Latest / Logosofia no Dia a Dia / Pallet Puzzle: Picking the Right Freight Tool
Transcript
- [Prompt] custom: Are there any applications that you can recommend for Android or computers that allow you to calcula | Hosts: corn, herman [Response] Corn: Daniel sent us this one — he's looking for an app, Android or desktop, that can calculate freight volumes for consolidation. The scenario is, you've got a product with a specific volume, you want to know how many units fit into common pallet sizes. He figures somebody must have built a calculator for exactly this purpose. And the short answer is yes, they exist. The longer answer is that the landscape is weirder and more fragmented than you'd expect, and the tool you pick changes the answer you get. Herman: It really does. And the core problem here isn't just "how many boxes fit on a pallet." That's the naive version. The real problem is consolidation across multiple SKUs with different dimensions, weight limits, overhang rules, stackability constraints — and the math explodes fast. You're not just dividing one volume by another. You're solving a three-dimensional bin-packing problem, which is NP-hard. There's no single perfect solution, only tradeoffs. Corn: It's one of those things where the question sounds simple — "just tell me how many fit" — and the answer is actually "well, which algorithm are you using, and what constraints do you care about? Herman: And most people don't realize they're making that choice implicitly when they pick a tool. A greedy algorithm gives you one number. A genetic algorithm gives you a different number. Same boxes, same pallet, different answer. Corn: Which is why Daniel's assumption that "somebody has made a calculator for exactly this purpose" is correct, but also undersells the complexity. Somebody has made several calculators, and they don't agree with each other. Herman: With freight rates still bouncing around — we've seen spot rates swing fifteen, twenty percent month over month depending on the lane — every cubic foot you waste is real money. If you're shipping regularly and your pallet utilization is eighty percent instead of ninety-five, you're buying empty air and paying a carrier to haul it. Corn: The invisible inefficiency. Nobody gets a bill that says "line item: wasted space." It just shows up as more pallets, more shipments, higher cost. And you never see the version where you did it better. Herman: And the tools to fix this are surprisingly obscure. They're not advertised. You don't see banner ads for pallet optimization software. It's this quiet ecosystem of web calculators, desktop apps, a few Android offerings, and a whole lot of spreadsheets that someone's operations manager built at two in the morning. Corn: Which is exactly what we're going to dig into. What tools actually exist, what algorithms they're running under the hood, and how to pick the right one without accidentally optimizing for the wrong thing. Herman: Let's define the actual problem, because it's worth being precise. Freight volume consolidation is taking a shipment — maybe it's a hundred units of SKU A, fifty of SKU B, seventy-five of SKU C — and figuring out the minimum number of pallets to pack them onto, given real-world constraints. And "real-world constraints" is where it gets messy. Corn: Because in the abstract, you'd think it's just a geometry problem. Divide total cubic volume by pallet volume, round up, done. Herman: Which works exactly once — when every box is identical and perfectly divisible into the pallet dimensions, and you don't care about weight, and nothing overhangs, and gravity isn't real. The moment you introduce a second SKU with different dimensions, you're no longer tiling a rectangle. You're solving a three-dimensional orthogonal packing problem. And the number of possible arrangements grows combinatorially. Corn: The gap between what Daniel's asking — "how many units fit" — and what the math actually demands is basically the gap between arithmetic and computer science. Herman: And it's not just academic. If you have three SKUs, say a twelve-by-ten-by-eight box, a fifteen-by-twelve-by-six, and an eight-by-eight-by-eight cube, all going onto a standard North American pallet — forty-eight by forty inches, forty-eight inches high — the number of ways to stack and arrange those is enormous. A greedy algorithm might give you five of the first, three of the second, two of the third per pallet. A genetic algorithm running overnight might find a layout that fits six, four, and one. Same boxes, same pallet, different answer. Corn: If you're shipping a thousand pallets a year, that one-box difference per pallet compounds into real money. Herman: Tens of thousands of dollars. And here's the thing — most people don't even know they're getting a suboptimal answer because they only ever run one calculator. They type in their dimensions, get a