Summary
Arjun joins an episode of the BlueOcean podcast by StartOps to discusses how consumer brands can reduce manual data entry and streamline finance operations with inventory automation. He outlines Mandrel’s vision to automate document data capture, SKU-level landed costing, and real-time inventory insights, enabling brands to scale efficiently without increasing headcount.
Episode Links
StartOps: https://startops.network/
Mandrel on StartOps: https://startops.network/companies/mandrel
Spotify: http://bit.ly/3DOYehN
Apple: https://apple.co/40oyJN4
Transcript
00:00 Introductions
01:41 Arjun’s journey to founding Mandrel
05:06 Challenges in inventory management for consumer brands
06:43 Why inventory operations are still so complex in 2024
10:00 Automating data updates in real-time
12:13 Mandrel’s approach to integration and automation
13:34 Core features of Mandrel today
16:00 Item-level data tracking and insights
18:04 Building a source of truth for inventory
20:26 Automated landed cost calculations
22:00 Replacing manual processes with automated tools
24:00 The future of commerce and Mandrel’s vision
26:29 Closing thoughts
~~
Arjun (00:00)
We're not building features here. We're trying to solve a problem. And the problem in my mind is not like I need to do inventory planning, or I need to manage my years. The problem is, I have too much to do because I'm trying to build a business, and I'm trying to do it in a lean way without scaling headcount linearly with my growth.
And so my problem is, I can't spend all this time putting stuff into Excel or Google Sheets, and then I can't buy a software that basically makes me just put that same stuff into the software instead of a spreadsheet. I need that to happen automatically without me thinking about it. So the data, when I look at it is all there and ready for me to make a decision on top.
That's the problem that we're trying to solve. And maybe there's a couple of different applications that we have to build to go do that. But the problem is not a feature problem. The problem is it takes too much time to do all this stuff. There's too much data entry, and we should be able to just automated it all.
Charles (00:58)
Welcome to Blue Ocean by startups. Each episode I sit down with leaders in consumer brands, supply chain finance and see to talk about the problems they're facing and the solutions they're using or building. I hope you enjoy the show.
Hi everyone. I'm Charles Cushing, the founder of startups. With me here today is Arjun Agarwal, founder and CEO of mandrel. Welcome, Arjun.
Arjun (01:19)
Hey, Charles. Thanks for having me.
Charles (01:21)
Why don't you just share the one liner on Mandrel? Just for anyone who's not familiar with what you guys do.
Arjun (01:26)
Yeah, sure. So at Mandrel, we are building inventory automation for consumer brands, which very succinctly means we're trying to automate all of your inventory and COGS tracking, costing, accounting, planning and reporting in a single place.
Charles (01:41)
There's a lot of product surface area and a lot behind that philosophy of inventory automation that we can get into. So good job condensing it to one sentence. And then I don't usually spend a bunch of time on people's backgrounds. But yours is really interesting and also I think is largely informed what you're building today. So would you mind just sort of taking us through your journey and picking up or highlighting the bits that have fed into this idea, or what you're building now?
Arjun (02:05)
Unknowingly, I think it goes back to the start of my career. Actually, I was an investment banker for technology companies, and so I was working with mostly like late stage private companies, helping them go public or get acquired or get sold. And I ended up spending a lot of time with consumer hardware businesses. As an example, I helped take Fitbit public and different from maybe other technology bankers, learned a lot about the space and understanding the ins and outs of a hardware business versus a software business.
And that actually became even more helpful as I moved earlier in the finance lifecycle. I became an investor at NEA for a few years, where I did spend some time looking at software businesses, but ended up spending quite a bit of time around industrial technology and looking at the same types of businesses building in the physical goods world, but from a even earlier stage, as early as a pitch deck, starting to launch their first products.
And you see a lot as an investor and you learn a lot about what makes these businesses really tough and having to deal with physical product is a big thing. As an investor, your primary focus is deploying capital, and especially when capital is not just used to build a team or hire or spend money on software tools, but it's actually going into physical product, you get a different appreciation for the challenges and uses of capital that you're giving out to these companies and the risk profiles.
