Listen to this episode:
About this episode:
In this episode of Behind the Data, Matthew Stibbe from Articulate Marketing interviews Nick Liles, IT Manager at Sisk Fulfillment Service. They discuss the advanced printing technology used in fulfillment, the importance of high touch fulfillment for nonprofit clients, and the challenges of managing data in fulfillment operations. Nick shares insights on how they implemented CloverDX to automate processes that were previously manual, improving efficiency and data integrity. The conversation highlights the significance of quality control and the transformative impact of automation on the fulfillment industry.
AI-generated transcript
Matthew Stibbe (00:02.19) Hello and welcome to Behind the Data with CloverDX. I'm your host, Matthew Stibbe, and I'm with Nick Liles today, who is IT manager at Sisk Fulfillment Service in sunny Maryland. Great to have you on the show, Nick.
Nick Liles (00:15.745) Thanks for having me.
Matthew Stibbe (00:17.57) And just before we started, you were telling me about your printer and I was having geek delight at it. So can you tell our listeners, tell about this amazing printer that you've got?
Nick Liles (00:28.92) Sure. So as a fulfillment company, we do personalized printing for a lot of our clients. We do, you know, a lot of it is black and white, but we do full color printing. And in the last year we have acquired, it's a Canon V1350. It is a full color digital press. It does Pantone color matching and it prints at about 145 pages a minute. And you can do that in tabloid size paper. It is a massive beast of a machine.
It's about the size of two full-size cars put together takes up an entire wall of a room. But it can do extremely accurate color output. So when you have to talk about a typical desktop printer, like it can print color. But if you've ever had to do Pantone or worked on corporate logos, people want very specific color matching. And that can be a painful process with a digital printer. So typically, you have an offset printer that will mix inks and make those colors.
You are a bit limited with the CMYK because certain colors that are pretty difficult for it to hit and you kind of got to know what the limitations are but as long as you're within that range this machine can actually sample that color in line while it's printing and make adjustments which is something that pretty much no other printer can do. You kind of have to print, make an adjustment and then go in and tell it what colors you need and this thing can do that all on the fly while it's printing at that speed. Just very impressive.
Matthew Stibbe (01:53.61) And this fits into the fulfillment business in what way?
Nick Liles (01:59.778) Yeah, so we do a lot of fulfillment for nonprofit clients, even some for-profit that want to do membership programs and kit fulfillment. So like a high touch fulfillment program. So an example might be you would donate to a nonprofit company and they're going to send you a free gift and recognize that gift. So we will print out, you know, the personalized letter, put the tax receipt at the bottom. So hit all those variable data points and you know, there may be a certificate that goes along with it that needs to get a printed name or some other printed material.
We also can do variable graphics on the fly so for some of our clients, they may have a different animal species that you can adopt, and they will print that on the letter in color on the fly. That matches an output of the actual products that have to get packed into the box for production.
They come out in sequence, right? So you could print all of your tigers and then all of your panda bears, but you'd have to then shuffle them out in production to match them up with the orders. So being able to do that variable data printing on the fly is a big help efficiency wise.
Matthew Stibbe (03:08.342) This and tell me a little bit about the fulfillment business generally. You've given me this wonderful example of adopting animals and things, but what sort of things are other than that are you doing? Tell me a little bit about Sisk.
Nick Liles (03:20.471) Sure, sure. So I usually tell people that, You know we're basically like an Amazon, but we don't do the sales part. So all of our clients manage their own storefronts. They maybe use Shopify, Magento, WooCommerce. I mean, if you name it, we probably interface with it in some way. Most of the major shopping carts. And we have some customers that sort of roll their own and integrate in with our API directly. So there's half the business that we would call like pick and pack fulfillment. So you're going to a warehouse, you're picking products, you're putting them in a box and you're getting them mailed, which is more like the Amazon side.
