240,184,209 messages delivered this month · 4,218 active workspaces across 38 markets · Median template approval · 18m · ▲ +34% revenue lift · 90 days post-onboarding · Now live: AI Flow Generation
240,184,209 messages delivered this month · 4,218 active workspaces across 38 markets · Median template approval · 18m · ▲ +34% revenue lift · 90 days post-onboarding · Now live: AI Flow Generation
Whaterakt
Home Features Pricing Blog About Contact
Start free Sign in
Team Inbox & Support

WhatsApp Analytics: 12 Metrics That Actually Matter

A measurement framework for WhatsApp business channels: 12 metrics across response, resolution, quality and revenue — with definitions, hedged healthy ranges and the weekly rituals that turn numbers into fixes.

Whaterakt Team · Jul 17, 2026 · 8 min read

Most teams judge their WhatsApp channel by a pile of grey delivery ticks. That measures the platform, not your business. Real WhatsApp analytics lives in three layers — delivery data from Meta, conversation events from your inbox, and business outcomes from your CRM — and this framework pulls out the twelve numbers worth putting on a screen in front of your team.

TL;DR

  • Group your metrics into four families: response, resolution, quality and revenue. Each answers a different management question, and you need all four.
  • The ranges below are directional starting points, not industry law — set a baseline for your own channel, then beat it month over month.
  • Delivery rate, contact count and messages sent are vanity; stop reporting them on their own.
  • Run a 15-minute Monday scorecard and a Friday retro — metrics only matter if they change Monday's behaviour.

Where the numbers come from

Meta provides per-conversation statistics — sent, delivered, read — and billing breakdowns by conversation category in its dashboards and API reference. What it cannot tell you is who replied, how long resolution took, or whether the chat ended in a sale. Those come from your tooling layer: a shared inbox that timestamps first response and resolution, tags conversations by intent, and pushes outcomes to a CRM. Whaterakt's analytics sit on exactly those events — per-agent response times, resolution rates and campaign performance in one place — because platform statistics alone cannot answer management questions. One definitional trap to settle early: decide whether a "conversation" means a thread, a customer, or a 24-hour session — Meta bills per session, but your team experiences per customer. Whichever definition you choose, keep it consistent across dashboards, or your weekly comparison quietly ends up comparing different things. For market-level context, our WhatsApp marketing statistics roundup covers what the industry reports; this post is about measuring your own channel.

Response metrics: are we fast?

The first family measures the only currency a customer notices before quality: speed.

MetricDefinitionDirectional targetWhy it matters
1. First response timeCustomer's message to first useful replyUnder 5 min in live hoursSlow first replies kill threads before they start
2. SLA hit rateShare of conversations answered within target90%+ over timeTurns "we're fast" into a testable claim
3. Reply coverageShare of inbound threads that got any replyEffectively 100%Unanswered chats are silent churn

Track the median, not the average — one 3-hour outlier will distort a small team's mean and hide the fact that most chats are answered in ninety seconds. And measure first response separately for bot and human replies; customers forgive a short bot acknowledgement, not a silent hour.

Resolution metrics: are we useful?

Speed of acknowledgement matters little if the underlying problem then ages for days.

MetricDefinitionDirectional targetWhy it matters
4. Time to resolutionFirst message to marked resolvedHours, not daysThe delay customers actually remember
5. Conversations per agentResolved conversations per agent per shiftYour own trendStaffing and coaching baseline
6. Repeat-contact rateResolved threads reopened within 7 daysLow single digitsCatches "closed" rather than "solved"

Conversations per agent is the one metric here you should never benchmark externally — product complexity, price point and queue mix make cross-industry numbers meaningless. Its value is as a trend line for your own staffing maths, and as a fairness check when you compare workloads across a team.

Quality metrics: do they like us?

Quality is where WhatsApp's informality cuts both ways — fast and sloppy is worse than slow and careful.

MetricDefinitionDirectional targetWhy it matters
7. CSATPost-resolution 1–5 ratingMid-80s%+ happyLeading indicator of retention and referrals
8. Opt-out rateMarketing recipients who stop, per sendLow and flatYour permission to keep messaging at all
9. Read rateDelivered messages that were readCommonly high vs emailTests creative and timing fit

Two caveats. Read receipts are approximate — privacy settings and batching distort them — so treat read rate as a relative signal between campaigns, not an absolute truth. And opt-outs deserve more respect than etiquette: sustained high opt-out and block rates feed your number's quality rating on the platform side, which can throttle reach precisely when you need it. CSAT carries its own bias — unhappy customers answer surveys more readily than satisfied ones, so silence is not neutral. Read response rates alongside scores: a 4.9 average from 8 percent of customers tells a different story than a 4.5 from 60 percent.

