The Descartes Systems Group Inc. (NASDAQ: DSGX; TSX: DSG) — The House That Sells Umbrellas in a Hurricane, Marked Down Because It Rained

Independent equity research note. Report date: 2026-07-04. Fiscal year ends January 31; all figures USD unless noted.

Note: The analysis below is deliberately position-free — it contains no buy/sell recommendation and no price target, and discusses valuation only as embedded expectations and scenarios. The sole exception is the clearly-labeled Author’s Take block immediately below, which is the author’s own subjective opinion.

⚡ Author’s Take

This is the author’s own independent opinion and general information only — not investment advice. The analysis that follows takes no position and carries no price target.

Verdict: HOLD / accumulate-on-weakness. A genuine wide-moat compounder — arguably the single highest-quality business in freight-tech — has quietly de-rated ~40% from a bubble multiple to its cheapest valuation in a decade, even as its organic growth is accelerating and margins are hitting record highs. This is not a screaming bargain; it is a great business finally trading at a merely-full (rather than absurd) price. Constructive accumulation zone ~$60–72 (≈17–20x forward adjusted EBITDA / ≈30–33x forward earnings); I’d get materially more interested sub-$62 (the Feb-2026 low), and I would trim, not chase, back above ~$95–100. Conviction: medium-high on the business, medium on the entry (still ~20x EBITDA).

Descartes is misread by the tape as a cyclical “freight-recession loser.” It is really a neutral network + trade-data utility that gets paid to sell certainty in chaos: its fastest-growing, highest-margin franchise — Global Trade Intelligence (tariff/duty content, sanctioned-party screening, foreign-trade-zones, Datamyne) — feeds directly on exactly the tariff whiplash and geopolitical disorder that spooked the stock. The market conflated a real but cyclical shipment-volume headwind (US trucking volumes −4%, ocean choked by the Iran war) with a structural problem, and missed that organic services growth actually re-accelerated to ~9.5% through the downturn while adjusted-EBITDA margin printed a record 46%. The framing is quality-compounder-at-a-fair-price, sharpened by a contrarian/abandoned-momentum setup: the factor tape shows DSGX at Momentum −0.31, ~40% off its peak, one-year total return −28% — a long-term winner (lifetime +15.8%/yr) sitting in a −57%-scale drawdown. It is a HOLD and not a table-pounding BUY for one reason — price. At ~20x EV/EBITDA and ~33x forward earnings on a business whose reported ROIC (~11%) is flattered-down by goodwill but is still not a 30%-cash-return machine, and whose insiders own under 1% (a net-selling CEO), you are paying a full price for durability, not exploiting a mispricing of the assets.

Catchy tag: “The house that sells umbrellas in a hurricane — marked down because it rained.” Bull trigger (flips me more bullish): organic services growth holds ≥9–10% into a recovering freight cycle (volume tailwind stacks on the compliance tailwind) while the NCIB keeps shrinking the count — or the stock revisits sub-$62. Bear trigger (flips me cautious): organic growth decelerates back toward mid-single-digits as tariff volatility normalizes (a GTI air-pocket) and a large out-of-lane acquisition or a goodwill write-down signals the buy-and-build runway is thinning.

📈 Stock Price Action — Five-Year Event Map

Over five years DSGX has round-tripped: from ~$69 (mid-2021) to a February-2025 all-time high of $122.50, then a grinding ~40% de-rate to a $62.82 low (Feb-2026), recovering to ~$72.80 today. The stock sits ~41% below its peak and still below its 200-day EMA (~$79) — a broken momentum name whose fundamentals never broke. 52-week range ~$63–$108. The five-year story is almost entirely multiple, not earnings: revenue and EPS compounded straight through while the EV/EBITDA multiple roughly halved (≈37x → ≈20x).

Cycle narrative. (1–3) The 2021–2024 action was a macro round-trip: DSGX fell in the 2022 rate shock purely on multiple compression (revenue grew ~22% that fiscal year), then re-rated as the durability of its growth reasserted itself. (4) The February-2025 ATH of $122.50 marked peak enthusiasm — a ~37x EV/EBITDA, ~69x P/E multiple pricing the tariff-complexity tailwind and AI optionality to perfection. (5–6) The 2025→2026 decline is the mirror image: as the freight cycle rolled over (US trucking volumes −4%, ocean shipping choked by the Iran war/Strait of Hormuz) and tariff uncertainty caused importers to freeze shipping decisions, the market re-cast a rich-but-durable compounder as a cyclical loser and compressed the multiple by nearly half, bottoming at $62.82 on 2026-02-23 amid a broad growth-software selloff. (7) The recovery to ~$73 tracks the realization that organic services growth actually accelerated through the downturn (three consecutive quarters, to ~9.5% in Q1 FY27) and margins hit records — with management initiating a Normal Course Issuer Bid into the weakness. Every price move is a Fact; the attributed driver is Interpretation, cross-referenced to earnings dates, the price history, and the Q1 FY27 call.

1. Executive Summary

The Descartes Systems Group is a Waterloo, Ontario cloud software company that operates the Global Logistics Network (GLN) — a neutral, multi-sided electronic network connecting hundreds of thousands of shippers, carriers, freight forwarders, customs brokers, and 3PLs — plus a federation of modular logistics and trade-compliance applications sold on a Software-as-a-Service and transactional basis. Founded in 1981, it has compounded revenue from $326M (FY20) to $729M (FY26) at a ~14% CAGR while expanding operating margin from 17% to 30% and adjusted EBITDA margin to a record ~46%. It is ~93% recurring, generates ~37%-of-revenue free cash flow, carries no debt and ~$377M of cash, and reinvests roughly 90% of that cash flow into disciplined tuck-in acquisitions of niche logistics-software leaders. It is, in short, one of the cleanest compounding machines in software.

The central tension is not business quality — that is largely settled — but cyclicality vs. durability, and price. A meaningful (undisclosed) slice of revenue is transaction-based and tied to shipment volumes, so the 2024–2026 freight recession, layered onto tariff-driven shipment paralysis and the Iran-war disruption of ocean freight, created a genuine volume headwind. Yet Descartes’ revenue is dominated by recurring, contracted, compliance-critical services, and its highest-margin franchise (Global Trade Intelligence) is counter-cyclical — it thrives precisely when tariffs, sanctions, and trade rules are in flux. The result: organic services growth accelerated to ~9.5% through the worst of the cycle. Meanwhile the market compressed the EV/EBITDA multiple from ~37x to ~20x, leaving the stock at the ~1.5th percentile of its own decade-long P/E history and the ~6.9th percentile on a composite basis — near the cheapest it has ever been, and a discount to factor-similar quality-software compounders (Manhattan ~35x, Tyler ~38x EBITDA).

What we like: a durable (if narrow) moat rooted in customs/compliance switching costs and proprietary regulatory-grade trade data; 77% gross margins with genuine operating leverage; a fortress balance sheet; a repeatable, self-funding buy-and-build with zero goodwill impairments in eight years; and a conservative “baseline calibration” operating model that protects margins in downturns. What gives us pause: reported returns on capital only modestly exceed the cost of capital (value is created by improving full-priced acquisitions, not buying them cheap); weak insider alignment (insiders own <1%, and the CEO is a net seller); a real, undisclosed exposure to shipment volumes; and a valuation that — while cheap versus history — is still a full ~20x EBITDA in absolute terms. This memo takes no position; the labeled Author’s Take above does.

