In one paragraph

Growth teams ran fewer experiments in 2025 (63% do just 2-4 per month) yet reached further up-funnel into acquisition and doubled-down on monetization—all while shipping AI at record pace (53% launched a customer-facing feature).

Organizational models splintered from “centralized vs de-centralized” into hybrid guilds, and ambition to be “very product-led” keeps inching up (+0.6 YOY).

The paradox: teams crave faster learning loops but allocate less capacity to run them—setting up 2026 as a make-or-break year for experimentation culture.

Only 7% of growth teams now run more than ten experiments each month—down from 18% last year—while a majority shipped at least one customer-facing AI feature.

“Growth alone can’t fix a lagging product—experiments are only as good as what you build next.”
— Product Growth Lead, Dropbox

I. Org Models Are Fragmenting

A neat centralized vs decentralized debate no longer captures reality. This year only 37% of respondents still run a fully centralized-growth org; nearly half operate hybrid “growth guilds” or have no formal team at all, echoing broader product-team trends towards hybrid structures.

“Basics, basics, basics—we can’t chase trends if the core isn’t rock-solid.”
— Product Growth, Box

II. Funnel Ownership Shifts Earlier —and Later

Acquisition (73%) has leap-frogged activation to become the stage most commonly owned by growth teams, with monetization/expansion close behind (70%). The pivot mirrors B2B SaaS advice to lower CAC while squeezing additional ARR from an existing base.

“Prioritization should come from product insight, not HIPPOs.”
— Growth Manager, Chegg

III. AI Moves From Tool to Product

Fifty-three percent of respondents shipped at least one customer-facing AI release last year—almost identical to the share of executives who rank generative AI among their top three strategic priorities. McKinsey 2025 AI Report. Internally, AI is most often applied to experiment ideation (67%) and content creation (50%).

“2025 bets solve the jobs customers dream about—not the ones we planned.”
— Director of Product, Faire

IV. The Experiment Paradox

Teams rate their product-led maturity at 5.1 / 10 today and expect 5.7 within a year, yet 63% cap out at four experiments per month. Research continues to link disciplined experimentation with sustainable product-led growth. Amplitude

“Simple fixes beat attention-grabbing band-aids.”
— Product Growth, LaunchDarkly

V. Action Playbook for 2H 2025

• Thin engineering bandwidth → Run premortems and use paint-by-numbers experiment templates to cut setup time.
• AI feature hype → Tie every AI launch to a clear activation or monetization KPI; 71% of CFOs say monetizing AI remains a struggle. TechRadar
• Fragmented teams → Hold a monthly growth guild to share wins and avoid duplicate tests.

Ninety-one percent of SaaS companies still plan to increase investment in product-led growth over the next year, heightening the competitive value of faster learning loops. VentureBeat

VI. Methodology & Acknowledgments

• Sample: 30 practitioners across SaaS, marketplace, and infrastructure sectors (Q1 2025).
• Margin of error: ±18%.
• Data handling: All PII removed; single-company answers anonymized.

Thanks to peers at Dropbox, Box, LaunchDarkly, Faire, Chegg and many others for candid insights on product growth in 2025.