Product teams in Australia and New Zealand have more building capacity than they've ever had, thanks to AI. What they don't have, in most cases, is a fast and reliable way to prove what that building achieved.
That's the headline from theANZ Product Intelligence Index 2026, a study Mixpanel commissioned with YouGov. We surveyed more than 200 product, engineering, data, and growth professionals across the two countries about how AI is changing the way they ship, measure, and decide.
Why the focus on ANZ? The region is home to some of the most innovative product-led companies, including a growing set of Mixpanel customers. We’re increasing our investment in ANZ with a growing local team and an office in Sydney (coming soon). To help us get the lay of the land, we wanted to understand what local product teams are up against right now.
What we heard is a story about a shift in pressure. AI has made it easier to build. The hard part has moved downstream, to knowing what's worth building and whether it worked.
Building got easier. Deciding got harder.
AI is already part of how most ANZ teams ship. More than eight in ten respondents (83%) say their teams release product changes more often than they did a year ago, and 88% say AI tools are a main reason why. Most also believe the extra speed is paying off: 82% say AI has helped their organisation both ship faster and improve customer or business outcomes.
But more capacity doesn’t eliminate pressure; it changes where it sits. Three in 47 respondents (76%) now say deciding what to build takes longer than building it.
That's a striking reversal for any product team. When building was the constraint, a slow roadmap meant a slow release. Now the constraint is judgement, and judgement runs on evidence. Every additional release creates another set of outcomes to interpret and another round of decisions about whether to scale a change, tweak it, or roll it back.
Mixpanel CEO Jen Taylor put it plainly: "Speed alone is no longer the advantage: Knowing what's worth shipping, and why, is."
Confidence is high until someone asks for proof
On the surface, ANZ teams feel good about measurement. We found 85% say their ability to prove impact has kept pace with how fast they release, and 87% trust the numbers in their product reporting.
The picture changes when a specific decision is on the table. Seven in 10 (71%) say their organisation made significant product decisions in the past 12 months without the evidence it wanted. Two-thirds (68%) have been asked for evidence of impact they couldn't produce. And respondents estimate that, on average, just over half (55%) of their releases hit the metric they were meant to move.
ANZ Product Intelligence Index 2026
Confident in general, short on proof when it counts
How ANZ product teams rate their measurement, compared with what happens when a specific decision is on the table.
The report calls this the product impact gap: the distance between shipping a change and being able to show, consistently and credibly, what it achieved. Most teams have trusted dashboards. What they're missing is the right answer for a particular release, decision, or stakeholder question, at the moment it's needed.
Consistency makes it harder. Almost three in four (73%) say different teams report different figures for the same metric, even though 84% say ownership of core metric definitions is clear.
As a global charity running highly seasonal campaigns, we’re under pressure to move fast, and AI helps with that. But faster decisions are only useful if we trust the data behind them. Before we can say an AI-powered initiative is working, we need to know the activity we’re measuring comes from real supporters.”
Prasad Murthi
Digital Analytics, Optimisation, and Personalisation Lead, Movember
Answers still take days, while releases take hours
If a release can go out in an afternoon, the answer to "did it work?" should arrive on a similar timeline. For most ANZ teams, it doesn't.
Only 17% of respondents say their team can typically get a reliable answer to a new question about product behaviour within the same day. Four in 10 (41%) wait one to two days, and a third (32%) wait three to five.
ANZ Product Intelligence Index 2026
Answers still take days, while releases take hours
How long it typically takes ANZ product teams to get a reliable answer to a new question about product behaviour.
17%
get a reliable answer within the same day
41%
typically wait one to two days
32%
typically wait three to five days
Full breakdown of responses
Less than an hour
Later the same day
13%
1–2 days
41%
3–5 days
32%
1–2 weeks
More than 2 weeks
We usually can’t get a reliable answer
Don’t know
Same-day answers (17% combined)
Base: all respondents (n=202). Q: When someone on your team has a new question about how people interact with your product, how long does it typically take to get a reliable answer? Figures may not sum to 100% due to rounding. Source: Mixpanel and YouGov, September 2026.
Asked what holds them back, nearly everyone (96%) named at least one barrier. The most common were the engineering effort needed to set up and maintain tracking (49%), the cost of tools (48%), poor data quality or limited trust in the data (46%), and tools that are too complex for much of the team (45%).
None of these are about whether data exists. They're about whether the right evidence can be produced quickly, and read the same way by everyone, when a decision needs to be made. Otherwise, teams are forced to go backwards: decide first and check later.
Who, or what, is using your product?
AI isn't only changing how ANZ teams build. It's changing what ends up inside their products. Nearly all respondents (96%) say their organisation has shipped at least one AI capability. The most common are AI-generated content or recommendations (64%), AI-powered search (58%), and conversational assistants (54%). Four in 10 (40%) have shipped an AI agent that completes tasks on a user's behalf.
Measuring those features is a mixed experience. About a third (32%) say AI features are harder to measure than conventional ones, while 36% say they're easier. The capabilities being shipped fastest are, for a sizeable minority, the hardest to evaluate.
The bigger blind spot is attribution. Only three in 10 (31%) say their organisation can reliably tell activity generated by human users apart from activity generated by AI agents acting on their behalf. Almost half (48%) can only do it partially, and one in five (19%) can't do it at all, including 8% who haven't considered the question.
That matters because the same chart can mean very different things. A spike in feature usage or completed actions could reflect stronger human engagement, more agent activity, or both. Without a clean split, it's hard to connect product behaviour to real customer outcomes.
When we shipped AI into our own product, we started tagging every action in Mixpanel by who was actually driving it: a person, an agent or the API. The surprise came straight away. Several customers told us they were already using us with AI, and when we jumped on demos, they had AI assistants driving our web app through their browser. For the first time, every product has two users: the person and the agent working for them. You have to run experiments for both.”
Chris Borzillo
Mixpanel CTO Anant Gupta sees this as the next frontier for product measurement: "To make significant product decisions with confidence and clarity, teams increasingly need access to accurate behavioural data that clearly understand human activity and the actions of AI agents."
From velocity to verifiability
AI has raised the ceiling on what ANZ product teams can produce. The findings suggest the teams that benefit most won't be the ones that ship the most. They'll be the ones that shorten the loop from release to evidence to decision to improvement, and run it at the same speed they build.
This is the problem we built Mixpanel to solve, and it's why we're doubling down on ANZ. In the words of our CEO, "That's the gap Mixpanel closes, and the demand we're seeing from local teams is why we're investing in ANZ with a growing team and a new office in Sydney."
We're looking forward to working with more of the region's product teams on what comes next.
About the research: YouGov conducted an online survey on behalf of Mixpanel of 202 professionals employed in Australia (150) and New Zealand (52), working in product, engineering, data and analytics, or growth roles at organisations that build customer-facing digital products. Fieldwork ran from 3 to 10 September 2026. Results are unweighted and reflect respondents' reported experiences.