2. Business Overview

What Descartes does. Descartes sells cloud-based software and data that help companies move goods across borders and across town more efficiently, compliantly, and visibly. Its offering is best understood as two intertwined things: (1) the Global Logistics Network (GLN) — a neutral B2B messaging and connectivity backbone over which trading partners exchange shipment bookings, status messages, customs filings, and documents; and (2) a portfolio of ~300–400 modular applications layered on that network, spanning six broad pillars:

• Global Trade Intelligence (GTI) / customs & compliance — real-time global tariff and duty content; sanctioned/denied-party screening; foreign-trade-zone (FTZ) management; customs declaration and filing (e.g., NetCHB for US entries); product classification/HS-code determination; and Datamyne, a global trade-data research tool. This is the highest-margin, most defensible, and (recently) fastest-growing pillar.
• Transportation management & real-time visibility — transportation management systems (TMS) for shippers/brokers; MacroPoint real-time freight visibility (tracking loads across carriers and modes); rating, load-matching, and freight audit/payment.
• Routing, mobile & fleet performance — route optimization and scheduling, mobile proof-of-delivery, telematics, and (via the 2026 Idelic acquisition) AI-driven driver-safety analytics.
• E-commerce, shipping & fulfillment — parcel shipping, e-commerce customs entries, warehouse/inventory management (Sellercloud, Finale), and delivery/appointment scheduling.
• Broker & forwarder enterprise systems — back-office systems for customs brokers and freight forwarders, including air-cargo systems.
• B2B connectivity / the GLN itself — the messaging, EDI, and network services that stitch the above together.

How it makes money. Revenue is ~93% “services” (FY26: $677M of $729M), which the company defines as ongoing transactional and/or subscription fees plus maintenance. The remaining ~7% is professional services/implementation/hardware, and <1% is perpetual license (a legacy rounding error). Within “services,” a portion is fixed subscription/maintenance and a portion is transaction-based — priced per shipment, per filing, per tracked load, or per message across the GLN. [FACT] Descartes does not disaggregate the transaction vs. subscription split; it concedes only that “some of our revenues…are tied to the volume of shipments.” [OPEN QUESTION] This is the single most important undisclosed number in the model, because it sizes the freight-cyclical slice.

Customers & geography. The customer base is globally diverse — carriers (air/ocean/truck), 3PLs, freight forwarders, customs brokers, retailers, manufacturers, distributors, and government agencies — with no disclosed customer concentration. Geographically (FY26): United States 68%, EMEA 23%, Canada 6%, Asia-Pacific 3% of revenue; ~72% of revenue is USD-denominated, though a material share of costs is CAD/EUR/GBP. [FACT] Deferred revenue is ~$119M and grew ~13% YoY, consistent with a healthy, expanding recurring base.

Recurring-revenue character. Management expects to lose only ~5–7% of aggregate annualized recurring services revenue per year absent new business — implying gross retention of ~93–95%. [FACT/INTERPRETATION] That is a very sticky base for a company with a diverse SMB-to-enterprise mix, and it is the foundation of the “baseline calibration” model discussed

Revenue by line (FY26 vs FY25 vs FY24, $M):

[FACT] The mix is the quality signal: the ~93% services line carries an ~80% gross margin and grows with both new bookings and the M&A cadence, while the low-margin professional-services line is shrinking in relative terms (a healthy sign — less bespoke implementation, more repeatable product). License is a vestigial <1%. This is a product-and-network business, not a services-and-integration one.

Why the “network/data utility” framing matters. Management is emphatic — and, on the evidence, correct — that Descartes is better understood as a neutral connectivity-and-content utility than as an application-software vendor. The GLN is plumbing: it moves billions of transactions a year among parties who would otherwise need to build and maintain a combinatorial web of point-to-point integrations. On top of that plumbing, the most valuable modules (GTI, customs, sanctioned-party) are essentially proprietary regulatory content sold by subscription — closer in economic character to a Bloomberg-terminal or credit-bureau data feed than to a seat-license SaaS app. That is exactly why the FactorsToday risk model loads Descartes partly on a “Financial Data Titans” basket rather than pure application software: the market’s own statistical fingerprinting sees the data-utility character. This framing is not spin — it is the reason gross margins are 77%, retention is 93–95%, and the business grew organically through a freight depression.

Verdict: A high-quality, asset-light, ~93%-recurring software-and-data business with a genuinely differentiated core (the neutral network + regulatory-grade compliance content) and a crowded periphery (visibility/TMS). The revenue mix is dominated by durable, contracted, high-margin services, with a real but unquantified cyclical transaction tail.

3. Industry Dynamics

Descartes straddles two very different software profit pools, and its quality is largely a function of being correctly weighted toward the better one.

Pool A — Global Trade Management (GTM) / trade-compliance: small, sticky, capital-scarce, counter-cyclical. Third-party sizings cluster around $1.3–2.5B in 2026, growing ~8–12% per year (e.g., Fortune Business Insights ~$1.37B→$2.63B by 2034 at ~8.5%; The Business Research Company ~$1.31B→$1.44B at ~10%). [FACT] Trade-compliance is the largest sub-segment (~31% share), and cloud is ~63% of it. This pool is content-heavy, regulator-facing, liability-bearing, and oligopolistic — the ideal structure. Getting a customs filing or a sanctioned-party screen wrong carries fines and shipment stoppages, so buyers pay for accuracy and certification rather than shopping on price. Descartes’ GTI/customs franchise sits squarely here.

Pool B — Transportation management & real-time visibility: large, crowded, converging, price-compressing. MarketsandMarkets sizes TMS at ~$18.5B (2025) → ~$37B (2030), ~15% CAGR. [FACT] It is a bigger, faster-growing pool but a worse one — a crowded knife-fight among enterprise incumbents (Manhattan, Oracle, SAP, Blue Yonder) and VC-subsidized visibility pure-plays (project44, FourKites, Tive). Differentiation is thin and pricing is under pressure. Descartes competes here via MacroPoint, its TMS, and routing — as a share-taker leveraging its network and balance sheet, not as a moat-holder.

Cyclicality. The end market — freight — is brutally cyclical; 2022–2024 was among the worst freight recessions on record (a theme well documented across the freight-brokerage and forwarding cycle), and 2026 is a slow, supply-driven recovery complicated by tariff/import headwinds and the Iran-war ocean disruption. But two structural features insulate Descartes: (1) ~93% recurring revenue, so even the transaction tail resets off a contracted base; and (2) compliance complexity is counter-cyclical demand — the more chaotic and rule-bound global trade becomes, the more customers lean on GTI. Descartes grew organically ~7–9% straight through the 2022–24 trough. [FACT/INTERPRETATION]

Capital-cycle lens (Marathon). The tell-tale sign of an over-capitalized pool is capital flooding in chasing high returns. That is unmistakably happening in the visibility layer: project44 has raised ~$912M (backed by Goldman, a prior ~$2.7B valuation) on ~$210M of revenue; FourKites reached ~$114M revenue at a prior ~$1B valuation; and WiseTech Global spent ~$2.1B EV to acquire e2open in August 2025. [FACT] By the capital-cycle framework, that portends mean-reverting returns — in the visibility layer. The customs/compliance + neutral-network layer where Descartes concentrates is capital-scarce, protected by content moats and regulatory friction that deter entrants. Descartes sits on the right side of the capital cycle; the visibility pure-plays sit on the wrong side.

Secular drivers — all tailwinds: rising tariff/sanctions complexity (the 2025–26 tariff regime is a catalyst for GTI); e-commerce customs formalization (the elimination of the Type-86 de minimis exemption means every low-value parcel now needs a formal entry — more filings for Descartes); nearshoring/supply-chain reconfiguration; and AI’s appetite for clean, structured, liability-grade trade data.