But our niche in the industry is the like high touch fulfillment. So again, it's the same sort of thing. You're picking products from the warehouse and you're packing them in a box, but maybe they're getting gift wrapped or tissue paper. And again, getting matched up with that personalized printing. So it's not about get it in a box and get it to my house as fast as possible.
You're, you know, this is a recognition of a membership or a donation. So you're the, you, our company is the is having to be the face of that organization. So we are interacting with their constituents and it's important that those products are presented in a way that I guess represents that company and not just tossed into a box with some bubble wrap and sent on the way.
Matthew Stibbe (04:34.892) And what sort of volume of things are you doing? How many shipments or how many bits of paper or how many packages are going out?
Nick Liles (04:41.719) Yep. we, uh, want to say last year we did somewhere and they were at about 4.5 million packages that went out the door. Now, some of those are just envelopes. Um, those are not all necessarily parcels, but 60% of those probably are a little better. Um, we do on average somewhere around 200,000, maybe 300,000 prints a month, um, personalized prints. Some of those are duplicates. Again, we have orders that have a letter and a certificate. So I mean, it counts as two prints but really would be one parcel going out the door. The majority of our stuff is single parcel. So it's just one package going somewhere. We do have sometimes they do multiple, maybe they have sets of things, two different boxes. One thing is it's two packages in that set. We will do just about anything you could think of that you would want to get mailed. If it takes custom manual labor to make it happen, we have the people in facilities to pretty much do it.
We have clients from nutritional supplements that require temperature controlled and humidity controlled storage. So that's something that we offer. We've done food fulfillment. So no direct contact with food, but say you are promoting a movie night, your organization's going to do a, you when everybody log into Zoom and we're going to watch this movie together sort of as a thing. And maybe we've mailed everyone like bags of popcorn and personalized buckets that go together and that kind of thing. Again, it's more of a high, you know, a high touch fulfillment than like I said, just get this person a box of popcorn, right? So that would go in, there'd be a letter about the night and why they should attend and that sort of thing.
Matthew Stibbe (06:28.078) So we've sort of talked about the volume and complexity of the physical stuff, but all of this is being driven by data that's coming in from Shopify and WooCommerce and what. So tell me, first of all, on your side, what sort of systems are you having to put in place to process that? And then from the client side, where is the data coming from and in what format?
Nick Liles (06:33.249) Mm-hmm.
Nick Liles (06:50.059) Yeah. So that, yeah, you've touched on a little bit. It is a wide array of things that come into us in pretty much every imaginable format. So everything from our most sophisticated customers that again are directly interfacing with our WMS API, and they're able to send us orders that way, all the way down to smaller clients that, you know, they're just putting names and addresses and details into an Excel spreadsheet and sending it to us. And pretty much everything you could think of in between.
Comma separated, tab delimited, fixed length, if you name it.... And the method of those getting to us is just as wide, right? Some clients want to email it, some post it to our FTP, some have us retrieve it from their FTP, some use a SharePoint site. If there's a way to obtain data from someone, we've probably encountered it at some point in time and had to work with it. So it's, I mean, yes, and it's a lot.
And what we need to do with some of that data varies greatly. Again, we have customers that have sophisticated IT departments and they give us exactly what we need. We process it and get it done and that's the relationship. And then we have other customers that sort of just dump everything onto us. They give us a set of rules about how we should, you know, maybe they have some rules about, you know, hey, if a person donates X amount, they get this package. If they donate Y amount, they get this package. If it's over, you know, Z amount, they get this package.
So we have some rules we have to put into place to figure out what should these people be getting. It's great that clients trust us to do that, that's also... the data integrity is super important there because if we get that wrong, we're going to send people the wrong stuff and potentially take a nonprofit client into the red because maybe they're not making any money off of all the gifts they've given out based on the donation levels.
Matthew Stibbe (08:41.358) And internally that data has to somehow presumably get to the warehouse for the people to pick the right things. It has to get to the printer to print the right things. You mentioned a piece of software, WDM, think I heard, but internally what are the systems that have to absorb the information?
Nick Liles (08:56.137) Yep.