number, and assume that's the number. Corn: The calculator becomes the authority, not the math. Herman: And the landscape Daniel's stepping into is genuinely fragmented. There's no Google Maps of pallet optimization — no single dominant tool that everyone uses. Instead you've got web-based calculators like PalletStacking, which is free and uses a greedy bin-packing approach. You've got desktop software like LoadMaster, which runs a genetic algorithm and costs about two hundred dollars one-time. And on Android, there's FreightOptix, which uses heuristic methods tuned for mobile — it's five dollars a month. Three different tools, three different algorithmic philosophies, three different answers for the same input. Corn: That's before you get into the spreadsheet underground. Herman: The spreadsheet underground is real. But the broader point is that the tool you pick is making an optimization decision on your behalf, and most tools don't surface their assumptions. They don't tell you they're ignoring weight limits. They don't tell you they're assuming all boxes can be stacked. They just give you a number. Corn: Which is why the search Daniel's doing — "is there an app for this" — lands in this strange territory where the answer is yes, but choosing the right one requires understanding what you're actually optimizing, and what you're willing to sacrifice to get there. Herman: Let's actually walk through those three tools, because the algorithmic differences are where the real story is. PalletStacking — free, web-based, runs right in your browser — uses what's called a greedy bin-packing algorithm. The way it works is essentially first-fit decreasing. It sorts your boxes from largest to smallest by volume, then places each one into the first pallet where it fits, scanning layer by layer from the bottom up. It's fast. You hit calculate and get an answer in under a second. But it's greedy — it makes the best local choice at each step without looking ahead to see if a different arrangement would leave less wasted space overall. Corn: It's the algorithmic equivalent of packing your suitcase by just grabbing the biggest thing and shoving it in, then fitting smaller things around it, without ever pulling anything back out to try a different layout. Herman: That's exactly the mental model. And for a lot of use cases, it's good enough. If you're doing a one-off shipment, the difference between eighty-two percent utilization and eighty-five percent might not matter. But the gap gets wider when you introduce mixed SKUs with different footprints. A twelve-by-ten-by-eight box and a fifteen-by-twelve-by-six box don't nest neatly. The greedy algorithm places them in sequence and never reconsiders. Corn: Whereas LoadMaster takes a completely different approach. Herman: LoadMaster runs a genetic algorithm. It generates a population of random packing arrangements — hundreds or thousands of them — scores each one on pallet utilization, then breeds the best ones together, mutating and crossing them over across multiple generations. It iterates until the improvement plateaus. This takes longer — could be thirty seconds, could be a few minutes for complex mixed loads — but it explores arrangements a greedy algorithm would never find. That's how you get from five of SKU A, three of B, two of C per pallet, to six, four, and one. The genetic algorithm found a stacking pattern that oriented the fifteen-by-twelve boxes in a way the greedy approach never considered. Corn: That one-box improvement per SKU per pallet is the difference between shipping eighteen pallets and shipping sixteen. Herman: Which at current LTL rates — you're looking at maybe a hundred fifty to two hundred dollars a pallet depending on lane and class — that's three or four hundred dollars saved on one shipment. The software pays for itself in two loads. Corn: Then there's FreightOptix on Android, which can't really run a genetic algorithm because your phone would melt. Herman: Right, the computational constraints on mobile are real. FreightOptix uses heuristics — rule-of-thumb shortcuts tuned for speed. It's not trying to find the global optimum. It's applying pre-computed pattern templates, essentially asking "does this SKU mix look like something we've seen before?" and adapting the nearest match. It's clever engineering, but the tradeoff is accuracy. On uniform loads it's fine. On mixed SKUs with odd dimensions, you'll leave pallet space on the table. Corn: There's also the data entry question. PalletStacking and FreightOptix are manual — you type in each dimension. LoadMaster lets you import a CSV, which matters when you're dealing with fifty SKUs and don't want to spend an afternoon transcribing numbers. Herman: The friction is non-trivial. I've seen warehouses where someone's using a free web calculator and manually re-entering the same twelve SKUs every week because there's no save or