Verdict: a GOOD industry where Descartes concentrates (compliance/GTM + neutral network), and a MEDIOCRE/crowded one at its periphery (visibility/TMS). On balance, favorably weighted — the profit is in the sticky, capital-scarce pool, and the crowded pool is a source of incremental share, not of the moat.

4. Competitive Position

The moat, named. Using the Greenwald (“Competition Demystified”) taxonomy, Descartes’ aggregate advantage is real but narrow, and rests on four legs of unequal strength:

1. Switching costs / customer captivity — the strongest leg. Customs filing (NetCHB), sanctioned-party screening, duty classification, and FTZ management are mission-critical, audited, regulator-facing workflows embedded in a customer’s daily operations. Ripping them out risks fines, shipment stoppages, and compliance liability. This shows up directly in the financials: 77% gross margins, ~93% recurring revenue, ~93–95% gross retention, and pricing power (GTI price is “largely the same” — growth comes from customers buying access to more countries/commodities, not from discounting). [FACT/INTERPRETATION]
2. Proprietary data / intangibles — strong on compliance content. The GTI global tariff/duty/sanctioned-party database, Datamyne trade data, and Idelic’s ~40-billion-mile driver-safety dataset are regulatory-grade, liability-bearing, and effectively un-scrapable. Management’s argument against AI-replication is credible: the data must be collected from every jurisdiction, normalized to a common format (“make Belgium look like Japan”), delivered in each customer’s system format, and — critically — stood behind when a government audits a filing. “I built my own AI tool and it was wrong” is not a defense a regulator accepts. This leg is weaker on visibility data, where rivals hold comparable track records.
3. Network effects on the GLN — real but partial. The GLN is a genuine multi-sided network: FedEx, UPS, DHL, Amazon, and Uber Freight are customers, not competitors, because maintaining hundreds of thousands of point-to-point connections is uneconomic for any single player, and a neutral intermediary is cheaper for everyone. [FACT — management framing] But it is not winner-take-all; much of the “network” value is really point-to-point integration switching cost, and WiseTech and EDI rivals coexist. Moderate, not dominant.
4. Economies of scale — moderate. Shared network and content amortize across 100,000s of connections, but Descartes is not the scale leader in most single pillars.

Financial fingerprints of the aggregate moat: 77% gross margin; ~30% operating / ~46% adjusted-EBITDA margin; ~93% recurring; net cash; negative working capital; and ~7–9% organic growth through a freight depression. A “moat” claim only counts if a financial outcome would deteriorate without it — here, retention, gross margin, and pricing power would all visibly erode if the switching costs and data advantage were illusory. They have not. [INTERPRETATION]

Head-to-head: Descartes vs. WiseTech Global (the sharpest comparable). WiseTech’s CargoWise is the closest analog and, by the raw numbers, the better core business: ~$700M revenue growing ~21–25% at ~53% group EBITDA margin, with a deeper single-platform network that runs a forwarder’s entire business rather than modules. [FACT] Descartes counters with a more diversified, lower-risk, capital-allocation-driven model: a federation of niche #1–#3 positions, disciplined ~$150M/year tuck-in M&A at sane prices, a permanently net-cash balance sheet, less single-platform concentration, and better positioning in counter-cyclical compliance. WiseTech is now digesting a large, debt-funded e2open acquisition (integration and margin-dilution risk). On pure quality, WiseTech wins; on risk-adjusted durability, the gap narrows sharply. [INTERPRETATION]

The competitive map, by pillar (named):

• Trade/customs/compliance: SAP GTS, Thomson Reuters ONESOURCE Global Trade, QAD Precision (ex-MIC), Aptean, e2open/WiseTech; trade data from Datamyne/ImportGenius/Kpler.
• Visibility/TMS: project44, FourKites, Tive, Transporeon (Trimble), Blue Yonder, Oracle TM, SAP TM, MercuryGate, Manhattan Associates.
• Forwarder/broker systems: WiseTech CargoWise (dominant), Magaya.
• Routing/fleet: Trimble, Samsara, Verizon Connect, Workwave, Omnitracs/Solera.

The competitive economics, quantified. The clearest evidence for “durable core, contested periphery” is in the rivals’ P&Ls. In the core, the incumbents are profitable, slow-moving, and complementary: SAP GTS and Thomson Reuters ONESOURCE are modules inside broader ERP/tax suites (not focused attackers), and the trade-data players (ImportGenius, Kpler) sell adjacent data rather than the compliance-workflow-plus-liability bundle. In the periphery, the attackers are large but unprofitable and capital-hungry: project44 (~$210M revenue, prior ~$2.7B valuation, VC-funded and loss-making) and FourKites (~$114M revenue) are burning capital to buy share in real-time visibility — exactly the profile that mean-reverts. Descartes’ MacroPoint competes there from a position of profitability and network scale (the 87%→93% track-rate lead is a network-density advantage rivals cannot cheaply replicate), which is why Descartes can take share in visibility even without owning a moat there. Manhattan Associates (~$1.07B revenue, cloud +22%, a Gartner TMS Leader) is the strongest enterprise-TMS overlap but plays up-market and is not a customs/compliance competitor. [FACT/INTERPRETATION] The net picture: nobody attacks Descartes’ profitable core head-on, and its periphery losses-leaders are structurally disadvantaged on unit economics.

The AI question — opportunity for the core, threat to the periphery.

• Bull (widens moat): GTI compliance content and GLN transaction data are exactly the clean, liability-grade data AI needs and cannot self-generate — Descartes becomes the “source of truth” feeding customers’ AI. Agentic monetization is already live: MacroPoint’s AI driver-calling agents lifted the shipment track rate from 87% to 93% in six months, and each incremental point is “a lot of $2 shipments.” Management is building an AI-agent layer on the GLN that orchestrates, permissions, audits, and bills for agent actions — a potential new high-margin revenue stream (“I may be charging $0.25 for giving this out…times a thousand more things”).
• Bear (commoditizes): frontier LLMs can increasingly classify HS codes and parse public tariff schedules, eroding the duty-database premium; API-first/hyperscaler connectivity could attack the GLN’s EDI switching costs; and AI accelerates convergence in the already-commoditizing visibility layer. [OPEN QUESTION] The swing factor is whether certification/liability keeps GTI’s pricing power once public schedules are trivially parseable.

Verdict: a DURABLE but NARROW moat. Verified where it matters — customs/compliance switching costs, proprietary regulatory content, and a real (if partial) neutral network — and over-claimed on the commoditizing visibility/TMS pillars. The 77%-GM / 46%-EBITDA / 93%-recurring / net-cash / through-cycle-organic-growth profile is genuine proof the aggregate moat is real, even if it is not the wide, single-platform fortress WiseTech is building.

5. Growth History and Forward Opportunities

The historical algorithm. Descartes compounds through a repeatable stack: ~9–10% organic services growth + ~4–5% acquired growth ≈ ~14% revenue CAGR, plus steady margin expansion ≈ ~20%+ EPS/FCF-per-share growth. The record:

[FACT] EPS nearly quadrupled over six years (~27% CAGR) — the product of revenue growth, ~700bps of operating-margin expansion, and only mild dilution. Growth is high quality: it is recurring, it is cash-generative, and it has been resilient to a severe freight downturn.

