Outbound calling metrics that matter

Outbound calling metrics come down to four numbers: connect rate, conversation rate, booking rate and cost per outcome. The 2026 benchmarks are clear: an average of 1.55 calls to reach a prospect, 65.6 percent of connected calls becoming conversations, and a 2.7 percent average success rate, all from Cognism research covering more than 200,000 cold calls. Track those four weekly, against stable definitions, and reporting stops being vibes. This guide defines each metric, shows what good looks like in 2026, and ends with a one-page weekly report you can build in an afternoon. It comes from 6 Seven Labs, which builds AI voice calling agents for Singapore businesses and tracks these four numbers on every campaign.

The four metrics that matter

Every outbound programme, human or AI, runs the same funnel: dials become connects, connects become conversations, conversations become booked outcomes, and the whole thing costs money. One metric per stage is enough, plus one honesty check: show rate, the share of booked meetings that actually happen. Add more and the dashboard starts hiding problems instead of exposing them.

Metric	How to work it out	2026 benchmark	What good looks like
Connect rate	Connected calls divided by dial attempts	13.3% answered on verified data; 1.55 attempts on average to reach a prospect (Cognism)	Double-digit answer rates on a clean, verified list
Conversation rate	Conversations divided by connects	65.6% of connects become conversations (Cognism, May 2026)	Two in three connects turn into a real exchange
Booking rate	Booked outcomes divided by total calls	2.7% industry average; 11.3% for a top team (Cognism)	Climbing quarter on quarter, counted on held meetings
Show rate	Meetings held divided by meetings booked	85.9% held by Cognism's own team; 6.5% overall no-show, mostly inbound-booked (RevenueHero, Dec 2024)	Roughly nine in ten booked meetings happen
Cost per outcome	Total calling spend divided by held outcomes	No public benchmark; calculate your own monthly	Falling while volume holds steady

As a funnel starting from 100 dials, the 2026 benchmarks look like this.

Three meetings from 100 dials is roughly what the industry average produces. The rest of this guide takes the stages one at a time: what each metric means, where the benchmark sits, and which lever actually moves it.

What is a good connect rate in 2026?

A good connect rate in 2026 is a double-digit answer rate: SDRs working from verified data with AI-assisted research answered 13.3 percent of dials in Cognism's Cold Calling Competitiveness Gap research (2026). Connect rate itself is connected calls divided by dial attempts. A related but different number: reaching a prospect now takes an average of 1.55 call attempts, down from 2.9 a year earlier (State of Cold Calling 2026, 200,000+ analysed calls). That one measures persistence with prospects you eventually reach, not the share of dials answered.

The driver is data quality, not charm. Connect rate is mostly a list metric; a bad number cannot be out-dialled.

Timing helps at the margin. The same 2026 report puts the best calling window at 10 to 11am, a secondary window at 2 to 3pm, and Thursday as the strongest day. Worth scheduling around, not worth obsessing over.

Persistence has a ceiling too. Cognism's cold calling statistics, updated May 2026, show 93 percent of conversations happen by the third call attempt and over 98 percent by the fifth. Three to five attempts spread across different windows is a sensible policy; after that, recycle the number and move on. And every list needs to clear Do Not Call checks before anyone dials: our guide to PDPA and DNC compliance for outbound calling covers who is responsible for what.

What is a good conversation rate?

A good conversation rate is about two in three: 65.6 percent of connected cold calls become conversations (Cognism, updated May 2026), higher than most people guess. Conversation rate is the share of connected calls that become a genuine two-way exchange instead of an instant brush-off. If yours sits well below two thirds, the problem usually lives in the first ten seconds: the opener, the tone, and whether the caller sounds like a person or a script.

Call length is the supporting signal. The average cold call now runs 82 seconds, down from 93 a year earlier (Cognism, 2026). Short is not the goal, though. Gong's cold call analysis, from a 2021 dataset so treat it as directional, found successful cold calls run nearly twice as long as unsuccessful ones. A rising average duration on connected calls usually means more real conversations, not wasted minutes.

The lever here is the script: the opener, the permission ask and the first objection response. We break down what actually works in cold calling scripts that work.

What is a good booking rate?

A good booking rate in 2026 is anything clearly above the 2.7 percent industry average, itself up from 2.3 percent a year earlier (Cognism State of Cold Calling 2026). Booking rate, called the success rate in most research, is booked outcomes divided by total calls. The average hides a wide spread: the same company's Competitiveness Gap research measured an 11.3 percent success rate for its own team working from verified data. A four-fold gap between average and good is a process gap, not a talent gap.