Nick Liles (09:00.865) Sure, sure. So our main system that we use is something called VeraCore. It is a WMS, which is warehouse management system, and an OMS, order management system, and they're paired together. So it's a unique piece of software for the fulfillment business. I've been working with it now for about 15 years of the 18 years that I've been at Sisk. Just for some context, prior in my early days here, we used an old AS400 system.
And people were out in the warehouse making changes, writing it down on a piece of paper and taking it into the office and sort of key it in on that little dumb terminal. So that was horribly inefficient and certainly not real time. So VeraCore was one of the first major projects I tackled here, implemented that system that allowed us to start using handheld scanners in the warehouse. So I was able to barcode all the inventory that came in.
So when orders would go through the WMS, it would generate a pull sheet that said, Hey, here's all the products for this batch of orders that we need to get, so those guys could go around the warehouse, scan that inventory, and we could get basically real time inventory movements around the building, with VeraCore.
So that, that's probably the linchpin in the whole operation here because you have the order management side, which is what clients care about and what they want to see. You know, they want to see when an order came in, how long it took for it to be fulfilled. They may want to get tracking information, that sort of thing from that. And then they care about their total overall inventory. But from an organizational standpoint, you know, we may have a customer that has 14 pallets of the same inventory. We need to know where that is in the building and exactly what's on that pallet at any given time. So those two systems work together. I know there are some disjointed things out there that you could probably help marry, but having those two things sort of embedded and grained in with one another, so the WMS sort of gives us our single pane of glass of what's happening inside the building where the OMS is like isolated like to a client so they can only see their data and their inventory and run reports upon that.
Matthew Stibbe (11:01.326) And so all this data from all these directions has to get into these two systems to make the magic happen. And that brings me to the next thing I wanted to explore with you. You told me when we were talking earlier a little bit about a project you've been working on recently to try and streamline or automate that. Can we start with what was happening before you did this project? What were you trying to automate?
Nick Liles (11:28.375) Sure. So I'll give you a specific example. So one of our clients, big clients is World Wildlife Fund. They have a program where you can donate to adopt a species of animal from them. so there's a, I don't know what the total number is, but I'm going to say over a hundred different species of animals that you can adopt from them. So we have each one of those plus in the warehouse, set in a location. That file comes in from us and there are a lot of different types of, I guess, processes that happen for their fulfillment. So there's different levels that you can donate. And depending upon that is the type of gift that you're going to get back from them. So it could be a small plush, could be a large plush. In some cases, it's a collection of different plushes. They have a group of meerkats.
Matthew Stibbe (12:16.04) So little toys? So you adopt a panda bear and you get little panda bear? That's the cutest thing ever.
Nick Liles (12:25.237) Right, right. So it's like a, that's right. Yeah, yeah, yeah. So when those orders come into us, the first thing is we have to get that data into a format that our WMS wants to ingest. And because this involves personalized printing, there are some extraneous pieces of data that we have to have. So obviously you need the name and address and where it's getting mailed to.
But depending on the client, everybody has sort of different requirements. They have specific points, they may have a specific way they want a name formatted to be printed on the envelope, different than the citation name, I guess, and it's formatted a certain way. The donation amount, the date, they may have an ask amount if they're going to say, hey, we appreciate you donating for this, if you can match this other amount, it would help us greatly. And again, it varies from program to program. But we have to get that data processed in, do some data integrity checks, and get that done.
Previously, we're doing most of this, what I would say, is manually. We would take that file and through a piece of software called Mail Manager that's designed to do postal presorts and those types of things. It has some ability to do error correction. Obviously has all of the tools you would need to correct an address. It allows you to apply some minor, you know, if statements and maybe some lookup tables for some very basic sort of things. But even then it's got some weird limitations because it's not an ETL software. It's designed to do the mailing.