Organic growth is accelerating, not fading. This is the crux of the variant view. Management disclosed organic services growth of ~8% in Q4 FY26 rising to ~9.5% in Q1 FY27 — the third consecutive quarter of acceleration — achieved in a down freight market. [FACT — Q1 FY27 call] The drivers: (1) GTI — the standout, as tariff/duty volatility, sanctioned-party complexity, FTZ adoption (importers deferring tariffs), and Datamyne research all surged; (2) e-commerce customs (NetCHB) — Descartes gained share when the Type-86 de minimis elimination forced formal entries and competitors’ non-network filing systems “fell apart”; (3) fleet/routing — demand rises with fuel costs and driver shortages; and (4) MacroPoint — the 87%→93% track-rate lift is pure high-incremental-margin revenue.

Decomposing the ~14% CAGR. The growth is a deliberate two-engine machine. Organic services growth has run ~7–10% (the low end in the freight trough, ~9.5% now), and acquisitions add roughly ~4–5 points on top. The M&A contribution is steady by design: at ~$150M/year of deals against a ~$700–730M revenue base, tuck-ins acquiring businesses at (inferred) high-single-digit to low-double-digit revenue multiples add on the order of $60–120M of acquired revenue over a rolling two-year window — i.e., the ~4–5 point contribution is arithmetically repeatable so long as the pipeline and balance sheet hold. Critically, the organic and acquired engines reinforce each other: acquired products are cross-sold into the GLN installed base (raising their organic growth post-deal), and the network’s reach makes Descartes a preferred acquirer (targets want distribution). This is the Constellation/Roper flywheel, applied to a single vertical. [INTERPRETATION]

A note on retention and expansion. Descartes discloses gross attrition (~5–7%/year of recurring services revenue) but not net revenue retention (NRR) or an expansion rate. [OPEN QUESTION] The ~9.5% organic services growth on a ~93–95%-gross-retention base implies meaningful net expansion (existing customers buying more countries/commodities of GTI content, more tracked loads, more modules) plus new-logo wins — but we cannot cleanly separate the two from disclosure. That the GLN track rate rose 87%→93% and GTI customers “buy access to more of the database” are qualitative confirmations of the expansion motion.

Forward opportunities. (a) Freight-cycle recovery — if volumes normalize, the transaction tail turns from headwind to tailwind, stacking on the compliance tailwind. (b) GTI/agent monetization — the AI-agent layer and expanded GLN “menu” of billable services are a genuine new revenue vector (management’s illustrative “$0.25 per call × a thousand more callable services” is optionality, not yet in the numbers). © M&A runway — global logistics software remains fragmented; the buy-and-build has years of targets, and private valuations are reportedly compressing, improving entry prices. (d) Cross-sell — the diverse installed base is under-penetrated across the 300+ product catalog. (e) De minimis / regulatory formalization — every structural increase in trade-compliance burden is Descartes revenue; the elimination of Type-86 de minimis is a durable, structural (not cyclical) uplift to customs-filing volumes.

Verdict: high-quality growth, and — contrary to the tape — accelerating. The main risk is not that growth is low quality but that a normalization of tariff chaos could create a GTI air-pocket.

6. Financial Quality

Descartes is a financial-quality standout: a “rule-of-~55” software business (organic services growth ~9–10% + adjusted-EBITDA margin ~46%) with pristine cash conversion, negative working capital, net cash, and unusually modest dilution for a serial acquirer.

Revenue & margins. Revenue compounded from $326M (FY20) to $729M (FY26) (~14.3% CAGR); Q1 FY27 was a $193.6M record (+15%). Gross margin expanded 73.7% → 77.1% (78% in Q1 FY27); operating margin nearly doubled 17.2% → 30.2%; and adjusted EBITDA margin set a record 46% in Q1 FY27, above the company’s own 40–45% target. Services gross margin is a stable ~80%. Incremental operating margins have run 30–48% — hard evidence the largely fixed-cost network absorbs incremental transactions at near-zero marginal cost (economies of scale). Notably, management is reinvesting the margin overage into AI rather than dropping it to the bottom line, and refuses to raise the target range until it is durably beaten — a tell of conservatism.

Cash flow & quality of earnings. Operating cash flow was $266M in FY26 (1.63x net income) and FCF $260M (~37% of revenue); capex is trivial (~$5.7M, <1% of revenue). Q1 FY27 OCF was $75M at 84% of adjusted EBITDA. The 1.63x OCF/NI ratio is explained cleanly: D&A of $87M vs. capex of $5.7M — ~$81M is non-cash amortization of acquired intangibles, a bookkeeping charge against cash-generative assets. Critically, stock-based compensation is only ~$21M (≈3% of revenue) — a fraction of the typical software company’s 15–25% — so FCF is real, not an SBC mirage. This is about as clean a P&L-to-cash bridge as exists in software.

Returns on capital — read past the goodwill. Reported ROE is ~16% and reported ROIC ~11%. The latter is a serial-acquirer artifact: the denominator is stuffed with $1.03B goodwill + $332M acquired intangibles ($1.36B of $1.90B total assets), and tangible book is just ~$3.04/share. On tangible/organic invested capital, the economics are extraordinary — 30%+ operating margins on <1% capex intensity with negative working capital (cash-conversion cycle −15 days; customers and deferred revenue fund the business). The correct read: incremental organic dollars earn very high returns; the ~11% blended ROIC is the price of the buy-and-build (paying fair multiples for acquired growth), not a sign of a low-return core ( Capital Allocation for the value-creation mechanism).

Balance sheet. Fortress: $377M cash (Apr-2026), effectively no funded debt (~$8M capital leases), an undrawn $350M revolver, net cash ~$349M. This funds continuous M&A and the new buyback without leverage.

Verdict: economics unambiguously improve with scale. The combination of 77% gross margins, expanding operating leverage, ~37%-of-revenue FCF, ~3% SBC, and negative working capital is elite. The only asterisk is that headline ROIC understates the core because of acquisition accounting — a feature of the compounding model, not a flaw in the business.

7. Capital Allocation

The model in one line: a self-funding, single-vertical buy-and-build. Over FY2019–FY2026 Descartes deployed ~$1.20B of cash into acquisitions against ~$1.33B of FCF — i.e., ~90% of all free cash flow recycled into M&A, with essentially zero dividends and (until December 2025) zero buybacks, while holding a net-cash balance sheet in every single year. The one equity-funded exception was the FY2020 Visual Compliance deal (~$238M bought-deal raise; share count jumped from ~77M to ~84M). Everything since FY2021 has been internally funded. [FACT]

The acquisition ledger (30+ deals since 2016), by pillar, upfront cash + max earn-out where disclosed:

• Compliance/GTI: Visual Compliance/eCustoms (2019, ~$250M — largest ever), OCR Services (2024, ~$90M), NetCHB (2023, ~$38.7M + up to $60M), Datamyne, QuestaWeb.
• Visibility/TMS: MacroPoint (2017, ~$107M), 3GTMS (Mar-2025, ~$112.7M), MyCarrierPortal (2024, ~$24M + $6M).
• Last-mile/fleet/routing: GroundCloud (2023, ~$138M + up to $80M), Idelic (Q1 FY27, ~$28M + $12M), Localz, GreenMile, PackageRoute.
• E-commerce/shipping/inventory: Sellercloud (2024, ~$110M + $20M), Finale Inventory (Aug-2025, ~$39M + $15M), ShipTrack, Peoplevox, XPS, OrderMine (Mar-2026), Kontainers.
• Forwarder systems/air cargo: BoxTop (2024), Aerospace Software Developments (2024), Portrix.

[FACT] The pattern: mostly small-to-mid tuck-ins ($5–140M), but deal sizes are creeping upward (recent flagships $90–138M vs. earlier $20–50M) — a sign they are moving up-market as the smallest niche targets get exhausted.