Revenue metrics: does it pay?

The family that justifies the budget line.

MetricDefinitionDirectional targetWhy it matters
10. CTA click rateClicks on link or button per message sentVaries by offerFirst filter of creative quality
11. Conversion rateConversations reaching the defined goalYour own baselineThe bridge from chat to pipeline
12. Revenue per conversationAttributed revenue divided by conversationsTrend upwardThe number your CFO actually wants

Attribution does not need to be perfect to be useful. Unique links with UTM tags, WhatsApp-only coupon codes and CRM stage changes each capture part of the truth; triangulate two or three and accept some undercounting. Our WhatsApp CRM ROI guide goes deeper on building the model, and the broadcast campaign guide covers measurement design at the campaign level.

Vanity metrics to stop reporting

  • Contact count. A million unengaged numbers is a liability — every marketing send to them burns conversation fees and reputation.
  • Messages sent. Measures effort, not outcomes; it quietly rewards spam.
  • Delivery rate alone. Delivered is not read and read is not acted on; only interesting when it suddenly drops.
  • Green tick status. Verification is credibility, not performance — it belongs in a branding update, not the metrics review.
  • Agent busy-ness. Messages handled per hour without resolution and CSAT alongside it rewards speed for speed's sake.

From metric to fix: a worked example

Say your Monday scorecard shows first response time sliding from three minutes to eleven over a fortnight. The vanity dashboard says nothing is wrong — delivery sits at 99 percent. The diagnostic path: check the distribution (is it everyone, or one shift?), check the queue mix (did a campaign start going out on Monday mornings?), and check the away flow (are overnight chats waiting for the 9 a.m. opener?). In this case the fix was routing — keyword rules sending order-status questions to a self-serve flow — and response time recovered the following week. That loop is the whole discipline: a metric moves, you form a hypothesis, you check a distribution, you ship exactly one fix. Reports that end at the number are trivia; reports that end at a fix are management.

Turn the dashboard into weekly rituals

Monday: the 15-minute scorecard

One screen, last week versus the prior week, all twelve metrics. Pick the single worst mover, name one fix, and give it an owner — not five fixes with no owner. If two metrics tie for worst, pick the one upstream: response problems eventually contaminate everything below them.

Wednesday: the SLA pulse

A two-minute check on the response family only, because it degrades fastest when a queue slips. If SLA hit rate has dipped below the mid-80s percent, act midweek instead of discovering it the following Monday.

Friday: the conversation retro

Pull the three lowest-CSAT and three longest-resolution threads and read them. Almost every systemic fix — a missing canned reply, a routing gap, an authority limit that forces escalation — shows up here first.

Monthly: the deep dive

Cohort the revenue metrics: which segments, campaigns and agent approaches produce the best revenue per conversation? Feed the winners into next month's support playbook and campaign calendar.

Finally, assign one owner per family — someone accountable for response, for resolution, for quality, for revenue — because metrics owned by everyone are owned by no one.

FAQ

Are Meta's own WhatsApp statistics enough?

They are a necessary floor: conversation counts, category splits, delivery and read data. They cannot see agent-level performance, resolution or revenue, which is where management decisions actually live. Layer your inbox events on top.

What is a good CSAT score on WhatsApp?

Directionally, well-run channels commonly report happy responses in the mid-80s percent or better. More useful than the absolute number is the trend, plus reading the verbatims behind every 1–2 rating.

How do I attribute revenue to WhatsApp?

Pragmatically, not perfectly. Unique tracked links, WhatsApp-specific coupon codes, a "how did you hear about us" capture at checkout, and CRM stage attribution each do part of the job. Triangulate two or three and accept some undercounting.

Which single metric should I start with?

First response time, paired with SLA hit rate. They are cheap to capture, easy for agents to feel ownership of, and the fastest leading indicator of everything downstream — resolution, CSAT and eventually revenue.

How often should we review WhatsApp analytics?

Weekly for the operational families, monthly for revenue. Real-time dashboards are seductive but mostly produce anxiety; rituals beat staring.

Twelve numbers, four families, three rituals — that is a measurement framework you can run in under an hour a week and defend in front of any stakeholder. Start with the response table, then add revenue attribution once the operations layer is stable. If you want the events pre-wired — per-agent response times, resolution tracking, campaign analytics — explore Whaterakt features and connect your number to see a real baseline within a week.

#analytics #metrics #kpi #csat #reporting

Keep reading

All posts →
Back to all posts