So a lookup table, you can only look up four characters. Which you can work around that, but sometimes it gets real tricky if you have something that's an eight character string, like, all right, how can I make this work if I'm only looking at four of the eight and those kinds of things? So we had a lot of little workarounds and quirky things that we just took the tools that were in front of us and made it happen. But again, a lot of it was happening manually in Excel. So we would open the file. We were taking a look at the data in there and making sure there were no errors and those types of things.
And about two years ago, we got CloverDX and we were able to use that to build those same processes in an automated fashion. So instead of having someone have to be present, in fact, personally, in their busiest time of the year in December, I used to have to get up at 2 o'clock in the morning and process these files. So their workload would get very heavy, obviously, in the Christmas time and the season, so the average would be...
Matthew Stibbe (14:49.004) Was it a case of manually checking Matthew at this address and putting it into the system?
Nick Liles (14:52.435) Yes.
Well, I mean, it was, you know, they would send a file over on the FTP and I would have to bring it into that mail manager software, run it through and make sure it didn't find any records that had glaring address problems. And there was some integrity to that data. And, know, that software had no ability to be automated. So I had to open that software up, click a box that said, here's the file, import this file, run the process, tell me what's bad, what's good, do some filters, output that into somewhere and then go into that WMS and say, okay, I now need you to import this file. And then I have to go in there and say, okay, now process these orders and get them out and get the output for the warehouse on the other side. And that, that process and start to finish required hands on. There just wasn't a way in that moment to automate that with the tools that we had in front of us.
Matthew Stibbe (15:41.132) Not great when you're doing it at two o'clock in the morning because you've got to get the orders out the next day.
Nick Liles (15:43.48) No, and certainly not your peak time to be finding issues. It took a lot to focus at 2 o'clock in the morning to make sure that I wasn't skipping any steps or was finding that. And that's tough, right? It's tough from an organizational standpoint because you're relying on somebody to physically be there and do that, which puts you in a vulnerable position if somebody's ill or something comes up or there's connectivity problems. I certainly wasn't driving into the office if I could help it at 2 o'clock in the morning, but you know, if there was a connectivity issue, I had no choice. You know, you have employees coming in in the morning. So it's critically important if that stuff is not ready, they're coming in with nothing to do in the morning. So it was, you know...
Matthew Stibbe (16:23.296) Each of those little steps somebody has to know how to do it and remember how to do it and if that you know they're sick or something I guess that's also a risk isn't it.
Nick Liles (16:27.223) Mm-hmm.
Nick Liles (16:32.555) Yep. So, you know, we organizationally had SOPs and standard operating procedures and we had other people that were trained to do it. But it's just, you know, again, if, you wake up at midnight and you're having stomach issues, you know, who are you going to call to say, Hey, in a couple of hours, can you cover for me at 2am? You know, if you plan ahead, sure. But, this is, it's just tough. And it was a vulnerable sort of spot for us for a long time that, I was glad that I'm a pretty reliable guy was got us through all those times, but it's still a vulnerable position to be.
Matthew Stibbe (17:03.5) And how did you start to automate that with Clover?
Nick Liles (17:07.029) Yep. basically, most of our jobs are sort of split into two parts. So we have what we call the import, and that's getting the raw files from the client, getting them into our WMS and all of the process that goes there. And then we have what we call the backend, which is the data processing with our variable data printing. We talked about the printer earlier happens. So the first step with that was to get the front end processes automated.
So we sort of looked at all of our clients. sort of just picked and choose the ones we felt like had the most manual operations that were happening on the front end. You know, over the years we had some clients, again, I mentioned some had more sophisticated IT and sort of gave us things the way we needed them to be as opposed to a client who maybe, you know, their system maybe isn't flexible enough and they're like, hey, this is just what comes out. We need you to work with this to get to this result. So we started at the top of the list and worked through and World Wildlife Fund is pretty high at the top. I mean, again, their program is pretty involved. They have a lot of different like intricate programs. there was, again, there was lot of manual steps were happening. So that was one of the first ones we tackled. It took that to a process now that was completely hands off inside of Clover. So everything that we were doing by hand happens completely overnight. I don't have to touch it. When that file shows up at two in the morning, it's done, ready to go. The only time I get woken up is if there's a major issue that, you know, with the files completely in the wrong format or something like that.