Deal economics — disciplined, but they pay full price. Goodwill rose monotonically from $378M (FY19) to $1,026M (FY26) with zero impairments in eight years — strong evidence deals have at least held value. [FACT] Discipline is enforced primarily through earn-out structures (GroundCloud +$80M, NetCHB +$60M, Sellercloud/Finale +$20M/$15M) that make large slices of consideration revenue-contingent, shifting risk to sellers. But Descartes discloses no acquisition multiples, so overpayment can only be inferred (zero impairments + rising margins), not measured — an [OPEN QUESTION]. The value-creation engine is not cheapness but improvement + cross-sell: they buy quality data/network assets at fair prices and re-base them to house margins via the “baseline” cost discipline, driving EBITDA margin from ~33% to ~46% while integrating 30+ deals. Reported ROIC (~11%) sits only modestly above a ~8–9% WACC, so the margin of safety is execution, not entry price. Relative to Constellation Software or Roper, Descartes is a lower-octane, lower-risk variant: single vertical, no leverage, lower ROIC — but a very clean, durable compounding machine.

The “baseline calibration” model. Descartes calibrates planned operating spend against a conservative baseline-revenue floor — its “visible, recurring and contracted revenues excluding any expected new sales” — and targets 40–45% adjusted EBITDA (now ~46%). Because opex is deliberately held below contracted recurring revenue, the business stays profitable even if it signs zero new business. Latest calibration (at 5/1/2026): Q2 FY27 baseline revenue ~$169M, baseline opex ~$102M, baseline adjusted EBITDA ~$66.5M (~39%). Since baseline revenue (~$169M) is ~87% of actual Q1 revenue ($193.6M), only ~13% of quarterly revenue is unmodeled “new business” — the mathematical core of the downside protection. [FACT]

Management incentives & alignment — the real weakness. Insider ownership is <1% of the company. CEO Ed Ryan owns only ~0.05% directly (~CA$4.45M) and was a net seller — disposing of ~34,193 shares (~43% of his direct stake) around April 2026 at ~CA$89.70 (appears routine/diversification, but there was no insider buying into the ~40% drawdown). Total CEO comp ~$10.4M, ~94% variable/equity. Top holder T. Rowe Price ~8% (largest disclosed 13G); BlackRock ~5%; institutions ~86%; public float ~17%. Governance is conventional (KPMG auditor, annual say-on-pay, a shareholder-rights plan renewed in 2026). [FACT] Verdict on alignment: weak skin-in-the-game and a selling CEO are a genuine negative — you are underwriting a process/culture, not an owner-operator, a real contrast with founder-led compounders.

The buyback pivot — good sign, not warning. The first meaningful buyback in company history: an NCIB commenced 12/11/2025 for up to 8.6M shares (10% of float) through 12/10/2026; Q1 FY27 repurchased ~305,000 shares for $20.8M (~$68/sh) plus ~196,800 more in May–June 2026 — while still doing tuck-ins (Idelic, OrderMine, PackageRoute). [FACT] Read: opportunistic capital return during a price dislocation, a healthy second lever, not evidence of pipeline exhaustion — though the up-drift in deal sizes keeps the “shrinking small-deal runway” question open.

Verdict: intelligent and disciplined capital allocation, with two asterisks — (1) value is created by improving full-priced acquisitions, so the reported ROIC only modestly clears WACC and the safety margin is execution, not price; and (2) weak insider alignment. The buy-and-build is repeatable and self-funding, the balance sheet is permanently conservative, dilution is minimal, and the new buyback is a rational addition.

8. Changes and Headwinds — Last Two Years

Acquisitions (thesis-positive). A steady cadence of tuck-ins: OCR, ASD (aerospace), BoxTop, MyCarrierPortal, Sellercloud (FY25); 3GTMS ($112.7M, Mar-2025, the largest recent deal); PackageRoute (Jun-2025); Finale Inventory ($39M, Aug-2025); OrderMine (AI demand-planning, Mar-2026); and Idelic (AI driver-safety, Q1 FY27). The M&A machine kept running throughout the downturn.

Cost discipline / demand signal (mixed). In mid-2025 (Q2 FY26) Descartes executed a ~7% global workforce reduction plus an opex cut “in light of economic uncertainty in the global trade environment” (a $4.6M charge, completed by Jan-2026). [FACT] This is simultaneously a positive (margin defense; the baseline model in action) and a caution (a demand signal — management saw enough softness to cut).

Leadership transitions (neutral, but a raised change-surface). CFO transition — Edward Gardner succeeded Allan Brett (effective 3/12/2026; Brett to advisory/retirement); CCO Andrew Roszko departed in Q2 FY26; and William Green was elevated to EVP Global Sales. Three senior changes in ~12 months, on top of a long-tenured CEO (Ryan since 2013), elevates key-person/execution surface even if each move looks orderly.

The tariff and trade-regime whirlwind (net tailwind for the franchise). A dizzying 2025–26 sequence: the Type-86 de minimis exemption ended for China/HK (~May-2025) and then all countries (EO 7/30/2025, effective 8/29/2025) — a structural tailwind forcing formal customs entries on every low-value parcel (Descartes gained NetCHB share as rivals’ non-network systems buckled); the Supreme Court struck down the IEEPA tariffs (2/20/2026) and CBP launched a refund process (4/20/2026) — adding complexity (work for the software) but also raising the risk that trade normalization eventually cools elevated GTI demand; and new China extraterritorial regulations plus a Supreme Court broker-liability ruling (both flagged by management on the Q1 FY27 call) that further increase compliance complexity (and demand for Descartes’ MyCarrierPortal carrier-vetting). [FACT — Q1 FY27 call + web-dated events]

Freight/geopolitical headwinds (thesis-negative, cyclical). The Iran war closed/curtailed the Strait of Hormuz in Q1 FY27, disrupting ocean shipping (oil, fertilizer, aluminum), raising rates and insurance, and lengthening routes; US trucking volumes were down ~4%, and overall shipment volumes fell. This is the genuine cyclical headwind on the transaction tail — but it coincided with accelerating organic growth, underlining the compliance/volume divergence.

Capital-return & events. The NCIB buyback (Dec-2025) is the first-ever buyback lever. Management is hosting an in-person Innovation Forum in Chicago, October 6–8, 2026, to lay out its AI strategy — a potential catalyst for the AI-monetization narrative.

Verdict: on balance, the changes strengthen the thesis. The structural (compliance formalization, de minimis elimination, M&A cadence, buyback initiation, margin discipline) outweighs the cyclical (freight/Iran volume drag) and the governance noise (CFO/CCO turnover, weak insider buying). The one change to watch is any normalization of the tariff regime that removes the GTI tailwind faster than the freight cycle recovers.

9. Risk Analysis

Catastrophic-loss risk is low. Descartes carries no debt, holds ~$377M cash, is ~93% recurring, and is diversified across products/geographies/customers — there is no realistic path to a total loss. The dominant risks are to the growth rate and multiple (a GTI air-pocket if tariffs normalize before freight recovers, or continued de-rating), not to solvency. The most under-appreciated positive skew is regulatory: unlike most software, Descartes benefits from more trade regulation; deregulation is the low-probability tail.

