Pattern's two-speed quarter: record 129% NRR, a first margin crack, and a 25% round-trip

Q2 2026: revenue +47%, EBITDA +54%, guidance raised again — but Q3 guides to y/y margin contraction for the first time, landing right as the stock gives back a quarter of its 126% run.

PTRN · Earnings Call · 2026-08-05

The NRR machine hit a new gear

Pattern's second quarter as a public company produced a fourth consecutive quarter of 40%+ revenue growth, and the rate of acceleration is what stands out. Net revenue retention hit a record 129%, up from 127% last quarter and 118% a year earlier — against a long-term target of 115%, the machine is clearly outrunning its own assumptions. “In Q2, NRR reached another record at 129%. Up from 127% last quarter, 118% a year ago.” — David Wright, Unknown · 2026-08-05 Jason Beesley walked through the three growth drivers — technology-led optimization, marketplace/geography expansion, and product-line expansion — with international revenue crossing $110M for the first time (up 87%) and non-Amazon revenue growing 93%. The fundamentals corroborate the inflection: the latest 10-Q shows net income of $29M and, more tellingly, free cash flow ex-stock-compensation of $55M, up 28% y/y — a sharp reversal, and consistent with the 92% trailing FCF growth Jason cited. Even a year ago, the company described its NRR as elite: “even a 3-year average going forward would put us an elite company on this metric” — Jason Beesley, Chief Financial Officer (CFO) · 2025-11-07. It's worth underscoring how company-unique this is: the global keyword tape for the same quarter is dominated by tariff refunds, ERCOT "Batch Zero," and retail noise — nothing about data moats or agentic commerce. Pattern is riding its own wave.

Agentic is now the operating thesis, not a roadmap slide

The most genuinely new material in the quarter isn't the financial beat — it's the productization of the agentic story. Pattern Intelligence (PI) launched in May and is running a sensor-actor framework writing "millions a day" of governed actions; the company was awarded a US patent covering True ROAS; and via the ROI Hunter acquisition, brand partners can now advertise directly inside ChatGPT. David Wright framed the data advantage in AB testing terms: “the data and the data moat that we have is invaluable at this point and is continuing to grow... it just keeps compounding” — David Wright, Unknown · 2026-08-05, citing 91 trillion data points, up 38% year to date. When pressed on PI's near-term economics, management pointed to software-factory productivity rather than direct monetization — “a unit of work... for a developer would be a pull request... we have doubled those units year to date” — David Wright, Unknown · 2026-08-05. The strategic framing — LLM surfaces on the discovery side and agentic transactions that still require fulfillment — is a deliberate attempt to position the company for both outcomes. This vocabulary is fresh: commerce infrastructure, "agentic surfaces," and "surface area" all appear high in the company's own 20263 keyword trajectory and were absent a year ago. The risk: this is the quarter where the forward narrative outran the deliverable — and the tape appears to agree.

The first crack in the leverage story

Full-year guidance was raised again to $3.4–3.5B (37–38% growth) and $211–213M of adjusted EBITDA (38–40% growth), and Q2 itself printed EBITDA growth (54%) ahead of revenue (47%) for the fourth straight quarter. But the Q3 guide implies the opposite: 31–34% revenue growth against 25–29% EBITDA growth — a year-over-year margin contraction for the first time since the public history began. Bernie McTernan pressed this directly, and Jason Beesley's answer deserves the full frame:

That causes some noise between Q2 and Q3 on both the revenue and the EBITDA side... When you control for that, both the growth rate in Q3 and the margin rate in Q3 make a lot more sense. And then the only real drag year over year is... increased investment in R&D faster than revenue.
Jason Beesley, Unknown · 2026-08-05

The "noise" is a ~4-point calendar shift — Amazon Prime Day, Walmart Deals, and Target Circle moved into Q2 this year. But the structural drag is real: the full-year call for margin accretion now leans on a holiday quarter Jason himself flagged for "seasonal pressure," and R&D investment is deliberately running ahead of revenue. Tellingly, a year ago the company guided NRR to “existing brand partner NRR converging to 115% by the end of the year, generally as our long-term target” — Jason Beesley, CFO · 2026-03-20; today it says the 2026 exit will be 123–124%. The direction of travel (moderation) is now the same — the question is whether that plus a margin-drag quarter is the start of normalization or a pause in the compounding story.

Turnover test: Thorne to P&G — and a 25% round-trip

The quarter's only genuine corporate event was external: P&G's announced acquisition of Thorne, a long-standing Pattern brand. The market immediately worried about change of control. “There is no change of control provision in our agreement... we have got runway to demonstrate our value to the partnership going forward.” — Jason Beesley, Unknown · 2026-08-05 David Wright's framing was the more strategic one — serial exits are, for Pattern, actually a feature: “some of our best and longest partnerships are inside those CPG conglomerates” — David Wright, Unknown · 2026-08-05. The keyword change of control appearing at rank ~14 in this quarter's trajectory is the market's flag — this is the new variable investors are checking. It's also the right moment to weight the tape: PTRN ran +126% in twelve weeks into a 7/9 peak, then gave back 25% of that. The report arrives with the stock already in a drawdown — hot fundamentals against a guide that, for the first time, whispers deceleration. Both can be true: the record NRR and deepening PI moat say the model works, channel shift is the strategy, and the marginal buyer will now demand that the price earn re-acceleration rather than beat-and-raise on prior running-room.