Booked is not held, and only held meetings pay. Cognism's own outbound team held 85.9 percent of its booked meetings in the same 2026 research, a useful reference for what a tight confirmation process achieves. For the wider no-show picture, RevenueHero's benchmark of 6,428 meetings (December 2024) measured a 6.5 percent overall no-show rate, mostly on inbound-booked meetings, which show up more reliably than cold-booked ones. The spread by sector is wide in that dataset: education at 18.1 percent and real estate at 15.1 percent.

Two rules follow. Report held meetings, not booked ones. And treat the gap between the two as a metric of its own: confirmations, reminders and a short wait between booking and meeting all close it.

How do you work out cost per outcome?

Divide total monthly calling spend by held outcomes. Total spend means everything: salaries or plan fees, data, telco, tooling and management time. The outcome is whatever the campaign exists to produce: held meetings, completed surveys, confirmed renewals. This is the one number finance will actually read, so honour the definition and keep it identical month to month.

The billing model changes the arithmetic more than any other input. On our published plans, billing runs per second and only on connected calls, so a dial that rings out costs nothing: Starter is S$680 a month for 500 minutes and Growth is S$1,980 for 2,000 minutes at list prices, before any founding discount, with current rates always in the pricing section. At the 2026 average cold call of 82 seconds, 2,000 minutes works out to roughly 1,460 connected calls a month; apply your own conversation and booking rates from the sections above and the cost of a booked outcome falls out of the division.

For the full comparison against hiring, including salary, CPF and turnover, the cost-per-conversation maths is in our guide to AI callers vs hiring a telemarketer in Singapore.

Build a weekly report that stops being vibes

One page, four rows, updated weekly: connect rate, conversation rate, booking rate counted on held meetings, and cost per outcome. Show last week, this week and the four-week trend. That is the whole report, and the template looks like this.

Metric	Definition used	Last week	This week	4-week trend
Connect rate	Connected calls / dial attempts
Conversation rate	Conversations / connects
Booking rate	Held meetings / total calls
Cost per outcome	Total spend / held outcomes

Three habits keep it honest. Freeze the definitions in writing, because a booking rate that quietly switches its denominator from calls to connects will flatter itself enormously. Review the trend, not the snapshot; single weeks are noisy at SMB volumes. And give each metric one owner and one lever: list quality for connects, script for conversations, offer and follow-up for bookings.

6 Seven Labs builds AI voice calling agents for Singapore businesses, and measurement comes built in: every call is recorded, transcribed and logged with an outcome in the client portal, so these four numbers come out of the system rather than a spreadsheet rebuilt at month end. How it works covers the setup, and a 20-minute demo will show you a live call against your own use case.

Frequently asked questions

What is a good cold call connect rate in 2026?

A good connect rate is a double-digit answer rate: teams working from verified contact data with AI-assisted research answered 13.3 percent of dials (Cognism, 2026). Reaching a prospect takes an average of 1.55 call attempts, but that measures persistence with those eventually reached, not the share of dials answered.

What is a good cold calling success rate?

The industry average success rate is 2.7 percent of cold calls, up from 2.3 percent a year earlier (Cognism, 2026). Top teams do far better: the same research measured an 11.3 percent success rate for a team working from verified data, so treat the average as a floor, not a target.

How many call attempts should you make per prospect?

Plan for three to five. Cognism data updated in May 2026 shows 93 percent of conversations happen by the third call attempt and over 98 percent by the fifth. Beyond five, extra conversations are marginal, so recycle the number back into the list and prioritise fresh prospects instead.

What is a normal no-show rate for booked meetings?

RevenueHero's December 2024 benchmark measured a 6.5 percent overall no-show rate across 6,428 meetings, mostly booked from inbound requests. Cold-booked meetings need tighter reminders and faster follow-up. Cognism's own team held 85.9 percent of booked meetings in its 2026 research, so a tight process can hold roughly nine booked meetings in ten.

How long should a cold call be?

The average cold call now runs 82 seconds, down from 93 a year earlier (Cognism, 2026). Short is not the goal: Gong's cold call analysis, from a 2021 dataset, found successful calls run nearly twice as long as unsuccessful ones. Judge calls by outcomes, not by seconds saved.