Matthew Stibbe (18:29.964) And tell me about your process of taking the manual steps in the software and sort of transposing that into a Clover workflow.
Nick Liles (18:40.535) Sure. So again, we have mentioned that we have standard operating procedures. And I think one of the things that happens, and it probably happens across many organizations, but it certainly happens here, is that over time, a client will make changes to a process, and relatively minor, right? Like, hey, we're going to add in this new program. So it adds a little twist to whatever you were doing before. And some of those things were getting lost.
Somebody, the customer service rep would come in and say, hey, the client wants to do X and Y. No problem. We'll make that change. They sort of mentally knew that they made that change and they did it every day, but it didn't quite make it into that standard operating procedure. And that was one of the first things I guess that we sort of looked at was we took that standard operating procedure. We recreated that process in Clover. So whatever steps we were doing manually, it's like, how do we redo this in Clover?
And on top of that, you have the issue of, you know, a person is manually looking at it, you can kind of pick out, like, this column is supposed to be a dollar amount and it has some long string of numbers. Well, we know not every one of these people donate millions of dollars. Like, what happened with that data? But that's not something necessarily, or was something we were concerned about moving into an automated process. Sure, numbers are numbers. If I tell it this field is numeric, how can I set some constraints against that to say, should be a reasonable number? And what is a reasonable number to fit in there?
So that you could, you know, sort of replicate that human, Yes, I've looked at the data. It's obvious to me that something is errant and teach a computer basically how to perform that process. So that was probably the biggest hurdle when we first got started was just sort of figuring out like, how do we replicate these things that are not really tangible actions, but more of like, well, somebody else is looking at it so they could see if something's wrong. How can we teach this software to, to replicate that?
Matthew Stibbe (20:28.172) Yeah. How did you do that? I mean, was that done? How did you figure out what those reasonable ranges were?
Nick Liles (20:35.187) Yeah. So a lot of it was taking historical data. So for most of our clients, and not all of them, again, everybody has different requirements for PII and data retention and things like that. So some of this is not always possible to go back and look at historical data. But some clients allow us or ask us to retain data for longer. And in some cases, we have like, compliancy reasons why we have to retain data for a certain period of time.
For this customer, we had about two years worth of data that we keep for them on a rolling basis for them to be able to run reports on and things like that. So we were able to leverage that to look at what's the range of donations that have happened in the last two years. So that gave us sort of a baseline for that particular data point. Some of the other stuff like addresses, we've been in the business, Sisk has been in the business for 42 years. I forget exactly what year, I think it was 1983, so 42, 43 years, but a long time. And over those years, we've obviously seen just about everything. So we had some internal knowledge of when I say someone was looking, I mean, you knew what you're looking for. You know how an address should be formatted. You know those types of things. So we kind of knew what pitfalls we were looking for every day. We just needed to create an expression like how do we replicate that.
So an example might be a street address needs to have the house numbers and the actual street name. And if all you have is a numeric value in the field, that's an indication that there's a problem. So we can look at that field and say, hey, if this only contains numbers, kick it out into this error process flow that's like, hey, this needs some attention. Probably can't deliver it to just one, two, three. We need to at least know which street it's going to. So yeah.
Matthew Stibbe (22:14.262) ...the actual street as well. What's been the implication for you and your colleagues when you automate a process that used to take some hours to do manually and now it's being done in minutes in the software? What changes in the organization as a result?
