SPS Commerce [SPSC]
Every balance-sheet input the screen used is correct. Shares 36,712,702 ✓. Net cash $154,271k ✓ (zero debt, no securities portfolio). TTM revenue $762,077k ✓. EV/Sales 3.09x ✓. EPS cross-check ties to the cent.
The failure is different, and larger. The screen ranked SPSC on an 18.6% "demonstrated CAGR" — a trailing three-year average — at a moment when:
| Q1-2026 actual revenue growth | +5.8% |
| Q2-2026 company guidance | +4% to +5% |
| FY2026 company guidance | +6% to +7% |
| What today's price requires | 8.1% |
The company's own guidance sits below what its price requires. The screen's +11.5pp margin becomes −2.3pp, and the Valuation Criteria returns FAIL — on management's numbers, not on a house forecast and not on a haircut.
Nine months of monotonic deceleration: +22.0% → +16.0% → +12.7% → +5.8%, 1,620 basis points. Q1-2026 was also the first sequential revenue decline in the disclosed series.
SPS Commerce is a serial acquirer: goodwill $143.7m (FY2021) → $540.8m (Q1-2026), four deals in fourteen months, $424m spent on the two large ones.
The FY2025 10-K publishes a Carbon6 pro forma, and it settles the question:
| (unaudited) | FY2025 | FY2024 |
|---|---|---|
| Pro forma revenue | $755.7m | $680.9m |
| Actual revenue | $751.5m | $637.8m |
Pro forma growth = +11.0%. Reported growth = +17.8%.
Carbon6 alone accounts for 6.8 percentage points — 38% — of reported FY2025 revenue growth.
Corroborated independently by the auditor's internal-controls scope exclusion, written for an entirely different purpose: "Carbon6 … represented … approximately 5% of consolidated revenues" ⇒ ~$37.6m, or 33% of the $113.7m FY2025 revenue increase, from one deal. And the pro forma still understates the acquired share, because it adjusts for Carbon6 only — SupplyPike, Traverse and Vision33 all closed mid-2024 and their incremental months remain in the "organic" residual.
Estimated organic FY2025 growth: ~9–11% against a reported 17.8%. Roughly 40–45% of FY2025 growth was bought.
The customer count is starker. Recurring revenue customers rose 45,350 → 54,600, +9,250 (+20%). The 10-K: "approximately 8,500 recurring revenue customers were added in February 2025 due to the acquisition … of Carbon6."
92% of net customer additions were acquired. Organic net adds were roughly 750 — +1.7%.
And Carbon6's ~8,200 3P customers are Amazon marketplace sellers, not retail-EDI trading partners — a different end market from the network whose effects the equity story rests on.
96% of FY2025 revenue is recurring (94% in FY2024 and FY2023) and the largest customer is <1% of revenue. That part of the claim holds.
But SPSC discloses no net revenue retention, no gross retention and no churn figure — anywhere. What exists instead is ARPU, computed as "annualized recurring revenues … divided by the average of the beginning and ending number of recurring revenue customers." When 8,500 explicitly-lower-ARPU customers arrive in February — mid-period — they enter the numerator for eleven months and the denominator at half weight. Reported ARPU rose +8% in the same year 15% of the customer base arrived below average. "ARPU" appears 0 / 0 / 5 times across the FY2023 / FY2024 / FY2025 10-Ks: the metric was renamed from "wallet share" and formalised in exactly the year the acquired cohort distorted the count.
op_margin_delta_pp: +1.8, correct for FY2025-vs-FY2024 and the wrong sign for the current trend.