The 988 Suicide & Crisis Lifeline is one of the clearest public examples of multi-channel contact reporting. SAMHSA publishes monthly counts for calls, texts, and chats, plus answer rate and average speed of answer. The important lesson for a business is the reporting design—not a direct performance comparison with a crisis network.

Keep channels and denominators separate

A combined “contacts handled” number can conceal a weak phone queue or a strong text queue. 988 performance tables separate channels because their arrival patterns and handling processes differ. A business should do the same for voice, SMS, web chat, and callbacks.

For every month, retain:

  • offered contacts by channel;
  • answered contacts by channel;
  • abandoned contacts and the chosen short-abandon rule;
  • average and percentile answer time;
  • transfers and final disposition;
  • staffing hours aligned to the same intervals.

Average speed alone is fragile. Ten near-instant answers can mask one very long wait. Report the median, 90th percentile, and percentage answered within a defined window. Keep the raw count beside every percentage.

Do not import a crisis benchmark

The 988 network has a public-safety mission, federal and state funding, specialized counselors, routing rules, and demand that differs sharply from sales or appointment calls. Its monthly results also change over time. Therefore, a business should not claim that a 988 answer rate is an appropriate commercial target.

Instead, borrow three measurement disciplines: publish the period, separate channel cohorts, and explain routing changes. FCC work on georouting illustrates why the route itself matters: the place associated with a phone number may not be the caller's physical location. Commercial teams face a milder version when area code, branch, language, and time zone affect routing.

Run an interval-based test

Choose four comparable weeks. Export arrivals in 30-minute intervals, record staffing and routing configuration, and identify calls that overflowed. Change one rule—such as overflow after 20 seconds—and repeat across four more comparable weeks. Compare answer distribution, abandonment, transfer rate, and the business outcome attached to completed calls.

Seasonality, marketing campaigns, outages, and staff learning can confound the result. A before-and-after comparison supports operational learning, not a causal claim by itself.

Businesses building an escalation tree can consult our customer support escalation workflow research. For staffed coverage options, see AnsweringServiceStaff services.

FAQ

Is 988 a commercial answering service benchmark?

No. Its mission, workforce, routing, callers, and funding differ. Its public measurement structure is useful; its numeric performance is not automatically transferable.

Why separate phone, text, and chat?

Each channel has different arrival, queue, and handling behavior. Combining them can hide the operational cause of a result.

Which answer-speed statistic should be reported?

Use a distribution: median, 90th percentile, and percentage within a documented target, each with its count and period.