10. Valuation Discussion (embedded expectations)

Where the multiple sits. At ~$72.80 (~87.6M diluted shares → ~$6.4B market cap; ~$6.0B EV net of $377M cash), DSGX trades at roughly 19.7x trailing GAAP EV/EBITDA, ~17–18x forward adjusted EBITDA, ~39x trailing / ~33x forward GAAP P/E, ~8.3x EV/sales, and ~24x P/FCF (~4.2% FCF yield). The decisive datum is own-history context: on the AZI percentile ranks the stock sits at the ~1.5th percentile on P/E and the ~6.9th percentile on the composite of its own ~decade range. Descartes has spent most of its life at 50–100x P/E and 29–37x EBITDA; the EV/EBITDA multiple has roughly halved from ~37x (FY25) to ~20x (FY26). On its own history, this is close to the cheapest the stock has ever been — the mirror image of a “great business at its richest-ever multiple,” and the reason the risk/reward has inverted versus two years ago.

Cross-sectional check. Against factor-similar quality-software compounders, DSGX is no longer a premium outlier:

[FACT] ~17–18x forward EBITDA for a business growing revenue mid-teens with 46% margins, net cash, and a repeatable M&A engine is a market-to-discount multiple relative to these peers — the offset being DSGX’s somewhat slower organic growth (~9–10% vs. MANH cloud +22%). It is not statistically cheap in absolute terms — 20x EBITDA is a full price — but it is cheap relative to its own history and to the durability of the cash flows.

Embedded expectations. At ~33x forward earnings with ~14% revenue growth and steady margins, the market is underwriting roughly low-to-mid-teens FCF/EPS compounding for the next several years — a continuation of the historical algorithm (~9% organic + ~4–5% M&A, modest further margin gains, a shrinking share count). That prices durability correctly but gives little credit for (a) a freight-cycle recovery stacking volume growth atop the compliance tailwind, (b) GTI/agent monetization as a new revenue layer, or © accelerated buybacks at a depressed multiple. Conversely, the price is too high if organic growth reverts to mid-single-digits as tariff volatility normalizes.

Reverse-DCF sanity check. Working backwards from ~$72.80 on a simple owner-earnings frame: FCF is ~$260M today (~4.2% FCF yield on the ~$6.4B market cap). A ~9% FCF discount rate less a terminal growth of ~4% implies a ~5% cash yield would be “fair” for a no-growth perpetuity — so the ~4.2% starting yield is pricing in only modest growth relative to the business’s demonstrated ~12–14% top-line and ~20%+ per-share compounding. Put differently, at ~24x P/FCF the market is underwriting roughly a decade of low-double-digit FCF/share growth and then a fade — a reasonable-not-cheap expectation for a franchise that has actually delivered ~27% EPS CAGR over six years. The embedded expectation is neither euphoric (as at 45x P/FCF in early 2025) nor distressed; it is a fair price for continued compounding, with the upside coming from either a growth surprise (freight recovery + GTI/agents) or continued buybacks at a depressed multiple, and the downside from a growth disappointment (tariff normalization air-pocket) that would justify a further de-rate.

Scenario framing (illustrative, not a target).

• Bear: tariff volatility fades, freight stays weak, organic growth slips to ~5–6%, multiple compresses to ~15x EBITDA → a stock in the low-$50s.
• Base: ~9% organic + tuck-in M&A, ~44–46% margins, ~18–20x EBITDA holds → high-$70s to high-$80s over 12–24 months, plus buyback accretion.
• Bull: freight recovers, GTI/agents add a growth layer, organic re-accelerates toward ~11–12%, multiple re-rates to ~24–26x EBITDA → $110–125+ (retest of the highs).

Why “cheap vs. own history” is the right lens, not “cheap absolutely.” At ~20x EV/EBITDA and ~33x forward earnings, DSGX is not a statistical value stock, and a value screen would never surface it. The correct comparison is intertemporal and cross-quality: the same business traded at ~37x EBITDA / ~69x P/E eighteen months ago, and factor-similar quality compounders (MANH ~35x, TYL ~38x EBITDA) trade far richer today. The valuation case is therefore “quality de-rated to a fair multiple,” not “cheap asset.” That is a weaker mispricing claim than a classic deep-value setup — which is precisely why the labeled view (the Author’s Take) is a HOLD/accumulate rather than a table-pounding buy.

No price target and no recommendation in this section (the single labeled view is in the Author’s Take).

11. Variant Perception

Consensus view. The sell-side and the tape treat DSGX as a high-quality but freight-cyclical software name that got ahead of itself and is now working off a bubble multiple — a “wait for the freight cycle” hold. The factor tape corroborates the abandonment: FactorsToday shows Momentum −0.31, one-year total return −28% (Sharpe −0.80), ~40% off peak, and a −57%-scale lifetime drawdown — a long-term winner (lifetime +15.8%/yr) that momentum investors have left for dead. Notably, the risk model also loads DSGX on Sector:Financials (+0.26) and a “Financial Data Titans”/Broker-Dealers basket — i.e., it partly reads Descartes as a data/information-services franchise, not pure app-software, which is the correct lens.

Strongest bull case. Descartes is a counter-cyclical trade-data + neutral-network utility mispriced as a cyclical loser. Its highest-margin franchise (GTI) feeds on the tariff/geopolitical chaos that scared the stock, organic growth is accelerating (9.5%) not fading, margins are at records (46%), the balance sheet is a fortress, and the multiple is at a decade-cheap level and a discount to quality peers — with a fresh buyback shrinking the count into the weakness. A freight recovery would stack a volume tailwind on top, and AI-agent monetization is genuine optionality. You are buying a durable ~14% compounder at ~20x EBITDA instead of ~37x.

Strongest bear case. The GTI tailwind is borrowed from a one-off tariff-chaos regime; when trade normalizes (IEEPA tariffs already struck down, refunds flowing), GTI decelerates into an air-pocket while the freight cycle is still weak, and organic growth reverts to mid-single-digits — at which point ~33x forward earnings and ~20x EBITDA is not cheap, it’s a de-rating waiting to finish (bear case: low-$50s). Reported ROIC barely clears WACC, insiders own <1% and are selling, and value creation depends on continuing to buy ever-larger acquisitions at full prices with zero disclosed multiples — a machine that works until it doesn’t. AI could commoditize the very compliance-data premium the bull case rests on.

The 3–5 assumptions that matter most:

1. Is organic growth structurally ~9–10%, or a tariff-chaos sugar high? (Bull: durable, compliance + share gains; Bear: borrowed from the tariff regime.) Falsifier: organic services growth over the next 3–4 quarters — sustained ≥9% confirms bull; deceleration below ~6% confirms bear.
2. Does the freight cycle recover, adding volume to compliance? Falsifier: shipment-volume trends in the transaction tail.
3. Is the moat (switching costs + proprietary compliance data) AI-proof? Falsifier: GTI pricing power and retention as frontier models parse public tariff schedules.
4. Is the buy-and-build still repeatable at acceptable returns? Falsifier: deal sizes/multiples, and any goodwill impairment.
5. Does weak insider alignment matter? Falsifier: capital-allocation discipline and any value-destructive large deal.

Where consensus may be offsides: the tape has extrapolated the volume downturn and ignored the compliance acceleration, treating a −0.31-momentum, decade-cheap compounder as broken when its organic engine is re-accelerating. That is the empirical signature of a name where the cyclical story has crowded out the structural one — the core of the variant view. The counter-risk is equally clear: if tariffs normalize before freight recovers, the bears are right that the “acceleration” was borrowed.