import function. That's not just time wasted — it's error-prone. One mistyped dimension and your entire pallet plan is wrong. Corn: The landscape Daniel's looking at is basically three points on a triangle: speed, accuracy, and convenience. Herman: The algorithm is only half the story. The real-world implications of your tool choice go much deeper than pallet count. You pick a tool, you get a number, you book the freight. But that number ripples through everything downstream. Warehouse slotting, for instance — if your pallet plan says six pallets and the warehouse has five slots in the staging area, someone's making a decision about where the sixth pallet lives. Maybe it sits in a lane, maybe it blocks something, maybe it gets double-handled. That's labor cost your calculator never mentioned. Corn: The difference between four pallets and five can bump you from one freight class to another, or from LTL to a volume quote. The tool says "here's your pallet count." The carrier says "that'll be an extra four hundred dollars because you crossed a threshold. Herman: There's also the carbon footprint angle that nobody talks about. If your consolidation tool leaves you at eighty percent utilization instead of ninety-two, you're running more trucks, burning more diesel, generating more emissions. A twelve percent improvement in pallet utilization — which is what we see in the real-world case of a mid-size e-commerce company that switched from a free online calculator to LoadMaster — that translated to eighteen thousand dollars a year in saved freight. But it also meant fewer trucks on the road. The environmental impact is a direct knock-on effect of algorithm choice, and almost nobody connects those dots. Corn: The free calculator is free in the same way that a free puppy is free. The cost just shows up somewhere else. Herman: Here's where most of those free tools really fall apart — they ignore constraints that actual carriers enforce. A standard pallet can't exceed two thousand two hundred pounds, but most free calculators don't even ask for box weight. They'll happily tell you to stack forty boxes of ceramic tile because the dimensions fit, and the carrier will reject it at the dock. Corn: Or overhang rules. Some carriers allow zero overhang. Some allow an inch. Some allow two inches but only on specific sides. If your calculator assumes flush-fit and you've got boxes hanging over the edge, your pallet count is fiction. Herman: Stackability is the other silent killer. Some boxes say right on them — "do not stack." Maybe the contents are fragile, maybe the corrugate can't handle top-load. Your calculator doesn't know that. It sees a flat surface and puts another box on it. In the real warehouse, that pallet gets restacked by hand, your nice optimized layout is gone, and you're shipping air anyway. Corn: Which is why the tool's output is really a starting point for a conversation with reality, not the final answer. You have to layer on the constraints yourself. Herman: This is where small-to-medium shippers get particularly burned. They tend to rely on carrier-provided calculators — FedEx, UPS, the big LTL carriers all have them. But those tools are black boxes. You put in dimensions, you get a quote. You don't see the algorithm, you don't see the assumptions, and you definitely don't see the alternative arrangements that might have been cheaper. Corn: They're optimized for the carrier's profit, not the shipper's efficiency. Herman: A carrier calculator assumes uniform boxes. It's built for the simplest case because that's what lets them quote fast and book the shipment. If you've got mixed SKUs — which is most real-world consolidation — it'll just treat them as separate line items and never attempt to co-mingle them on a pallet. You get a pallet count that's higher than necessary, and a quote that's higher than necessary, and the carrier gets paid for the empty space. Corn: The tool that's supposed to help you is actually helping them. Herman: Most shippers never question it because they assume the math is neutral. Math isn't neutral when you can't see the equation. Corn: What's the alternative for someone who doesn't want to drop two hundred dollars on LoadMaster but also doesn't trust the free web calculators? Herman: The dark horse here is spreadsheets. Excel with the Solver add-in, or Google Sheets with a custom Apps Script. You can write a bin-packing macro — it's not trivial, but it's also not rocket science. The Solver engine in Excel can handle a simplified version of the three-dimensional packing problem if you're willing to set up your constraints properly. Weight limits, overhang rules, stackability flags — you can encode all of that. And once it's built, it's yours. No subscription, no black box, no carrier optimizing against you. Corn: The catch being you have to build it. Herman: The learning curve is steep, no question. You're