Chapter 1

Network, Marked Down

SPS Commerce sells something most investors never see and every large retailer depends on: the plumbing that lets a supplier and a store trade purchase orders, invoices, and shipping notices without either side building the connection by hand. It has done this for long enough — a 100th consecutive quarter of revenue growth closed out 2025 — that the business now reads like a utility, with 96% of its $751.5 million in FY2025 revenue recurring and no single customer worth as much as 1% of the total [1]. And yet, over the two years to July 2026, the stock lost roughly two-thirds of its value — a deeper fall than any company it competes with except one.

That gap is the reason to look closely. This report follows the split inside it. SPS runs a full-service retail EDI network whose connection flywheel still shows up in peer-leading growth and clean cash conversion; the 2024–2026 stock collapse came from a reset in the growth algorithm after acquisition-flattered customer additions rolled off and the Amazon third-party detour was sold. The core Fulfillment network may still be compounding — while the founder-exited management era must prove that richer wallet share and margin expansion can replace customer-count growth without leaning on M&A or buybacks. This chapter establishes the network, the field it plays on, and the paradox that makes the price interesting now. The chapters that follow trace the mechanism, the margins, and the people spending the cash.

Connect once, transact with thousands

The market SPS serves solves a specific, unglamorous problem. A supplier that wants to sell to Target, Kroger, and a regional grocery chain must exchange structured electronic documents — the format known as EDI, electronic data interchange — with each of them, and every retailer keeps its own rules for how those documents must look. Built by hand, each connection is a small integration project that has to be maintained forever. SPS replaces that with a shared network: a customer connects once to SPS and can then transact with any trading partner already on it, while SPS maintains the compliance rules, onboards the partners, monitors the transactions, and resolves the errors [2].

Source: FY2025 Annual Report (Form 10-K), Item 1 Business — revenue, recurring share [3]; recurring-customer count from the Q4 FY2025 earnings call [4].

The industry sorts itself into three delivery methods, and the distinction is the whole story of where SPS sits. Suppliers can license traditional on-premise software and run the integration themselves; they can buy managed services; or they can hand the entire function to a full-service cloud provider that "customizes, optimizes, and operates the technology" on their behalf [5]. SPS is a full-service provider. That choice — doing the work rather than selling the tool — is what makes the service sticky, and, as later chapters show, it is also what holds its operating margin below the software-only names it is measured against.

The value of such a network compounds with its own size. In the company's words, "the value of our network increases exponentially with scale": a new customer typically finds that many of its existing trading partners are already connected, so it can transact immediately, and every partner already on the network gains one more counterparty to reach [6]. By the end of 2025 SPS carried roughly 54,600 recurring-revenue customers spending an average of about $14,300 a year [7]. That is a genuine network effect — a self-reinforcing customer-acquisition channel, not just a good product — and it is the mechanism to hold onto as the rest of the report questions whether it is still turning.

The field is crowded and split. SPS itself calls the market for cloud-based supply-chain management "fragmented and rapidly evolving," competing on the breadth of pre-built connections and a history of reliable ones far more than on price or features [8]. Management sizes the opportunity, via a third-party consultant, at $11.1 billion globally — about 275,000 potential customers at roughly $40,500 of average annual spend [9]. Against $751.5 million of revenue, that puts SPS near 7% of the dollars and a fifth of the potential accounts — a long runway on the company's own math, with the honest caveat that the figure is self-sourced and no independent estimate in the record confirms it.

That the network structure is real, rather than one company's marketing, is visible in how competitors describe their own businesses. OpenText markets its Business Network Cloud as a way for "organizations of all sizes" to "rapidly onboard new trading partners, comply with regional mandates … [and] provide electronic invoicing" — SPS's exact value proposition, sold by a company more than six times its size [10]. Descartes runs a logistics network on the same logic. So the competitive set is not a single scaled rival but a fragmented mix: giant diversified players such as IBM and OpenText for whom B2B networks are one line among many, plus a long tail of small value-added networks, with SPS concentrated in the retail supplier-onboarding niche.

The share-taker

Inside that mature field, SPS grows like something much younger. FY2025 revenue rose 17.8%. The median of its named public peers grew 3.7% — a gap of 1,409 basis points — and the spread runs from OpenText shrinking 10.4% to IBM at 7.6% and Manhattan Associates at 3.7%.