And I think what really informed my experience and why we started mandrel was at Desktop Metal, which was an investment that we made at NEA and a company that I ended up joining at that two and a half years into its journey. It's a 3D printing company. We started by designing and manufacturing metal 3D printer, but expanded to offer 3D printers for not just metal but polymers, plastics, sand, biomaterials, a whole bunch of different materials that we supported. And we were primarily selling into commercial and industrial use cases.
But I joined there, and I ran product for a good chunk of the five and a half years that I was there. And I quickly learned that in a physical goods business, product has this physical operations component to it that's very analogous to in a software business, if you're a PM, you have a bizops element to your job. And the physical operations component to a product ended up being me spending a lot of time thinking about and understanding our inventory -not just how much of it do we have and where is it, but what are the costs of the inventory? How much cash have we deployed into this SKU or that SKU? And what are the off balance sheet commitments that we've made? And especially as interest rates rose, margins and cash became a really, really high priority alongside growth, but maybe growth had been more of a focus early on.
As cash and margins became that much more of a focus, the time spent around understanding inventory really came into focus. And we had a whole finance team. We had cost accountants, we had financial analysts, we had accounting people. We had an ops team that was working with our suppliers and getting stuff through the supply chain. And yet I still ended up in spreadsheets trying to understand all this stuff about inventory.
And you would think that in today's day and age, there would be a better tool to do that. But the reality is there isn't. And the reality is that the numbers that your cost accountants give you is different from the numbers that your operations team gives you is different from the numbers that you get when you try and go to the system and piece it all together.
And so I figured that there had to be a better way. And for me, it was again, not just about the quantities, but the dollars and sense of it all. So that was the spark for mandrel.
Charles (05:33)
I feel like it gets even more complex when you're first developing the product because you've got like the actual development costs themselves, which you have to manage, and then you also have to think about the scale costs. You've got like 16 versions of a part coming through that you're trying out or whatever. And I just feel like can get messy really quick.
Arjun (05:49)
It's an interesting challenge because in the discovery process of what is mandrel, what exactly are we trying to build? We talked to companies that were very small and just kind of getting started to companies that are a lot larger, and it's funny how even the companies that are really small and kind of in the product development process, their supply chain inventory position, is just as complex as a much larger company with a lot more products because they're going through the process of trying out new vendors, running iterations of the product, and trying to arrive at an understanding of what does my product actually cost.
When you go through those iterations and through those variety vendors, that's actually a really tough challenge. Even more so if it's not just the finished goods that you're buying, if you have some involvement in raw materials. It's an interesting conundrum as you're developing and launching new products, that understanding the financial elements of that inventory is just as hard as a product that's at scale and being sold pretty widely around the world.
Charles (06:42)
How did you cross the chasm from "Okay, cool. I'm a senior level exec at a publicly traded hardware company" to "I've gone and raised venture capital and hired a team to be building this"?
Arjun (06:52)
We took Desktop Metal public during Covid, and as a public entity, I ran product. I also oversaw some corporate development activities. We acquired a number of companies, and we were going through this process around really looking at inventory and consolidating product lines to get to a better, less overlapping product portfolio. And there was a lot of operational activities that ended up happening as a result.
And I just kind of felt like it was super interesting, actually, for a period of time thinking about product portfolio and product consolidation and strategy around that. But after a while, it kind of just started to feel a little bit too fortune 500. Not that we were at that scale, but what I imagined like a mid-level corp or BD job at a large company would be.
And I really wanted to get back to something earlier stage where you're really focused on building rather than optimizing. And as much as I loved being in the physical goods space, I knew that I didn't want to go build a physical goods company myself, but I really was interested in taking what I had learned about what makes those businesses hard, and using those learnings to help other people trying to do that.
Charles (07:57)
So maybe is like a good way to dive in. Inventory management is like the lifeblood of any business that with a physical product. And yet in 2024 companies still struggle to do this. Why isn't this is not a problem yet? Where do these things break down? Like, can you sort of take us through step by step kind of what the status quo look like and where the friction points continue to arise?
