the missing layer in product analytics
The issue with most product analytic tools is that they model the world as a series of disembodied streams of events, that either make a graph go up or down on a dashboard somewhere. The burden is then placed on PMs/dev to breathe life into these charts and translate them to something remotely useful. These events lack intent, outcome, and economic consequence, rendering them quite useless when trying to decipher what on earth happened.
Here’s what I think, the next era is revenue-first analytics that natively understand why users behave in a certain way and what it costs or gains.
The problem isn’t the events themselves, they’re just primitive, unopinionated facts with some utility; a click or page view does tell you something about what happened. But the real problem is what you’re left to deal with: you’re staring at a dead pile of event data and expected to bring it to life. Was this friction? Did the user abandon it, or were they simply exploring? And more importantly; what does any of this mean for the product, for revenue, or for what we should fix next? When there’s no obvious bug, PMs are basically going off vibes and hoping they’re right.
The absence of a coherent model built on top of events that assembles them into the high-order concepts that are considered when decisions are being made is the missing piece; intent, outcome, revenue impact.
Events are only as useful as the decision semantics they’re contextualised in; what was the user trying to do, what did they did, what was the revenue impact. Once you know that, figuring out what’s working and what’s not stops being an investigation.
What do WAU, feature usage, and session length actually tell us, no really? What story does it paint about our product and how users are interacting with it? Because WAU can increase because a bunch of free trial users are bouncing around with zero intent on converting. Sign ups can also go up, but if activation isn’t increasing, it’ll be a couple of months before these users churn.
While not entirely useless, these metrics are still proxies to what we really care about - they’re rough signals about whether people find our product valuable, whether they’ll stick around, and ultimately whether they might eventually or continue to pay. But they’re several steps removed from actual revenue impact, and a lot of the time nothing but vanity metrics.
So how do you do revenue attribution accurately? It’s difficult and not that straightforward, but I’m certain that it can’t be done accurately at event level (for the most part). Some events do act as strong litmus tests i.e, completing onboarding, activating, or inviting team mates, but for the most part, revenue attribution is more accurate when it’s a byproduct of a combination of behaviours; completing a workflow, repeating it, inviting others into it, and doing it consistently overtime. Until we can model the multi-variable patterns that actually lead to value creation, we’re stuck trying to connect atoms.
This is where I think product analytics is bleeding into customer success, but there’s still data lost in translations; billing teams think in MRR and lifecycle, CRMs in accounts and renewals, and product still use WAU, feature usage to try and reverse-engineer measure business impact. Once all of these teams operate off the same revenue model, where every meaningful action in the product maps to a real revenue impact, retention stops being guesswork and starts being a science.
So what would this actually look like? A tool that doesn’t just track events, but understands them in the context of accounts, workflows, and revenue. One that bridges the gap between what users do and what it means for the business. Where product, customer success, and billing teams can finally speak the same language because they’re looking at the same model.
This is exactly what I’m trying to do with Onbored. Onbored is an open-source revenue-first product analytics tool that treats accounts as first-class-citizens.