Source: reported financials, FY2025 — SPS FY2025 revenue of $751.5 million [11] and each peer's FY2025 filing; OpenText Business Network Cloud [12].

This is the first anomaly the report owes an explanation: SPS grew nearly five times the peer median while the one rival it competes against most directly, OpenText, was contracting. Growing that much faster than the field, and faster than the incumbent selling the identical promise, is the signature of a company taking share rather than defending it — and the network effect is the plausible engine. Two structural conditions support the pace. The first is switching cost as inertia: SPS's contracts are largely auto-renewing, its revenue is 96% recurring, and no customer matters enough to move the numbers if it leaves [13]. The second is a demand tailwind that shows up in more than one company's filings.

That tailwind is government e-invoicing regulation. As more countries mandate electronic invoicing for tax purposes, a compliance obligation becomes recurring network demand. SPS's management notes that "e-invoicing capability is a frequent requirement for customers implementing fulfillment outside of the U.S. in countries where there is an e-invoicing mandate" [14], and OpenText independently attributes network expansion to the same rising requirement [15]. A condition two competitors describe on their own is a sturdier fact than one the company claims alone. The benefit is real but geographically concentrated in Europe, while SPS's base is still overwhelmingly North American — so it depends on an international expansion SPS has not yet scaled.

The harder-to-call force is AI, and the industry is genuinely of two minds. Descartes argues that proprietary network data becomes more valuable as AI spreads — "Data is the fuel for AI solutions," and it must come from a trusted source [16]. Read that way, the compliance and connection data these networks hold deepens the moat. The opposite reading is equally available: the single most important competitive basis SPS names is the breadth of pre-built integrations, and AI is precisely the technology that lowers the cost of building an integration. The same shift could widen the entry barrier or erode it, and nothing in the current record settles which. Meanwhile the adjacent players are not standing still — Manhattan Associates reported competitive win rates above 70% with most new bookings from net-new logos, evidence that well-capitalized neighbors keep pressing on the same retailer-and-supplier base SPS serves [17].

Marked down harder than anyone

Here the operating record and the share price part company. From a high of $215.42 on 31 July 2024, SPS closed at $73.39 two years later — a 65.9% drawdown, against a peer median of 39.1%. Only WiseTech, at 73.9%, fell further; OpenText, IBM, and Manhattan each gave back between 32% and 40%. The 2,682-basis-point gap to the median says the market repriced something specific to SPS, not a sector or interest-rate move that would have taken the whole group down together.

Source: derived from daily price data through 31 July 2026 (SPS and peers); peer median −39.1%.

What makes the fall unusual is when it happened. The collapse ran across earnings days — the stock fell 13.8% the day it reported in February 2025, 22.0% at the end of July 2025, and 20.8% at the end of October 2025 — and on each of those reports, and every one in between, normalized earnings per share came in above consensus. Across the eight quarters through mid-2026, EPS beat by between 10% and 18% each time.

Source: consensus estimates and reported results; each quarter's normalized EPS exceeded consensus.

A stock that falls on beats is not being punished for the quarter it delivered. It is being repriced for the growth it is now expected to deliver. That is the shape of this decline: a growth-rate reset and multiple compression, not an earnings failure. The arithmetic of the current price bears it out. At $73.39 the market values SPS at roughly $2.8 billion, about 14.9 times forward normalized EPS and a 6.5% forward free-cash-flow yield — a low-double-digit multiple for a business with 69% gross margins, roughly 20% free-cash-flow margins, and net cash on the balance sheet. The multiple that once priced high-teens compounding is gone. In its place, consensus embeds forward revenue growth of about 5.3%, down from the 17.8% just delivered, and the sell-side has converged its price targets to a mean of $72.73 — at or below the spot price. Analysts, in short, are modeling the reset as the new normal rather than a recovery.

That reset is the question the rest of the report exists to test, and its mechanism is hiding in a number already on this page. The customer count reached about 54,600 at the end of 2025 — but the growth the market is now pricing does not assume that figure keeps climbing. It assumes the opposite. Why the unit that has always signaled network expansion is now set to shrink, and what part of the recent growth was borrowed from an acquisition, is where the story turns next in The Customer Count Break.