Arjun (08:16)
A lot of the pain comes from the fact that you have to, like, actively use systems to get value out of it. You have to actively enter data into them and maintain the data in them to get value out, whether that is a spreadsheet if you're just getting started, you have to enter data into spreadsheets, enter formulas in the spreadsheets, or that is a true system - you know, software system, IMS, CRP - whatever three letter acronym you want to use. These are tools that you have to get set up. You have to enter data into them on a daily, weekly, hourly basis. You have to maintain that data. When things happen in the real world, you have to make sure that the software world reflects what is happening in the real world.
All these companies spend a good chunk of their time just updating these systems. And it's not that the functionality of these systems isn't good, and they don't ultimately do what the brands are trying to do or understand. It's just that they need to be used to get value out of. Our perspective is that you shouldn't ever have to enter or maintain data in a system of record around inventory.
Fundamentally, that's what we believe, and that's what we're trying to solve for is all of the like, okay, "I'm creating purchase orders", and "oh, now they've been moved into In-transit" - and that requires you going and updating something in the systems. And "okay, now they've been received at the warehouse" - now I have to take an action in software, enter something into Excel. You shouldn't have to do any of that. The system should be automated to reflect what's happening in the real world, so that you can just use the intelligence and the data that it generates to make better decisions.
Charles(09:53)
In some sense, it's almost like you're using human eyeballs and human brains as like an API between what's happening in the real world and what's happening in your system. So how much of the problem is like, okay, there's just friction introduced by the fact that humans aren't robots versus just, I don't know, like supply chains are long, there are lots of stakeholders.
You know, you can get an email that says that there's a ship on the water with two containers of your stuff, and then actually it never left. How much of like the first category versus the second category is the problem today?
Arjun (10:18)
I think it's a little of both. I'm not going to be black and white here and say it's 100% water, 100% the other. I think there's absolutely a component that it can be cumbersome and expensive to have an army of operations analysts, and it's getting cheaper to do that. You can offshore some of this, or you could hire, you know, out of college folks to help you do this, but to get all of the information into a system and to make sure it stays accurate, it does take a lot of time and it takes human APIs, as you said, which I think is a great analogy. There is certainly a component of like "is the data that I'm getting from my various sources accurate"? And there has to be some kind of basic assumptions that you work on. So you assume that it is, and you just make sure that your system is flexible enough to handle updates when you get new information.
So as an example, I have a target date for a PO. Great. Nothing changes. Three weeks go by, four weeks go by. At week six, your vendor says, "hey, sorry, this is going to be late. Here's the new date." That is something that a human API has to understand and now update a date in the system. But there's no reason that you can't take that human API and make it fully automated so that instead of, again entering that data and making the update information where prior information is incorrect or just needs to be updated because statuses change all the time, you know, there's no reason that you shouldn't be able to do that in a more automated way so that that human is now spending time trying to figure out, "okay, this PO is late, what do I need to do now that the PO is late? Do I need to cancel it? Do I need to pay extra to accelerate it? How do we resolve the situation given the new set of data?"
Charles (11:54)
We can break this down a couple of different ways. One is on the spectrum of we are an interconnector that sits as middleware, between sort of existing systems of record and pulls them all the data and makes it all makes sense versus we're building the full stack, replaces your ERP, replaces your IMS, replace your OMS, whatever else you need. How do you think of yourselves on that spectrum? And then how do you think about sequencing as well?
Arjun (12:17)
I'll kind of break it down how we think about it, which is you're either a company that hasn't implemented a system like an IMS or an ERP or you are. If you haven't implemented one of those, we can come in and handle a lot of the functionality that they offer, but a lot of stuff that they don't offer that you would still be handling in spreadsheets or other, maybe more flexible, but less robust solutions like that - whether it's Airtable or Notion or file storage tools or what have you. And so we could come in and I think help put off or eliminate the need for, for some of these tools.
