Measure the workflow, not the technology
Messaging and voice automation fail in one of two ways: they never get adopted because nobody can tell whether they are working, or they run for months without a baseline to compare against. A small measurement framework fixes both. This article defines operational metrics for WhatsApp conversations, campaigns and AI calling — a framework you can run with the data your workspace already produces.
A framework, not a performance report
No invented benchmarks, no customer numbers, no fabricated dashboards. Define the metrics, measure your own workflow, and compare your own before/after.
WhatsApp messaging metrics
For day-to-day conversations, the useful metrics describe how quickly and how well the team responds: first-response time, resolution time, the share of conversations escalated to a person, and the share resolved by automation where chatbots are in the flow. Each one is a number you can define, measure and trend.
| Metric | What it shows | Watch out for |
|---|---|---|
| First-response time | How fast a customer gets an answer | Chasing speed over usefulness |
| Resolution time | How long until the conversation closes | Closed conversations can still mean unsatisfied customers |
| Escalation rate | How often a human must take over | A spike is often an intent or template problem |
| Automation resolution rate | How much routine work the bot absorbs | Only meaningful when escalation still works |
None of these require special tooling — conversation history and delivery states in the workspace are enough to compute them for a fixed period and compare periods.
Campaign metrics
For campaigns, measure the full path from send to outcome: delivery rate, reply rate, opt-outs, and the business result the campaign was built for (appointments booked, forms completed, orders confirmed). An opt-out spike is the earliest warning that frequency or relevance is wrong — treat it as signal, not noise.
- Delivery rate: how many sends actually reached customers.
- Reply rate: how many recipients engaged with the message.
- Opt-out rate: how many recipients unsubscribed after the send.
- Outcome rate: how many reached the campaign's defined goal.
AI calling metrics
For AI calling, the core questions are about the workflow: how many calls were handled, how many reached their defined success condition (a booking, a qualification, a resolved question), how many escalated to a human, and how long callers waited before that happened. Transcripts and call outcomes in the workspace make these measurable per agent and per period.
| Metric | What it shows | Watch out for |
|---|---|---|
| Calls handled | Volume the workflow absorbs | Volume without outcome tells you little |
| Resolution rate | Share reaching the defined success condition | Define success before launch, not after |
| Escalation rate | How often a human takes over | Expected for sensitive intents; a problem if it is everywhere |
| Abandonment | Callers who give up | Compare against your pre-automation baseline |
A simple cross-channel funnel
Contact arrives
A WhatsApp message or an inbound call.
Conversation starts
The thread or call is opened in the workspace.
Resolution attempted
An agent, a chatbot or a voice agent works the case.
Outcome
The defined success condition is reached — or the case escalates.
Review
Weekly metrics review drives the next change.
The same funnel shape works for both channels, which is the point of a platform that keeps messaging and voice in one workspace: the metrics compare like for like.
A review cadence that sticks
- Weekly: volume, response times, escalations, opt-outs — a ten-minute check.
- Monthly: trends, campaign outcomes, automation resolution rate.
- Quarterly: the workflow-level questions — is this still the right channel for this workflow?
The cadence matters more than the dashboard. A simple weekly review of six numbers will improve the operation more than a polished report nobody reads.
Run the framework on your own workspace
Messaging, campaigns and voice calls with transcripts and outcomes in one place — ready to measure.
Talk through your metrics with us
A call-flow and measurement walkthrough for one of your workflows, before you launch.
Frequently asked
Do I need analytics software to measure this?
No. Conversation history, delivery states, transcripts and call outcomes in the workspace are enough to compute the framework's metrics for a fixed period and compare.
What if my escalation rate is high?
Start with the intent and template setup: a spike is usually a sign that an intent is mislabelled or a template is confusing. Escalation is not failure — unmeasured escalation is.
Is there a target percentage for any of these metrics?
No. Benchmarks without your workflow data are meaningless; define success for your own operation, measure it, and trend it.
Does this framework measure the public website too?
No. This article is about the workspace operation. Site-side analytics (GA4) is a separate, operator-controlled surface.