writing VBA or JavaScript, defining objective functions, setting up constraint matrices. Most warehouse managers aren't going to do that. But for a technically inclined shipper — someone like Daniel, who works in tech and automation — it's a powerful option. And there are templates floating around. People share them on logistics forums, GitHub repos with pallet optimization scripts. It's an underground ecosystem. Corn: The spreadsheet underground you mentioned earlier. Herman: It's real. And the beauty of a spreadsheet solution is transparency. You can see exactly why the algorithm placed box seventeen on layer three instead of layer two. You can override it. You can add a constraint mid-plan because you just found out the customer wants all SKU A on top for easy unpacking. Try doing that with a carrier calculator. Corn: The landscape Daniel's actually looking at isn't just three tools on a triangle. It's four quadrants — free web, paid desktop, mobile, and roll-your-own. And the right choice depends on volume, technical skill, and how much those hidden constraints actually matter for your specific freight. Herman: With all that in mind, here's what you should actually do. If you're doing a one-off calculation — you've got a single shipment, maybe three or four SKUs, you just need a ballpark — use PalletStacking. It's free, it runs in your browser, and for simple loads the greedy algorithm is perfectly adequate. You're not leaving enough money on the table to justify buying software. Corn: The "good enough" zone. One shipment, low stakes, don't overthink it. Herman: But if you're consolidating mixed SKUs regularly — weekly or daily — invest in LoadMaster. Two hundred dollars, one-time, and that genetic algorithm will find arrangements the free tools never touch. The mid-size e-commerce company I mentioned earlier — twelve percent fewer pallets, eighteen thousand dollars saved in a year. The software paid for itself in the first week. Corn: That's the thing about the paid tool — it's not a cost, it's an arbitrage. You're paying two hundred dollars to stop overpaying your carrier by thousands. Herman: And then there's FreightOptix on Android for the warehouse floor. Five bucks a month. If you need to do a quick check while you're standing next to a pallet — "can I fit six more of these on here or am I wasting my time" — it's fine. Not optimal, but adequate. The heuristic approach means you're leaving some space on the table, but for on-the-go estimates, speed beats precision. Corn: The decision tree is basically: how often are you shipping, and how weird are your box dimensions? The weirder the mix, the more the algorithm matters. Herman: Here's the pro tip that applies regardless of which tool you pick. Always validate the output against your actual carrier's pallet rules. I cannot stress this enough. Your calculator says eight pallets. Does your carrier allow overhang? On which sides? What's the weight limit per pallet — is it two thousand pounds or twenty-five hundred? What's their policy on mixed-SKU pallets — do they charge a reclassification fee? None of these questions appear in the calculator interface, and any one of them can invalidate your beautifully optimized plan. Corn: The algorithm says "this fits." The carrier says "not on my truck. Herman: The carrier wins that argument every time. So before you book, pull up your carrier's rules sheet. If you're using FedEx Freight or UPS Freight, they publish detailed pallet specifications — maximum dimensions, overhang tolerances, weight caps by class. Compare them line by line against what your tool assumed. If the tool assumed flush-fit and your carrier allows two inches of overhang, you might actually fit more than the calculator said. If the tool ignored weight and you're shipping cast iron cookware, you might need to split pallets the calculator thought were fine. Corn: The tool gives you a starting point, not the final answer. You're the human in the loop, applying the constraints the software doesn't know about. Herman: Which brings me to the call to action, and I mean this. Download one of these tools today. PalletStacking is free — there's no barrier. Grab three of your actual SKUs, the ones you ship all the time, and run the numbers. Then run the same numbers through whatever you're currently using — your carrier's calculator, your gut instinct, the back of a napkin. Compare the results. Corn: If the numbers are different, that difference is money you've been leaving on the table. Herman: And if you're currently using a FedEx or UPS pallet calculator, I can almost guarantee you're overpaying. Those tools assume uniform boxes. If you've got mixed SKUs, they're not even attempting to optimize — they're just summing volumes and dividing. The gap between that and what a proper bin-packing algorithm produces is often ten to fifteen percent more pallets. That's not