For companies that are bigger and may already have an IMS or already have an ERP system deployed, if you think about all the time that you spend updating the systems, think about us as a way to save some of that time. You can send some of that data to us, and I'll I'll get to that in a second, and we can update a bunch of the actions and then right back into the systems or give you a tool that you can use to analyze the data coming from those systems in a much easier to engage with format and interface. So there's a use case wherever you are in the spectrum of technology adoption.
Charles (13:27)
And so what is the product today? This is - again it's a big mission to automate inventory. Where did you start out and where are you in the journey?
Arjun (13:34)
There is a handful of stuff as you allude to. I think about breaking down inventory automation into a couple of steps. First is data capture, and this is something that I think we do that is a little bit different from some of the tools out there. I think about it from an ins and outs perspective. You have inventory coming in and you've inventory going out, and there's different data sources for each of these elements of the supply chain. The inventory going out side - there's actually a decent number of tools out there that can help you understand this. So, Shopify obviously or other ecommerce platforms: this is where you get your order data, you understand what you need to deliver. Oftentimes, you'll even have fulfillments and shipments recorded in this tools. Even better than that is a warehouse management system. And there's a lot of fantastic cloud based warehouse management systems, or WMS is out there, and we plug into about two dozen of these.
And that's great because they're an even better, more timely source of truth around what is actually being shipped out. And they're a little bit more holistic because they give you not just ecommerce data, but retail data marketing fulfillment. If I'm selling stuff out to influencers that isn't coming in through Shopify or what have you, and so we plug into some of these tools and understand what's happening on the outbound side through these third party platforms.
The ins is a little bit more interesting. There isn't a great source of data on what is happening to my inventory as it goes from P0 to in the warehouse. And the most comprehensive best source of data is actually your inbox, because almost every document and almost every update, whether it's contained in a document or not, comes through your inbox.
Maybe there's some additional tools like WeChat or WhatsApp, and these are things that we think about down the road as wanting to kind of bring into our platform as well. But the inbox is a really good place to start. Every PO, every invoice, every shipping document, whether it's a BOL or a packing list or commercial invoice comes through there. We want to sit on top of that and help you triage all of those communications with your vendors and your supply chain, and extract all that data out and structure it so that now you have all of your ins data and all of your outs data on inventory.
Charles (15:37)
I just want to pause on that for one moment, because, correct me if I'm wrong here, but I feel like one of the things that you're doing slightly differently is really getting to like, ground truth on what container shape something is in, if it's like an item in a carton on a pallet in a LCL container, on a ship somewhere in the ocean, you're going to sort of have that whole picture put together, right?
Arjun (15:59)
We get the whole picture and really importantly, we get it at the item level. So for us at Mandrel, we think businesses make decisions about products. I'm not thinking about making a decision on inventory at the business level. I'm deciding how much of this red sweatshirt should I be buying. And so to do that and to make those decisions easier, all the data in the business has to happen at the item level.
And I think this is where some financial tools have challenges and some operations tools have challenges. Because in finance is you're always thinking about things at the aggregate. My total inventory, my total PO, my total invoice, it's really being broken down and digested at the actual object level. And the same goes in the ops world. It's my shipment or my container or my PO. And we are always trying to bring data back down to the SKU level to understand exactly "for this PO, for this item on this PO, where is it? How much should we order? Or what's the expected cost?" When we get freight costs, "how do those accrue to that specific item?" Everything in our database is oriented around this.
Charles (16:59)
It's beyond the SKU right. Like I think you were showing me a demo a while ago and it's like, all right, you have the SKU, but you swapped out the thread. You could save it as a new version of that SKU.
Arjun (17:08)
As a software provider, maybe our customers work with finished goods or buy finished goods primarily from their vendors or co-man, but we have a lot of them that also work with raw materials. We actually use the word item to convey not just finished products or finished goods, but also raw materials. We're agnostic. We are always tracking at that individual item level to understand what's happening in the business.
Charles (17:29)
Am I remembering this correctly, this sort of like the main interface for this is this really sophisticated, detailed item master. Every time there's an update, it writes to this source of truth, of where everything is in the supply chain.