12. Fact vs. Interpretation Table

13. Open Questions

1. What is the transaction-vs-subscription split inside “services”? Descartes does not disaggregate it, so the freight-cyclical slice cannot be precisely sized — the single biggest analytical gap.
2. What EBITDA multiples does Descartes pay for acquisitions? Undisclosed; overpayment can only be inferred (zero impairments) not measured.
3. How durable is GTI growth if the tariff regime normalizes? Is ~9.5% organic a new baseline or a tariff-chaos peak?
4. What is true net revenue retention (not just gross)? Management gives gross attrition (~5–7%/yr) but not NRR/expansion.
5. Does the AI-agent layer actually monetize at scale, or is it a moat-defense narrative? (October 2026 Innovation Forum may clarify.)
6. Is the up-drift in deal sizes a sign the small-deal runway is thinning, or simply Descartes moving up-market with a bigger balance sheet?

14. What Must Be True

For the bull case (the stock is a durable compounder mispriced as a cyclical):

• Organic services growth stays ~9–10% and is not merely a tariff-chaos sugar high. Falsification test: organic services growth decelerates below ~6% over the next 2–3 quarters even as management laps the tariff surge → the acceleration was borrowed, and 33x forward P/E is too high.
• The moat (compliance switching costs + proprietary data) survives AI. Falsification test: GTI pricing power or gross retention visibly erodes as public tariff schedules become trivially parseable by frontier models.
• The buy-and-build stays repeatable at returns above WACC. Falsification test: a goodwill impairment, or a large out-of-lane acquisition at a visibly rich multiple.

For the bear case (the stock is a full-priced compounder facing a growth air-pocket):

• Tariff normalization removes the GTI tailwind faster than the freight cycle recovers. Falsification test: shipment volumes recover and GTI growth holds, so total organic growth stays ≥9% — the air-pocket never materializes.
• Reported ~11% ROIC signals a low-return core. Falsification test: incremental organic margins stay 30%+ and FCF/revenue stays ~37%, confirming the core economics are far above the goodwill-diluted headline.

The single most important number to watch is organic services growth: sustained ≥9% vindicates the bull/structural read; a slide toward mid-single-digits vindicates the bear/cyclical read. The single most important qualitative watch item is capital-allocation discipline given weak insider alignment.

15. Source Appendix

Primary and quantitative sources (full detail in the standalone Source Appendix):

• Descartes 40-F (FY ended 1/31/2026), filed 2026-03-11 (SEC EDGAR CIK 0001050140); interim 6-K filings (5-year corpus).
• Q1 FY2027 earnings call transcript, 2026-06-03 (ROIC.ai) and FY26-Q4 / FY27-Q1 press releases (GlobeNewswire, 2026-03-11 / 2026-06-03).
• ROIC.ai financial statements, ratios, enterprise value, and valuation multiples (FY2019–FY2026; MANH/TYL comps), accessed 2026-07-04.
• AZI valuation-index own-history percentiles (2026-07-02) and daily price CSV (full history).
• FactorsToday stock-loadings, leaderboard, stock-info, related-stocks (2026-07-02/04).
• Industry sizings: Fortune Business Insights, The Business Research Company (GTM); MarketsandMarkets (TMS). Competitor/deal data: company/FreightWaves/BetaKit press releases (2017–2026); WiseTech Global filings. Ownership: SC 13G filings (T. Rowe Price, BlackRock); SEDI/MarketBeat (insider activity).

All non-obvious facts are cited to primary or clearly-attributed third-party sources. Third-party aggregated data (ROIC.ai, AZI, FactorsToday) is reconciled to the filings; where a filing and an aggregator disagree, the filing governs.

APPENDIX A — Standard Diligence Questionnaire — The Descartes Systems Group Inc. (NASDAQ: DSGX)

Supplemental to the research memo. Report date 2026-07-04. FY ends January 31; USD.

General

What thoughtful questions have other investors asked about this company? The recurring debates, per the Q1 FY27 call and sell-side exchanges: (1) At what point does trade complexity flip from tailwind to headwind? — management’s answer: complexity almost always helps (customers use more software); it only hurts when complexity depresses the economy and shipment volumes fall, at which point Descartes “goes right along with the economy.” (2) Can customers use AI to replicate the GTI tariff database? — management (and our analysis) says no: the moat is data breadth + normalization + delivery-format + compliance liability, not raw computation. (3) Is the ~40–45% EBITDA target too conservative given ~46% actuals? — management deliberately keeps the range to avoid missing, and to reinvest overage into AI. (4) Is the M&A runway shrinking? — deal sizes are creeping up; the new buyback raised the question of pipeline health (management says private valuations are coming down and it remains “very busy”). (5) Does the ~40% drawdown mark value or a broken cyclical?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Mixed and nuanced: margins are at a cyclical/structural high (record 46% adj. EBITDA), but shipment volumes (the transaction tail) are at a cyclical low (freight recession, trucking −4%, Iran-war ocean disruption). Organic growth (~9.5%) is accelerating despite the volume trough because the compliance/GTI franchise is counter-cyclical. Net: earnings are neither obviously peak nor trough — margins are high, but a freight recovery would add volume growth on top.

Driven by external environment or internal actions? Both. Margin expansion (33%→46% over six years) is internally driven (the “baseline calibration” cost discipline + operating leverage on a fixed-cost network). The recent GTI acceleration is externally driven (tariff/sanctions chaos) — the key reason to watch for a normalization air-pocket.

How stable are revenues? Very — ~93% recurring, ~93–95% gross retention, ~$119M deferred revenue, and a conservative “baseline” (~87% of quarterly revenue is contracted/visible). The ~13% unmodeled slice plus the transaction tail are the variable pieces.

Outlook for products/services; how big is the market? GTM/trade-compliance ~$1.3–2.5B growing ~8–12%; TMS ~$18.5B→$37B growing ~15%. Growing, global, and structurally expanding with trade complexity, e-commerce customs formalization, and nearshoring.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More, at the periphery (visibility/TMS is a VC-subsidized knife-fight — project44, FourKites; WiseTech bought e2open for ~$2.1B). Less/stable at the core (customs/compliance is capital-scarce, content- and liability-gated).

How profitable is the business (ROIC, ROE)? ROE ~16%; reported ROIC ~11% (a goodwill artifact — the tangible/organic core earns far higher returns: 30% op margin, <1% capex, negative working capital). 77% gross margin, ~46% adj. EBITDA.

How profitable is the industry; barriers to entry? The compliance/GTM pool is oligopolistic and high-margin; entry barriers are the tariff/sanctioned-party content, regulatory certification, and network integration. The visibility pool has low barriers and thin differentiation.

Can the business be easily understood? Yes at the model level (recurring SaaS + data + network + tuck-in M&A); no at the product level (300+ modules across six pillars). The “network/data utility” framing is the right mental model.

Undermined by foreign low-cost labor? No — it’s software/data, not labor-arbitrageable; if anything, offshore development lowers Descartes’ own cost.

Do brands matter? Nature of competition? Switching costs? Brand matters modestly (MacroPoint, NetCHB are known names). Competition is by pillar (see memo). Switching costs are high in customs/compliance (mission-critical, audited, liability-bearing) and moderate elsewhere.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Yes — the organic network/content/customer base is largely internally generated and under-carried; conversely, ~$1.36B of goodwill+intangibles (72% of assets) over-states tangible capital. Tangible book is only ~$3/share.

Off-balance-sheet liabilities? Contingent earn-out consideration (up to ~$9M for the balance of FY27, plus deal-specific earn-outs) and operating leases — modest. No debt.

How conservative is the accounting? Conservative — low SBC (~3% of revenue), no revenue-recognition red flags, OCF/NI ~1.63x (amortization-driven, benign), and a deliberately conservative “baseline” disclosure. KPMG auditor.