a rounding error. That's a recurring surcharge you're paying because you trusted the wrong calculator. Corn: The most expensive tool is the one that's free and wrong. Herman: The cheapest tool is the one that tells you the truth about your own freight. So pick one, test it, and see what you've been missing. The savings might surprise you — and if they don't, you've lost twenty minutes. If they do, you've just found a permanent cost reduction that compounds with every shipment. Corn: As you start optimizing, here's what's coming next. Herman: The thing I keep thinking about is whether any of this matters in two or three years. Because the real shift coming isn't a better greedy algorithm or a faster genetic solver. It's reinforcement learning models that don't just pack pallets — they learn from every shipment they've ever seen. Corn: Instead of "here's a packing algorithm, good luck," it's "here's an AI that has watched ten million pallets get loaded and knows which arrangements actually survive transit. Herman: That's the promise. A reinforcement learning agent doesn't need to be told the rules about overhang or stackability. It learns them from outcomes. Pallet arrived damaged? That arrangement gets penalized. Carrier rejected the load? Negative reward signal. Over time, the model internalizes constraints that no human programmer would think to hard-code. Corn: Which makes today's tools feel a bit like the GPS unit you bought for your car in two thousand five — functional, useful, and about to be absorbed into something much bigger. Herman: The something bigger isn't just AI packing. It's integration. Real-time freight pricing APIs are becoming standard — every major carrier and half the digital freight brokers have them now. The logical endpoint is a tool that doesn't just tell you how many pallets you need, but simultaneously queries five carriers, finds the cheapest option for that exact pallet configuration, and books it. Corn: The consolidation calculator becomes the front end of a booking engine. You're not optimizing pallets in a vacuum. You're optimizing pallets against live rates, and the algorithm is making tradeoffs — maybe an extra pallet is cheaper if it lets you use a different carrier on a different lane. Herman: And that's where the AI piece gets interesting. A reinforcement learning model could optimize across both dimensions simultaneously — pallet configuration and carrier selection — because the objective function isn't "minimize pallets." It's "minimize total landed cost." Those are different things. Corn: The algorithm that saves you the most money might be the one that deliberately leaves a pallet half-empty because it bumps you into a lower freight class. Herman: No human planner is running that math in their head. But an AI trained on a few years of shipping data? It spots those patterns instantly. "Hey, if you leave four inches of headspace on pallet three, you drop from class one twenty-five to class ninety-two, and that saves you six hundred dollars on this lane with this carrier on a Tuesday." That's the kind of optimization that makes today's tools look like arithmetic. Corn: The question isn't really whether these tools become obsolete. It's whether they become features inside larger platforms. PalletStacking doesn't die — it becomes the packing module inside a freight booking platform. Herman: We're already seeing the early moves. Some of the digital freight brokers are building basic consolidation calculators into their quoting tools. They're not good yet — mostly uniform-box assumptions, not real mixed-SKU optimization — but the direction is clear. The company that nails the integration first wins. Corn: Which means the spreadsheet underground might actually have the last laugh. If you've built your own solver in Excel, you can bolt it onto any API you want. You're not waiting for a vendor to integrate. Herman: The spreadsheet people are always the last ones laughing. They're also the ones who never get invited to parties, but that's a separate issue. Corn: A noble sacrifice. Herman: Now: Hilbert's daily fun fact. Hilbert: In the nineteen thirties, on the island of São Tomé, Portuguese colonial administrators recorded that plantation workers consumed a fermented beverage made from the sap of the oil palm tree, known as "vinho de palma," which was reportedly drunk within six hours of tapping because it soured so quickly in the tropical heat that it became undrinkable by the following morning. Corn: That's a tighter shelf life than my leaf medicine tinctures. Herman: I have so many questions about the fermentation process and I'm going to choose not to ask any of them. Corn: This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop, and thanks to Daniel for the question. If you're enjoying the show, leave us a review wherever you listen — it helps other people find us. We're back next week.