Arjun (17:43)
That's right. So you can come into Mandrel, and kind of the foundational thing that we have, or the foundational module is an item master - product data management as we like to call it. And that's both raw materials and finished goods. We help you store, manage, keep this information updated. We sent it to and from third party channels like your commerce platforms, your warehouse management system.
And so when you onboard, you can just plug those in. Or a lot of brands that we work with have a pretty robust spreadsheet, especially if you're buying raw materials, which you might not store in a system like a Shopify. But you can get that information into Mandrel, and then use us as your source of truth there.
And then every time we're ingesting a document or ingesting a fulfillment or an order from a third party system that always looks back to that item record in Mandrel. And so you can see not just general information like "what are the dimensions? What are my UPC codes? you know, what are the weights? How many items go or how many units go into a carton and into a pallet?"
All that information we store, but then we're also grabbing all the information on "what are all my orders for this item? What are all my fulfillments that have this item? What are all my POs that have this item and downstream documents that have this item? And what is every basically event that happens and this item's lifetime?" So we start to build out this ledger of activity at the item level.
Charles (19:06)
So just like for a sort of extreme example here: if I place an order at the raw material level for 100 gallons of pink dye to go into some plastics that go into my product, can I trace them through your system to say, and those landed at the doorsteps of these thousand addresses, like, is that the sort of level?
Arjun (19:26)
Yeah, you should be able to do that, because we're tracking all the way from we placed a PO, to it got on a boat, to it got sent to the vendor, to it got consumed during a manufacturing run, to hey, now there's cost associated with that consumption and the freight associated with that the shipments of the raw materials are now getting sent as part of the finished goods to my warehouse and out to a customer. So you can see how not just the quantities, but really importantly, the costs all accrue to the end customer. And that's really the second big thing that we're helping to automate for these brands is SKU level landing costing. And we do this in a way that is fully automated.
Charles (20:06)
So you get all the data into the system. You could do manipulations on it. What are like the sort of key outputs that a user will get from using this. Right. Like sounds like one just accurate data entry into other systems of record. So if you have an ERP, it's not just like this jumbled mess of how do I attribute these freight costs of these cases of goods. It's like, no, you have that broken down. Another sounds like it's some kind of analytics reporting, decision making. How do you think about this categorically?
Arjun (20:30)
The best way to think about this is what are the workbooks that a brand may already have. So one is they probably already have a workbook and Google Sheets or something that is all their documents that they received. Or maybe it's a file storage folder where they've organized it. And if you need to access a document, you got to do a bunch of searching and opening up and closing. We help you with all your document management, and we extract all that data to make it searchable, sortable, filterable - we have it all there and we look it back to the item.
The second thing that you can do is there's usually a cost allocation or a costing workbook, where every week and ops person or finance person takes the data from these documents, enters them in, and splits out the cost to arrive at a landed all-inclusive cost for items. We do that automatically and we give you a view of landed cost for every item that you have - raw material or finished goods - that is live. And if you click into an item, you can see how that changes over time by shipment. So that's kind of the second workbook that we're replacing.
The third workbook that we handle is your overall inventory quantities. Because we are tracking all the ins and we're tracking all the outs, wecan give you a live, kind of responsive view of both raw material and finished goods by every location and by what state is it in. So "is it in transit on the way to warehouse A? Is it on hand at warehouse B? Is it on hand at vendor A or co-man A?"
We help you track all of that, and in addition to quantity, we show you value and unit costs at the item level. So again not just the ops side of things with quantities but the dollar side of things. And that is kind of what we use to help derive the landed costs that I mentioned earlier, as we're looking at not just the landed costs overall, but of what's in my warehouse right now, so the next item that I ship, it's going to be at this cost.
The next thing that we help you handle is creation of POs, revision of POs, and monitoring those POs as the inventory comes inbound into your warehouse. So every brand has this PO tracker spreadsheet that's like, "okay, here's my PR number, here's the line items, here's the current status, in-transit. Have I receive my freight invoice or my invoice yet? What were the quantities and costs? Have I received my freight invoice? Have I received my commercial invoice? Have my receive my goods receipt?" And they're like entering in all that data. We basically have digitized this spreadsheet and fully automated your inbound PO tracker to help you manage all of your POs as they come into the warehouse. So that's another workbook that we're helping to replace.
