# Descartes _The network you can't replicate._ --- ## The bankruptcy In 2004, Descartes Systems Group was nearly dead. The company had been public since 1998, had burned through capital chasing a dot-com-era vision of supply chain marketplaces, and was running out of cash. Revenue was collapsing. The share price had cratered from over $40 to below $2. Arthur Mesher, a former Gartner analyst, took over and did something unusual. He fired the entire sales team, turned the engineers into customer-facing roles, and cut 35% of the workforce. What remained was the one asset that still had value: a messaging network that connected logistics companies electronically. That network, linking carriers and shippers and customs brokers, processed transactions that its participants couldn't easily route elsewhere. Mesher's insight was that the network itself was the product. Not the software that sat on top of it. Not the consulting services wrapped around it. The connections. Twenty years of compounding followed from that insight. --- ## The network Descartes calls it the Global Logistics Network. Today it connects roughly 200,000 parties across 160 countries, processing about 20 billion messages a year. Shippers file customs declarations through it. Carriers transmit shipment data. Freight forwarders coordinate multimodal moves. Customs authorities receive regulatory filings. The economics are straightforward. Every participant on the network makes it more valuable for every other participant. A shipper joining the network gains access to 200,000 potential trading partners and dozens of government customs systems, without needing to build individual connections to each. A carrier joining gains visibility to the shippers already using the platform. A customs authority accepting electronic filings through the network reduces its own processing costs and increases compliance rates. This is a genuine network effect, not the diluted version where "more users means more data." Each connection is a live integration between two parties who transact regularly. Replicating the software is possible. Replicating 200,000 interconnected trading relationships built over two decades is not. The network generates 93% of Descartes' revenue through subscriptions and transaction fees, recurring payments from participants who depend on the connections to move goods across borders every day. --- ## Regulation as a tailwind Global trade is not getting simpler. Every new sanctions regime, every customs modernisation programme, every forced-labour import ban, every change in tariff schedules creates compliance obligations that logistics companies must meet. Miss a denied-party screening and you face criminal penalties. File a customs declaration late and your shipment sits at the border. Get the tariff classification wrong and you pay duties you shouldn't owe, or worse, underpay and trigger an audit. Descartes connects directly to customs authorities in dozens of countries, including US Customs and Border Protection's Automated Commercial Environment and the Canada Border Services Agency's advance information system. When a government changes a filing requirement, Descartes updates the connection once and every participant on the network benefits. A logistics company trying to manage this on its own would need to track regulatory changes across every jurisdiction it operates in, maintain direct electronic links to each customs authority, and update its systems every time something changes. The more complex global trade regulation becomes, the more valuable an embedded compliance network is. This is the opposite of most technology markets, where simplification threatens the incumbent. Here, complexity is the moat's reinforcement. A survey in 2025 found that 42% of freight forwarders and customs brokers cited tariffs as the single biggest change affecting their industry. Every one of those tariff changes generates transaction volume for Descartes. --- ## Acquisitions that feed the network Descartes has completed roughly 50 acquisitions since the mid-2000s, deploying about $1.4 billion at an average deal size of around $40 million. The cadence is steady: two to five deals a year, rarely more, never a bet-the-company transaction. The logic is different from [[Constellation]], which acquires vertical market software businesses across dozens of industries and runs them as autonomous units. Descartes acquires within a single domain. Every target is a logistics technology company whose customers and capabilities can be folded into the Global Logistics Network. A transportation management system brings carrier connections. A customs declaration provider brings regulatory links. An inventory management platform brings e-commerce shippers. Each acquisition extends the network's reach, which increases the value for every existing participant. 3GTMS, acquired in early 2025 for $115 million, plugged Descartes into North American less-than-truckload carrier networks. Sellercloud, acquired for $110 million in 2024, brought inventory and order management for e-commerce retailers. MyCarrierPortal, acquired for $24 million, added carrier risk monitoring. Each deal brought participants who immediately connected to the existing 200,000. The flywheel is visible in the financials. Recurring revenue funds the acquisitions. The acquisitions extend the network. The extended network generates more recurring revenue. Over the past five years, revenue has grown from $326 million to $651 million and adjusted EBITDA has expanded from $110 million to $285 million. Organic services growth runs at mid-single digits. Acquisitions contribute another four to seven percentage points. The combination has compounded market capitalisation at roughly 16% annually since 2002. --- ## The switching cost underneath Network effects get the attention, but [[Switching costs]] do the quieter work. A freight forwarder running Descartes for customs declarations has configured its systems around Descartes' filing formats, trained its staff on Descartes' workflows, and built integrations between Descartes and its own transport management and accounting systems. The connections to customs authorities, to carriers, to shipper counterparties all run through the network. Unplugging means rebuilding every one of those connections individually. Customer retention runs above 95%. Contracts are typically multi-year. Services revenue, at 93% of the total, renews with the reliability of a utility. The two moat sources reinforce each other. The network effect makes joining attractive. The switching costs make leaving painful. A competitor would need to offer both a better product and a comparable network. The product gap can be closed with engineering. The network gap cannot. --- ## What Descartes is not Organic growth runs in the mid-single digits, slower than [[Jack Henry]] and well below [[Veeva]] in its expansion phase. The stock has historically traded at premium multiples, with the trailing P/E in the high 40s. The product manages customs filings, freight messaging, and route optimisation for logistics companies and customs brokers out of Waterloo, Ontario. None of this makes headlines. WiseTech Global, an Australian competitor building CargoWise as a single integrated platform through organic development, has pursued the same logistics vertical from the opposite direction. WiseTech writes the code. Descartes buys the connections. Both approaches work, which tells you something about the size of the opportunity and the difficulty of dominating it outright. But Descartes has compounded for twenty years since its near-death experience, retains 95% of its customers annually, generates 77% gross margins and 44% EBITDA margins on $650 million of revenue, carries no debt, and sits on a network that took two decades and 50 acquisitions to assemble. Edward Ryan, who joined through an acquisition in 2000 and led the company as CEO from 2013 to 2024, has described the strategy simply: grow and operate the most complete logistics technology platform. The acquisitions serve that goal. The network enforces it. --- ## What this teaches [[Jack Henry]], [[Tyler Technologies]], and [[Veeva]] all built [[Verticals|vertical]] dominance through depth. Their primary moat is switching costs: customers embed the software so deeply that leaving becomes unthinkable. Descartes adds a second layer. The network effect means that each customer's presence makes the platform more valuable for every other customer. This is closer to the [[Complements]] logic in ASSA ABLOY, where each acquired component raises the value of the whole system, than to the autonomous portfolio model in Constellation. Switching costs protect the installed base. Network effects grow it, because each new participant attracts the next. The two together produce a flywheel where retention funds acquisition, acquisition extends the network, and the extended network deepens retention. Global trade is getting more complex: regulations multiply, supply chains fragment, compliance obligations increase. Every one of those trends makes the network more essential to the companies that depend on it, and harder for any competitor to reproduce. The company that nearly went bankrupt in 2004 survived because one person recognised that the network, not the software, was the asset worth saving. Everything since has been the slow, disciplined work of making that network larger, stickier, and more essential. ---