How CapEx-hungry? Barely — capex ~$5.7M (<1% of revenue). Asset-light network model.

Capital Allocation & Management

How much FCF, and how is it used? ~$260M FCF (37% of revenue); ~90% historically recycled into tuck-in M&A, with a new NCIB buyback (up to 8.6M shares) added Dec-2025. No dividend.

Significant acquisitions recently? Yes — 3GTMS ($112.7M, 2025), Sellercloud ($110M), Finale ($39M), Idelic, OrderMine, PackageRoute — a continuous cadence of ~$150M/year.

Buying back shares? Yes, newly — ~$21M in Q1 FY27 + more May–June 2026; first-ever buyback, opportunistic into the drawdown.

Issuing large amounts of stock to insiders? No — SBC is low (~3% of revenue); dilution has been minimal (~85M→~87.6M diluted shares over six years) despite heavy M&A (deals are cash-funded).

Compensation policy / motivations of management? CEO comp ~$10.4M, ~94% variable/equity. Weak alignment — insiders own <1%, CEO ~0.05% and a net seller. This is a process/culture compounder, not owner-operator.

Valuation & Market Data

ADR, MLP, or K-1 issuer? None — it is a Canadian MJDS foreign private issuer with common stock dual-listed on NASDAQ (DSGX) and TSX (DSG). Files 40-F/6-K, not 10-K/10-Q. No K-1. Reports in USD.

Dividend policy? No dividend; capital returned via M&A and (newly) buybacks.

How profitable? Net income vs. cash from operations? Highly profitable (22% net margin). CFO ($266M) exceeds net income ($164M) by ~1.63x — a favorable divergence driven by non-cash amortization of acquired intangibles, not an accrual red flag.

Risks & Downside

What would cause the stock to decline? A GTI growth air-pocket if tariffs normalize before freight recovers; continued multiple de-rating; a value-destructive large acquisition or goodwill impairment; AI commoditization of compliance content; or a broad software/growth selloff (the stock is high-beta to the software factor).

Risk of catastrophic / total loss? Very low. No debt, ~$377M cash, ~93% recurring, diversified. The risk is to growth and multiple, not solvency.

Recent News & Events

Has the business environment changed recently? Yes — a whirlwind of trade-regime change (de minimis elimination = tailwind; IEEPA tariffs struck down + refunds = complexity/normalization risk; new China extraterritorial rules; broker-liability SCOTUS ruling), plus the Iran-war ocean disruption and a weak freight cycle. Net tailwind for the compliance franchise, headwind for shipment volumes. (Note: the AZI news feed returns no rows for DSGX; this timeline is built from filings, the Q1 FY27 call, and dated public events.)

Significant acquisitions / accounting changes / other recent changes? Acquisitions ongoing (see above); a ~7% workforce reduction in mid-2025; CFO transition (Gardner ← Brett, Mar-2026) and CCO departure; NCIB buyback initiated; and an October 2026 Innovation Forum to detail the AI strategy. No accounting-policy changes of note.

APPENDIX B — Source Appendix — The Descartes Systems Group Inc. (NASDAQ: DSGX)

Report date 2026-07-04. Primary sources prioritized over secondary; third-party aggregated data reconciled to filings.

Primary — Company Filings (SEC EDGAR CIK 0001050140 / SEDAR+)

• Form 40-F, FY ended 2026-01-31 (filed 2026-03-11) — annual report wrapping the Canadian AIF + MD&A + audited financials. Revenue by type/geography, gross-margin by line, recurring-revenue attrition (~5–7%/yr), goodwill/intangibles, baseline-calibration methodology, risk factors (“Certain Factors That May Affect Future Results”), competitor list. Primary source for /////
• Interim 6-K filings (5-year corpus, mirrored locally) — quarterly results, MD&A, press releases, NCIB announcements.
• Q4 FY2026 results 6-K / press release (2026-03-11, GlobeNewswire) — FY26 full-year figures.
• Q1 FY2027 results 6-K / press release (2026-06-03, GlobeNewswire) — quarterly record ($193.6M revenue, 46% adj. EBITDA, $48.5M net income), baseline calibration for Q2 FY27.
• SC 13G / 13G-A filings — T. Rowe Price (~8%, event 2025-12-31), BlackRock (~5%, event 2026-03-31); passive holders, no 13D/activist.
• SD (conflict-minerals) filings (2025, 2026) — immaterial to thesis.

Primary — Management Commentary

• Descartes Q1 FY2027 earnings call transcript, 2026-06-03 (Ed Ryan, CEO; Ed Gardner, CFO; Scott Pagan) — organic services growth ~9.5% (accelerating 3 quarters); GTI/e-commerce/fleet/MacroPoint growth drivers; MacroPoint track rate 87%→93%; AI-agent layer; freight/Iran disruption; broker-liability ruling; China extraterritorial regulations; NCIB buyback; M&A framework. Accessed via ROIC.ai.

Quantitative Data Providers (reconciled to filings)

• ROIC.ai MCP — income statement, balance sheet, cash flow, profitability/valuation ratios, enterprise value, per-share data (FY2019–FY2026); comps MANH/TYL FY2025 multiples. Accessed 2026-07-04.
• AZI (azitrading.com) — valuation_index own-history percentiles (P/E ~1.5th, composite ~6.9th; 2026-07-02); daily price/OHLCV CSV (full history) for the five-year event map. AZI news feed returned no rows for DSGX (feed unavailable for this ticker; noted in memo).
• FactorsToday (factorstoday.com/api) — stock-loadings (Momentum −0.31, Growth +0.25, Financials +0.26; R²~41%), leaderboard (y1 −28%, lifetime +15.8%/yr, max drawdown −57%), stock-info (rs_12m −28.7, rs_peak −40.6%), related-stocks (MANH, TYL, BSY, OTEX). Accessed 2026-07-02/04.
• SEC EDGAR XBRL / filings index (scripts/edgar.sh) — CIK resolution, filing enumeration, corpus mirror (62 documents).

Industry & Competitor Data (secondary, attributed)

• Fortune Business Insights, The Business Research Company — Global Trade Management market sizing (~$1.3–2.5B, ~8–12% CAGR).
• MarketsandMarkets — Transportation Management Systems market (~$18.5B 2025 → ~$37B 2030, ~15% CAGR).
• WiseTech Global (ASX:WTC) filings/results — CargoWise ~$700M revenue, ~53% EBITDA; e2open acquisition (~$2.1B EV, Aug-2025).
• project44, FourKites funding/valuation data (press/PitchBook-type disclosures).
• Company/FreightWaves/BetaKit deal press releases (2017–2026) — acquisition dates/prices (MacroPoint, Visual Compliance, 3GTMS, GroundCloud, Sellercloud, OCR, NetCHB, Finale, Idelic, OrderMine, etc.).
• MarketBeat / SEDI — CEO Ed Ryan insider sale (~34,193 shares, ~2026-04-13, ~CA$89.70); insider ownership.
• Simply Wall St — ownership breakdown cross-check.

Public-event dating (trade-regime timeline)

• Type-86 de minimis elimination (China/HK ~2025-05-02; all-country EO 2025-07-30, effective 2025-08-29); SCOTUS IEEPA-tariff invalidation (2026-02-20); CBP refund processing (2026-04-20) — dated via public reporting/CBP notices.

Internal / Cross-read

Every material number in the memo reconciles to a filing; ROIC.ai/AZI/FactorsToday are third-party aggregated inputs used for acceleration and cross-check, not as primary authority. Management commentary is treated as hypothesis and validated against filings and external data.