes next. If you're building a company or a firm, build as if you've already passed that milestone, that first million and you're going to keep going. So here are my takeaways from my conversation with Ben. 1. Solve key problems. Don't just keep adding services. 2. Response time matters. 3. Use AI to boost efficiency, not perfection. I think that one's really key going through these number one, solve key problems. Don't just add services. When you start and as you grow, focus on addressing real pain points that add value. Then get really, really good at doing that. Only once you've gotten really, really good at doing that and you've created space to possibly do more, should you look at spreading out. Too many people get caught by the shiny object and they think, oh, I could do that or I could do that or this client's asking for that.
Matt Tait [00:46:34]:
It's much better to be really great at a few things than okay at a lot of things. And that focus, both for a software company and for a professional services or accounting firm, is really, really important. Get good at a few things first and then expand. Don't be okay at everything. You will generate much happier clients, much better relationships, much greater lifetime value, and I think you'll just genuinely be happier. Number two, response time matters. This is something that we have ingrained in Decimal from the very beginning. So to hear it from Keeper was absolutely amazing.
Matt Tait [00:47:13]:
Responsiveness is the most underutilized asset, I think, of companies and people today, let alone accounting firms and bookkeeping shops. It's easy to give it lip service. It's easy to do when you're one person. But thinking about how do you do it at scale, how do you truly look at breaking it down and how do you teach your team to do it is really important. I thought what was really cool to hear Ben say was he he talked about prioritizing quick responses. Even if it's just an acknowledgement, it's something that I do quite a bit when I get a voicemail from a client and it's an escalation, I'll immediately text them back, hey, got your voicemail. I'm going to check in with my team. We'll circle back when I have more information.
Matt Tait [00:47:55]:
It's a quick response. It shows that I heard them, that I cared and then that I'm going to solve the real problem. Ben does something similar. He's also put the systems in place with a specific inbox where all client emails go to. It allows him to prioritize his response time. Systematically putting those habits and systems in place makes it so that you can scale the theory or thought of responsiveness. Acknowledging client concerns is also really important, as is keeping them informed. But making sure that you're doing it in a way that allows you to do it as well with one client as you can with a hundred and then a thousand is I think one of the keys.
Matt Tait [00:48:34]:
And it's where everybody breaks in accounting and quite frankly all professional services. Finally, number three, use AI to boost efficiency, not perfection. I hear so many people when they talk about AI saying I'm waiting for 100% perfection. People aren't perfect. You can't replace perfect imperfect with perfect and hope it is going to be that way. It's just not going to happen. AI should reduce manual work. It's never going to replace or at least not that I can see human beings entirely focus on where can it make you more efficient and then fine tune it as it evolves.
Matt Tait [00:49:11]:
If it can be a 100% right in 60% of the ways, in 60% of the places, all of a sudden you've knocked out 60% of the human work. Look at it that way than rather than looking at how can I make it a hundred percent effective at a hundred percent of the things I'm really excited to see firms and companies that prioritize standardizing and streamlining workflows and minimizing switching costs and creating that seamless experience using systems like Keeper and there are others that are out there I think is really important to building a good foundation for work. The most successful firms and growth in this space will come from implementing the right technology as the right foundation to help empower people to be the advisors that these small businesses want. Now is the perfect time to be in accounting. You just have to be ready to embrace the tools, the growth and the constant change. Thank you.
Matt Tait [00:50:09]:
Thanks so much for listening. After the First Million is presented by Decimal. To listen to more episodes and find tips to help make running a business easier, visit decimal.com/afm. Want to join the conversation? Reach out to me on LinkedIn and let's explore the messy middle.