In-Product Analytics (PostHog)

Voiso is a CCaaS platform processing tens of millions of calls a month. The product had rich operational reporting for customers, call detail records, historical reports, real-time dashboards, but the behavioural analytics layer didn't exist yet.

Product analytics at this scale carries a real tension: the more you instrument, the more you risk both runaway event cost and privacy exposure, yet under-instrument and you are blind. At tens of millions of calls a month, a naive per-call tracking approach would have buried the signal and blown past cost limits, and a CCaaS platform handles sensitive customer conversation data that must never leak into an analytics tool. The mandate I took on, without being assigned it, was to stand up a product-analytics practice from zero: pick the approach, set the rules, and make it safe enough and disciplined enough that the rest of the org could build on it.

I drove this initiative from the front, as the sole owner. I set the strategy, wrote the tracking plan, defined the metrics, documented the event taxonomy, handed the spec to engineering for implementation, and then built the dashboards and insights in PostHog myself. Engineering owned event firing in the product. Everything from strategy to the finished measurement layer was mine.

The guiding principle: observe only what matters. Track the high-signal workflows, not the noise that's hard to formulate into a clear metric. The goal was never a wall of charts. It was a well defined set of numbers a domain team could open and immediately see where the experience was failing.

• Introduced a product analytics culture into the organization.
• Enabled teams to validate decisions using behavioral data rather than assumptions.
• Improved visibility into user journeys, adoption patterns, and product performance.
• Created a foundation for experimentation and data-informed product development.