Social customer care is measured honestly by three linked metrics: first response time, resolution rate and customer satisfaction, with volume and cost per contact as context rather than headlines. The platforms themselves hard-code urgency — Meta's messaging infrastructure closes the standard customer conversation window 24 hours after a user's last message, per the company's Business Help Center (2024), which means a slow reply does not just disappoint, it can lock the thread. Teams that report only response time are measuring the greeting; teams that report only CSAT are measuring the afterglow. The three metrics only mean something together.
Why First Response Time Is the Anchor Metric?
First response time (FRT) measures the gap between a customer's message and the first substantive human or system reply — an acknowledgment that work has started, not necessarily a solution. It is the anchor because it is the customer's most vivid memory of the interaction and the easiest metric to benchmark internally: median FRT per channel, per shift, per language. The measurement rules matter. A bot auto-reply should not count as first response unless it actually resolves or routes the issue, and business-hours versus 24/7 coverage must be declared, because an overnight median and an always-on median are not comparable numbers.
Targets should be set per channel and per tier. Public comments and complaints carry reputational urgency that DMs do not; a same-business-day reply that would be fine in email is slow in a public thread. Median FRT is the reporting figure; the 90th percentile catches the worst experiences that averages hide.
What Does Resolution Rate Actually Capture?
Resolution rate is the share of inbound issues closed within the social channel without escalation to another system — no "please email us," no ticket handoff the customer must chase. It is the metric that distinguishes care from deflection. A team can post an excellent FRT while resolving a minority of issues, because fast acknowledgments are cheap and fixes are not. The definition must be written down: what counts as resolved (customer confirmed, agent marked, or auto-closed after inactivity), and within what window. Auto-closure inflates the rate and should be reported separately so the inflation is visible.
Time to resolution (TTR) adds the second dimension — median elapsed time from first message to confirmed close. Together, resolution rate and TTR describe whether the operation finishes what the response time starts. If resolution rate is high but TTR runs to days, the bottleneck is rarely the social team; it is the internal process the social team is queueing into.
How Should CSAT Be Measured in Social Channels?
Customer satisfaction in social care is measured with a single post-interaction survey question — a rating request delivered in-thread or immediately after closure — with a response scale the team keeps constant. The two discipline points are sample and timing. Not every interaction should be surveyed, because survey fatigue depresses response quality; a sampled subset is enough. And the survey must fire at resolution, not at the agent's convenience, or the score drifts toward the easiest cases. CSAT is a lagging indicator and a blunt one: it tells the team how the experience felt, not why. Pairing it with a theme code from the interaction record turns the score into something a manager can act on.
Related stories: How to Measure an Employee Advocacy Program: Participation, Reach and Conversion · A Social Media Crisis Response Playbook for Brands: Stages, Roles and Escalation.
How Do the Metrics Fit Together?
The three metrics form a funnel with distinct failure signatures.
| Metric | Question it answers | Common failure mode |
|---|---|---|
| First response time | How fast do we acknowledge? | Bot replies counted as responses |
| Resolution rate | How often do we finish in-channel? | Auto-closure inflating the rate |
| Time to resolution | How long until finished? | Hidden queues outside the social team |
| CSAT | How did it feel? | Surveying only easy cases |
| Cost per contact | What does it cost? | Optimizing cost into worse service |
A dashboard that shows FRT improving while resolution rate falls is describing deflection, not care. CSAT falling while FRT improves usually means responses are fast and wrong. Reading the metrics as a set is the entire skill.
What Is a Sensible Measurement Procedure?
A care program that wants comparable numbers month over month runs a fixed routine.
1. Define each metric in writing, including what counts as a response, a resolution and a surveyed interaction.
2. Tag every inbound item with channel, language, issue theme and product line at intake.
3. Report medians and 90th percentiles, never averages — social care distributions have long tails that means misrepresent.
4. Segment by channel and shift before drawing conclusions; a blended FRT hides that weekends are unstaffed.
5. Reconcile social metrics with the contact-center stack monthly, because double-counted contacts distort both sides.
6. Review theme codes quarterly and route the top three recurring themes to product or operations owners, which is where care data starts preventing contacts instead of handling them.
What Role Does Escalation Play in the Numbers?
Escalation is a metric boundary, and it must be drawn deliberately. Every social care operation has issues it cannot close in-thread — refunds above a threshold, account security events, legal exposure — and each of those moves the interaction out of the resolution-rate numerator. The fix is not to stop escalating; it is to measure the handoff itself: what share of escalations receive a named owner within the service window, and what share of escalated customers come back to social channels to complain about the handoff. That second figure is the clearest early-warning signal a care program has, because it detects broken internal processes while the customer is still willing to say so publicly.
What About Automation and AI Assistants?
Automation changes the accounting, not the metrics. A chatbot that deflects password resets genuinely improves resolution rate and cost per contact; a bot that frustrates customers into leaving improves nothing but the FRT dashboard. The honest approach is to segment bot-handled and human-handled threads and report each stream separately, with CSAT sampled in both. When vendors demonstrate AI resolution rates, the platform says the numbers are high — the team's own segmented data is the only version worth budget decisions.
How Do Benchmarks Work in Social Care?
External benchmarks exist — contact-center and social-management vendors publish FRT and CSAT ranges by industry — but the same caution as any vendor sample applies: composition skews toward customers of the tool, and definitions differ. The defensible target is a stretch against the team's own trailing quarter, with vendor ranges used to check plausibility. A team moving median FRT from six hours to two and lifting resolution rate ten points has done more for customers than a team chasing an industry number it cannot decompose. Benchmarks describe other people's trade-offs; the metrics stack describes your own.