Does Product Ops have a Data Problem? And will AI force us to finally fix it?
This has been on my mind for quite a while now.
Product Operations, in JDs and conference slides, sounds genuinely exciting when it comes to data. Connecting dots across the portfolio. Surfacing patterns in customer feedback. Shaping product strategy rather than just reporting on it.
The reality for most of us: spending 80% of our time getting data into a usable state and 20% doing anything meaningful with it.
Not a new observation. But we can no longer just get on with it.
The infrastructure gap is the real bottleneck.
The tactical layer is well understood. Build dashboards. Aggregate feedback. Centralise metrics. All necessary, all valuable, and all harder than they sound.
Feedback is scattered across a dozen systems. Data is dirty, incomplete, or buried so deep it becomes a dark art. Product Ops professionals with genuine analytical capability end up spending their time on plumbing.
AI changes the conversation, but not how most people think.
The AI narrative focuses on automation: AI will build your dashboards, summarise your feedback, and generate your reports with the advent of MCPs and no-code API building. I’m doing this now.
But the more interesting question is what happens after the automation.
If AI handles the mechanical work: aggregation, summarisation, pattern recognition, then what is left for Product Ops?
Judgment. Context. Interpretation.
An AI tool can tell you 340 customers mentioned “onboarding” in negative-sentiment feedback last quarter. It cannot tell you which signals matter most given your strategy, which are symptoms of a deeper problem, and which are noise. It cannot navigate the conversation with a PM who does not want to hear that their onboarding flow is broken.
AI shifts Product Ops from a data collection function to a data interpretation function.
The questions I am left with.
The tools are changing faster than our practices. The potential is outpacing the infrastructure.
How do we build data foundations that make strategic insight possible, rather than perpetually firefighting the plumbing? How do we use AI to amplify judgment without outsourcing it? How do we ensure customer voice informs product decisions rather than just decorating them?
I do not have all the answers. But these are the right questions…
And so when Mixpanel reached out to see if I’d like to attend MXP London, this seemed like a perfectly timed opportunity to get some answers. Excited to hear again from speaker Debbie McMahon about the AI adoption gap; an incredibly relevant topic in stitching together the narrative of data usage by humans, by AI and what comes next…
So here I am herding other product and data professionals to join me on May 21 in London to answer these questions together, along with 300 other leaders in our field.
Want to join me?
Graham