Built on Traffic

A page view is traffic. A phone call is traffic. A chat message is traffic. Why does every analytics tool pretend they're different things?

Every campaign you run — Google Ads, social, email, a podcast mention — has one job: drive people toward your product. That traffic used to land on a web page. Now it lands on a web page, triggers a voice call, or starts a chat session. The intent is the same. The channels multiplied. But analytics is still measuring each lane of the highway separately and never counting the total. You can optimize individual lanes. You can't see the road.

That's the gap. Not a tooling gap — a definition gap. The first principle of analytics hasn't changed: what is happening with our traffic? The definition of traffic has.

One number, not three

Inside Sentrix we run voice agents, chatbots, and web properties across dozens of projects. Each channel has excellent observability on its own — LangFuse for AI chains, provider dashboards for call metrics, GA4 for the website. None of these tools are broken. The problem is simpler than that: none of them can give you one number for total traffic volume across everything, broken down by source.

When you run an ad campaign and web traffic goes up 20% but call volume goes up 40%, that's a single insight that lives in two separate dashboards. When organic search drives 60% of web visits but only 15% of voice calls, that's a resource allocation decision hiding across three tools. These are basic traffic questions. They just need all channels in the same pipeline to be answerable.

That's what Metrone is. One schema, one pipeline, one dashboard — web events, voice calls, chat sessions, all of it. Not because the specialized tools are inadequate. Because traffic is the abstraction that connects them, and no one was counting it.

The metrics are already obvious

Once you accept that a voice call is traffic, the metrics you need aren't new. They're the same ones web analytics has always tracked — applied to channels that didn't exist when those metrics were invented.

Completion rate is conversion rate. On the web, it's how many visits result in a goal completion. For a voice agent, it's how many calls resolved the caller's intent. Not "the call connected" — that's a telephony metric. Resolution is a traffic metric. Our VAPI agent had a 95% connection rate and a 60% resolution rate. Same channel, very different story depending on which number you treat as the real one.

Intent distribution is top pages. On the web, you look at which pages get the most visits to understand what your audience wants. For AI channels, you look at which intents get the most volume — "pricing inquiry" at 40%, "scheduling" at 25%, "support" at 20%. Same question, different surface.

Escalation rate is bounce rate. When a voice agent hands off to a human, the channel didn't fully serve the visitor. A 12% overall rate might look fine — until you break it down by intent and discover "billing inquiry" is at 45% while everything else is under 10%. That's the same insight as one landing page with a 90% bounce rate. The traffic is there. The channel isn't converting it.

Response latency is page load time. Web analytics has always cared about speed because slow pages lose visitors. Voice has the same dynamic — above 3 seconds to first response, callers start talking over the agent, and the conversation breaks down. Our P95 was 4.2 seconds during peak hours while the average sat at a comfortable 1.8s. Averages lie. The percentile is where the traffic leaks.

And then there's the one that only becomes visible when all traffic is in one place: channel correlation. Weeks with higher resolved voice call volume consistently correlate with higher web conversion rates in our data. We can't prove causation — the events come from different channels. But the pattern is there, in the same time series, the same chart. Try spotting that when your voice data is in Twilio and your web data is in GA4.

The definition expanded

For twenty years, "traffic" meant web visits. That was the whole surface area. Analytics tools were built for that world, and they're very good at it.

The surface area expanded. Every voice agent, chatbot, and copilot your team ships is a new channel where traffic flows — from the same campaigns, the same referrers, the same audience. The first principle didn't change. Traffic is still the thing you measure. The definition just grew past what single-channel tools can see.

Metrone is a unified traffic analytics layer. One schema for every channel. One pipeline for every event. One source of truth for the total picture. Because traffic is traffic, regardless of which channel carries it. The analytics layer should reflect that.