AHT, connect rate, abandonment: contact-centre KPIs that actually matter

Contact-centre dashboards have an interesting problem: every vendor adds another metric, no one removes any, and after 18 months the dashboard has 20 numbers and no signal. Then teams start optimising whatever is easiest to move rather than what matters.

This post is the short list of contact-centre KPIs that actually drive decisions, what they mean, and the traps that come with each.

The "always-watch" tier (5 metrics)

These are the only metrics where week-over-week movement should prompt action.

1. Connect rate

(# of calls answered by a human / # of calls placed) × 100

Tells you: list quality + dialing time + DID reputation.

Healthy ranges:

• Outbound B2B prospecting: 25-40%
• B2B with prior engagement: 50-60%
• B2C collections / D2C: 35-50%
• Inbound (different metric — abandonment): N/A

If your connect rate drops 10% in a month, one of three things is happening:

1. List quality decayed (stale data)
2. Time of day shifted (calling 7-9 PM gets worse pickups)
3. Your DIDs got flagged as spam — TNS / First Orion / Hiya analytics are showing your numbers as "Scam Likely"

Connect rate is the canary for spam-flagging. Watch it weekly per-DID; rotate aggressively.

2. Abandon rate (outbound dialing)

(# of calls picked up by a human but no agent connected / # of calls picked up) × 100

This is the TRAI metric. TCCCPR 2018 caps it at 3% per campaign per day. Cross it consistently and your DIDs get suspended from the operator side.

Healthy: < 3%, ideally < 1.5%.

Predictive dialers move this most. If you are running predictive on a small team, abandon rate will climb. Switch to progressive and abandon rate goes to zero (progressive only rings when an agent is free).

3. Average Handle Time (AHT)

Total call duration + after-call work / number of calls

Includes: ringing, conversation, hold time, post-call CRM logging.

Healthy ranges depend wildly on use case:

• Sales discovery / qualification: 4-8 minutes
• Customer support (simple): 3-5 minutes
• Customer support (complex): 8-15 minutes
• Collections: 3-6 minutes

The trap: AHT optimisation is misleading on its own. A team that shaves AHT by 30 seconds via aggressive coaching often sees CSAT drop because agents are rushing customers. Always pair AHT with CSAT.

4. First Call Resolution (FCR) — inbound

(# of issues resolved on the first call / total calls) × 100

For inbound contact centres, this is the single best proxy for customer satisfaction. A call that did not need a follow-up is a call that solved the problem.

Healthy: 70-85%. Below 60%, you have a knowledge / training / process gap.

The trap: FCR is harder to measure than it sounds. "Resolved" usually means "no callback within 7 days on the same issue", which requires linking calls by recipient + topic. Worth investing in.

5. Conversion rate (outbound) / CSAT (inbound)

The only metric that pays the bills.

For outbound:

• Conversation → demo / meeting: 5-15% (B2B), 10-25% (B2C)
• Demo → deal: 10-30%

For inbound:

• CSAT (post-call survey, "rate this call 1-5"): healthy is 4.2+ average

Conversion / CSAT is the truth check on every other metric. If AHT is improving but conversion is dropping, you are optimising the wrong thing.

The "review monthly" tier (3 metrics)

Less actionable but useful for trend-watching.

Service Level (inbound)

(# of calls answered within X seconds / total inbound calls) × 100

Standard SLA is 80/20 (80% of calls answered within 20 seconds). Below 80%, agents are short. Above 90%, agents may be over-staffed.

Schedule Adherence

(actual hours worked / scheduled hours) × 100

Measures whether agents are showing up on time and staying on shift. Healthy: 90%+. Below 85%, the schedule does not match reality or agents are disengaged.

Occupancy

(time spent on calls + after-call work / total logged-in time) × 100

How "busy" each agent is. Healthy: 70-85%. Above 90% causes burnout; below 60% means you are over-staffed.

Metrics not worth chasing

A few that vendors push but rarely drive decisions:

• Calls per hour — meaningless without conversion. Dial faster, talk less, miss every deal.
• Average wait time (outbound) — only relevant if you are running power and the wait is the bottleneck. Otherwise included in AHT.
• NPS instead of CSAT for contact centres — NPS is a brand-level metric, not a call-level one. CSAT is more granular and actionable.
• Total minutes on phone — same as calls-per-hour, no conversion signal.
• Vanity counts (total dials made this month) — used in board decks, not in operations.

KPI traps

Trap 1: optimising AHT without watching CSAT

Aggressive AHT pressure causes agents to rush, cut off concerns, and skip resolution steps. CSAT drops. Customers churn. Net negative.

Fix: pair AHT and CSAT on the same dashboard. Reward agents for the AHT × CSAT product, not AHT alone.

Trap 2: ignoring per-DID metrics

Connect rate as a team average hides the truth. If half your DIDs are flagged and the other half are fine, the team average is 25% but the fixable issue is concentrated in 50% of the numbers.

Fix: monitor connect rate per DID, weekly. Pull anything < 5% from rotation, replace with a fresh number.

Trap 3: campaign-level vs agent-level analysis

Comparing agent performance against the team average without controlling for campaign / time-of-day produces nonsense. Agent A on the morning campaign vs Agent B on the evening campaign will look very different even if they are the same skill level.

Fix: compare within-campaign, within-time-window. Aggregate at the agent level only after controlling for these.

Trap 4: chasing 100% recording compliance vs sample-quality QA

Some teams brag that they record 100% of calls and listen to 100% of recordings via AI. Sounds good. Reality: AI scoring is correlated with manual scoring at ~70-80% — fine for trend signals, terrible for individual coaching decisions.

Fix: AI-score 100% for trends + flag outliers; manual-review 5-10% of calls per agent per week for coaching.

What dialque dashboards default to

• Connect rate per-DID + per-campaign, weekly
• Abandon rate per-campaign, daily, with alerts at 2.5% (below TRAI's 3% threshold)
• AHT + after-call work time per-agent, weekly
• Conversion by disposition code, per-campaign
• CSAT if recorded (manual or via post-call IVR)
• DID rotation alert when connect rate drops > 10% week-over-week

Everything else is on-demand drill-down, not on the always-watch surface.

Frequently asked questions

How often should I review dashboards?

• Always-watch tier: daily, glance
• Review-monthly tier: monthly, with trend lines
• Per-agent coaching metrics: weekly 1-on-1s

What is a good benchmark for SDR conversion?

Highly variable by industry. B2B SaaS: 6-12% conversation → meeting, 8-15% meeting → opportunity, 15-30% opportunity → deal.

Should I track call duration or talk time separately from AHT?

Yes if you want to coach on conversation skills vs CRM logging. Long calls + short post-call = conversational agent. Short calls + long post-call = process bottleneck.

Does TRAI's 3% abandon rate apply to all outbound, including B2B?

Yes. No B2B carve-out under TCCCPR 2018.

The right contact-centre KPI is one that, if it moved, you would change something about how you operate. Most metrics on most dashboards fail that test. Trim to 5-8 metrics that drive decisions, get rid of the rest, and the team's discipline improves within a quarter.