Nick Liles (22:33.835) Yeah, so some of the biggest changes are that, you know it allows my team to have a better focus on their quality control, right? So when you're in the weeds and you're having to do these things, the first thing that I notice is that you tend to get complacent, right? You do the same thing every day and you never see a problem and you just sort of zone out on it after a while, right? Like it's hard to look at a process every day and think what's wrong with this. And it even more so if you have to do multiple steps with that process. So if you look at it once and you've came to the conclusion that it all looks good, and then you have a secondary process, it's very difficult to look at it a second time because you looked at it once. This has to be correct. I know it has to be. And occasionally it's not. And that would come out to bite us every once in a while. We'd miss a step and everybody would tap back and it's like, how did I miss that? It was so obvious that there were no city names on these addresses. How did I miss it? But I think that just is what happens, you get complacent.
So being able to automate that doing every day can allow them to focus a lot more on the output and do quality checks to make sure that, this final output is definitely what it should be. And they can really focus on that. Because again, when you're in the weeds and you're stressed about, I've got to get this done in a certain amount of time, because again, the warehouse is counting on me. Other people need this job. The first thing that tends to slip is your quality control, which is not good, right? I mean, it's probably the most critical piece of our business is ensuring that data integrity. So we know it reflects poorly on us and it's very important that we take time to make sure that that stuff's happening.
Matthew Stibbe (24:13.994) It seems to me always better to have human beings do what humans do well and let computers do the dull stuff, basically.
Nick Liles (24:19.831) Exactly,
Matthew Stibbe (24:13.994) So as we're sort of almost out of time, I'd like to ask you one last question. If you went back to the beginning of that project, say six months ago, not the very beginning of Clover, but the beginning of this round of automation, what's the one thing you know now that you wish you had known then.
Nick Liles (24:44.789) Yeah. So I, one of the biggest things there is just the amount of time that people were spending doing those manual processes. You know, we had SOPs, I was aware of how much time it was taking them every day to get a job done. But once we started digging into that process with Clover to try to replicate all of those steps, it really became apparent of how intricate and how there were just so many things that they were doing every day.
You know, working in Excel, copying and pasting into eight different workbooks and making sure that that stuff was all straight. I mean, that's mentally taxing for somebody. So it's no surprise to me that there were occasional slips in quality because you spent so much focus on that. And it's hard to believe that you just did something wrong when you go to check the output because you did the front end.
Matthew Stibbe (25:32.268) very hard to mark for an exam paper, isn't it?
Nick Liles (25:43.124) Exactly exactly. So that was probably the biggest thing that I that was, I wished I would have known more in the beginning or would have realized sooner is just how much work had come out of it because those processes have sort of developed organically. It was a program that probably 15, 20 years ago only took us 15 minutes to do. But over the years, as the complexity increased, the more demands, the more requirements, those things just got built upon, built upon. And somebody who did it every day was very efficient at it. So they didn't make it seem like there was a lot happening there. And it wasn't until we sat down to dig into that process that it was like, there are hundreds of steps that you're doing every day. It's amazing that you don't get this wrong all the time.
Matthew Stibbe (26:17.23) This knowledge management seems kind of almost reverse engineering human know-how. It seems like the hardest thing of automation. I think Wernher von Braun said, going to the moon is easy, but the paperwork's really difficult. And it sounds like the capturing this and distilling it into the software is the challenging and intellectually satisfying part of it, isn't it?
Nick Liles (26:23.319) Mm-hmm.
Nick Liles (26:39.287) Yeah, absolutely. And it is so rewarding when we get finished one of these projects to take a step back and realize how much time we've saved. So in that project we just mentioned, we took a process that was about two hours of manual labor every day down to something that takes 12 minutes and happens without anyone touching it, which is to me an incredible time savings. Like you said, it frees up my team to use their brain for things that are much more apt to their skill set than just doing those motions every day.
Matthew Stibbe (27:11.806) Well, that's an amazing result. think getting rid of two hours a day of tedious manual error prone labor is, you know, on the side of the angels. So Nick, I'm grateful for that story. And I think that's a great place to bring this episode to a close. Nick, thank you so much for being with me. And everyone listening at home, if you'd like more practical data insights, or if you want to learn more about CloverDX, please visit cloverdx.com/behind-the-data. Thank you for listening and goodbye.
Share
