Outbound metrics & benchmarks

Overview

Outbound is a funnel, and almost every argument about "good" numbers is really an
argument about denominators. Before trusting any benchmark, pin down two
things: what it's divided by and whether auto-replies were stripped. The same
"reply rate" is reported anywhere from ~0.45% to ~5% across reputable studies -
not because results differ that much, but because the math differs.

Trust reply rate, positive-reply rate, and meetings booked. Treat open rate as
directional only (see the Apple MPP note). Always state your denominator.

The funnel

Sent
 └ (1 − bounce%) → Delivered
     └ reply rate → Replied
         └ positive share → Interested
             └ positive→meeting → Meetings booked
                 └ held & qualified → Opportunities

Meetings ÷ Sent = (1 − bounce) × reply_rate × positive_share × meeting_conv

Benchmark ranges (as reported - directional, measure your own)

Metric	Healthy	Great	Reported by
Bounce rate (ceiling)	< 2-3%	< 1.5%	Instantly 2026; Woodpecker 2026
Spam complaint rate	< 0.10%	- (never hit 0.30%)	Google sender guidelines
Reply rate (replies ÷ sent, loose)	3-5%	10%+ (5.5%+ top quartile)	Instantly 2026 (3.43% avg); Hunter 2025 (4.1% avg)
Reply rate (unique human ÷ sent, strict)	0.45-0.70%	-	Belkins / sales.co 2026
Positive share of (human) replies	~34%	40-60%	sales.co 2026; Cleverly 2026
Meetings booked ÷ sent	0.5-1.5% (mid-market)	1.0-2.5% (SMB)	Cleverly 2026
Positive reply → meeting	30-60%	60%+	Cleverly 2026
LinkedIn connection acceptance	~24-30%	40%+	Expandi/HeyReach 2026
LinkedIn reply (per message sent)	~10%	15%+	Expandi 2026 (10.4% avg)
LinkedIn reply (per accepted conn.)	~22-25%	33%+	HeyReach 2026
Multichannel lift (email+LinkedIn vs email)	~2-4× reply	-	Pin 2026 (recruiting data)
Cold-call success (→ meeting)	~2.3%	6-11% (elite)	Cognism 2025

Why open rate is broken (Apple MPP)

Apple's Mail Privacy Protection (iOS 15, 2021) pre-fetches the tracking pixel
at delivery, logging an "open" whether or not a human reads the mail. Apple Mail
is reported as ~65% of all tracked opens (Litmus, 2026). So a high open rate
can hide a dead campaign. Use open rate only as a relative deliverability
tripwire against your own baseline - never as an A/B win condition or a KPI.

What to optimize (and what not to)

1. North star: meetings booked (and opportunities) - the metric that maps to pipeline.
2. Optimize: positive-reply rate, then reply rate - the most trustworthy real-
   engagement signals post-MPP, and closest to meetings in the funnel.
3. Do NOT optimize open rate (inflated) or raw volume (burns the domain).
4. Guardrails you drive down, not up: bounce (< 2%) and spam complaints
   (< 0.10%). These are survival metrics - breach them and every downstream number
   silently zeroes.

The leverage lesson: fixing targeting + copy beats adding volume. A weak
campaign (3% reply, 30% positive, 30% book) needs ~9× more emails for the same
meetings as an optimized one (10%/50%/50%) - and the lower volume also protects
deliverability. Always tighten the list before scaling the send.

Worked example - "how many emails for 10 meetings/month?"

Mid-of-road sourced inputs: delivered 97% × reply 5% × positive share 35% ×
positive→meeting 40% = 0.68% meetings ÷ sent (consistent with Cleverly's
mid-market 0.5-1.5%). → 10 ÷ 0.0068 ≈ ~1,470 emails/month (~70/business day).

Leading vs lagging

• Leading (daily/weekly): bounce, spam-complaint, pacing, reply, positive-
  reply, LinkedIn acceptance, speed-to-first-reply.
• Lagging (monthly/quarterly): meetings booked/held, opportunities, pipeline,
  closed-won, cost per meeting.
• Rule: a leading metric rising without the lagging one following = a quality
  problem (replies that are all "not interested"). Watch positive reply, not raw.

Per-campaign checklist (track with explicit denominators)

• Reply rate (÷ delivered) with auto-replies stripped (53% of replies are automated).
• Positive-reply rate and positive-share-of-replies.
• Bounce (alert ≥2%, halt ≥5%) and spam complaints (alert ≥0.10%).
• Meetings booked (÷ sent and ÷ positive reply); meetings held; opportunities.
• LinkedIn: acceptance rate; reply rate labeled per-sent vs per-accepted.
• Reply contribution by sequence step (~40% typically from follow-ups).
• Segment every metric by ICP segment and sending identity (no blended averages hiding a bad list).

Do / Don't

• Do fix one internal definition of each metric and track the trend vs your baseline.
• Do judge campaigns on reply/positive-reply/meetings, not opens.
• Don't compare reply rates across vendors without normalizing denominators.
• Don't let bounce/complaints drift - pause sending before they poison the domain.

Go live with Chronos (optional)

• geo_get_campaign_benchmarks - pull reply/positive/meeting health per campaign.
• geo_list_campaigns / geo_get_account_state - program-level rollup.
• geo_get_credit_status - usage and capacity.

Chronos measures send health per campaign (not a blended average that hides a
bad list) and guards bounce/complaint ceilings automatically.

Connect in ~30s → `CONNECT-CHRONOS.md`. Free, no card.

Sources

Figures above are as reported by these public 2025-2026 studies - treat as
directional and measure your own funnel:

• Instantly 2026 Cold Email Benchmark Report (reply 3.43% avg; omits open rate).
• Hunter.io State of Cold Email 2025 (4.1% avg reply; 11M emails).
• lemlist 2026 (reply "where the truth lives"); Litmus 2026 via lemlist (Apple ~65% of opens).
• sales.co 2026 (1.29M emails; 53.7% of replies automated; 34% of human replies interested).
• Cleverly 2026 (meetings ÷ sent by segment; positive→meeting 30-60%).
• Expandi 2026 & HeyReach 2026 (LinkedIn acceptance/reply); Belkins 2025-26.
• Pin 2026 (multichannel lift, recruiting context); Cognism 2025 (cold calling).
• Google email sender guidelines (spam-rate thresholds).

Honesty flags: the popular "multichannel = +287%" and a hard "0.15-0.5%
opportunity rate" could not be traced to a credible primary - derive opportunity
rate from your own reply→meeting→opp chain instead.

See also: email-deliverability, multichannel-sequences, linkedin-outreach.