And then we get into downstream uses of all of this data, which is inventory and COGS accounting. So we can, at the end of every month, take that inventory value and use it for accounting purposes. We can take the over time cost of the goods that you've sold within a month and use that for writing into your accounting system for your monthly COGS.
And we can use the same data to help you with inventory planning. So if we understand the ins and outs, we understand where your POs are inbound. We can really easily update the actual status and kind of the inventory forecast to make inventory and demand planning a little bit easier and ultimately recommend your POs.
Charles (23:40)
Just zooming out to like the market level a little bit. Obviously, there's a lot changing, generally speaking, in commerce and in the tools that serve commerce. What would you say is your view of the macro? How is the space that you're serving going to evolve? And like how do you see yourself fitting into that?
Arjun (23:56)
I think this is already starting to happen or things are moving in this direction, but we hope there's a future where one person can run a $100 million brand, or a $500 million brand, or even $1 billion brand someday. We spend a lot of time thinking about what needs to be true about the set of tools - whether that's software or services or some combination of those - what needs to be true of those tools to enable that future?
And for me and for the team at Mandrel, it kind of starts with, if I'm one person running this whole company, I need to ruthlessly prioritize my time. I can't be spending time updating data and systems. I need the systems to tell me what's happening in my business, so that I can make a decision that much faster.
That's where our focus has been today, is how do you automate a lot of the processes, particularly around inventory operations and finance. But it's a future that's already coming to pass. You see influencer brands that have a one person marketing, sales, everything go-to-market machine, and those brands are able to be really efficient. There's no reason that all brands shouldn't be able to be so efficient with the variety of marketing channels that are out there today. I think a big thing that gets in the way is ops and finance. There's so many great tools out there for marketing and sales, and we really want to build these awesome, cool automated tools for operations and finance teams as well.
Charles (25:21)
It's a big mandate. Software is very specific, and the conventional wisdom often has been like, pick a point and and focus on doing one thing really well. But I guess in this case, the thing is just a really complex thing. And so you kind of have to do it all.
Arjun (25:35)
We're not building features here. We're trying to solve a problem. And the problem in my mind is not like I need to do inventory planning or I need to manage my PO's. The problem is I have too much to do because I'm trying to build a business, and I'm trying to do it in a lean way without scaling headcount linearly with my growth.
And so my problem is, I can't spend all this time putting stuff into Excel or Google Sheets, and then I can't buy a software that basically makes me just put that same stuff into the software instead of a spreadsheet. I need that to happen automatically without me thinking about it. So the data, when I look at it, is all there and ready for me to make a decision on top.
That's the problem that we're trying to solve. And maybe there's a couple of different applications that we have to build to go do that. But the problem is not a feature problem. The problem is it takes too much time to do all this stuff. There's too much data entry, and we should be able to just automated it all.
Charles (26:29)
Arjun, thanks so much for your time today. It was great talking about all this and appreciate you joining us on the show.
Arjun (26:34)
Yeah. Thanks so much for the time, Charles. And happy holidays. Yeah.
Charles (26:37)
You as well. Thanks to everyone listening. And if you want to learn more about Mandrel, we'll link all the relevant details in the notes. Till next time.
I hope you enjoyed this episode of BlueOcean. BlueOcean is produced by StartOps, a growing collection of resources for operations leaders at ecommerce brands. A few of the things we built include a growing, vendor free slack community for operators to ask their most important questions and get answers from trusted peers. A job board that curates jobs from 1500 modern brands, a media hub that includes this podcast as well as articles and templates on a variety of operations topics, and a comprehensive directory of the software and services used by our community members.
If you want to learn more, head to startops.network and to keep up with the latest information from StartOps or catch the most recent episodes of BlueOcean, follow the startups page on LinkedIn or subscribe to our newsletter. Thanks for listening and hope to catch you next episode.