What to measure in outbound, and what to ignore

Outbound reporting drifts toward whatever number looks best. It happens gradually and without bad intent: a campaign underperforms on replies, someone points out that opens are strong, and by the next month opens are the headline. Six months later the dashboard has thirty tiles and nobody can say whether the program is working.

The fix is a fixed, short set of metrics agreed on before launch and reported unchanged every week. Not the most metrics you can produce — the fewest you can defend.

The set worth keeping

Deliverability health. Bounce rate and inbox placement per sending domain. This is a leading indicator for everything else. A domain that starts landing in spam will show a reply-rate collapse two weeks later, and by then the damage takes months to undo.

Positive reply rate. Replies indicating willingness to talk, divided by contacts touched. Not total replies — out-of-office and "remove me" are not the same signal as "tell me more."

Qualified conversation rate. Positive replies that clear the written criteria. This is the number that separates a program producing pipeline from one producing activity.

Handoff acceptance. Qualified conversations sales agreed were worth taking. The only metric that involves a second party, and therefore the only one that cannot be quietly graded on a curve.

Four numbers. Every one of them maps to a specific failure you can act on.

How the funnel narrows

Read the dotted lines as a diagnostic. Each drop-off point has one primary cause, and treating a targeting problem as a messaging problem — rewriting copy when the list is wrong — is the most common wasted month in outbound.

What to ignore

Open rates. Unreliable since privacy proxies started prefetching images. Apple Mail Privacy Protection alone inflates opens for a large share of B2B recipients, and the inflation is not evenly distributed, so you cannot even correct for it. Treat opens as a rough directional signal at best, never as evidence a campaign is working, and never in a report shown to anyone making a budget decision.

Click rate on cold email. Low volume, heavily distorted by link scanners in security appliances. A "click" is frequently a firewall.

Total emails sent. An input, not a result. Reporting it as an achievement creates pressure to raise it, which is exactly the pressure that destroys deliverability.

Connections, follows, and profile views. Real activity, not pipeline. Report them separately if at all.

The weekly report

One table, same shape every week, no commentary in the numbers:

The four-week average column matters more than the weekly figure. Outbound is noisy at weekly volume — a single large account replying can swing a percentage point — and reacting to weekly noise is how programs get rewritten every fortnight and never accumulate a baseline.

Benchmarks are context, not targets

Published reply-rate benchmarks span wildly different markets, deal sizes, list qualities, and definitions of "reply." A 2% positive reply rate is excellent in one segment and mediocre in another, and no external number knows which one you are in.

Use your own first four weeks as the baseline and measure change against it. The question worth answering is not "are we above average" but "is this month better than last month, and do we know why." That question has an answer you can act on. The benchmark question does not.

Attribution, briefly

Outbound attribution gets contentious when a prospect who was emailed later arrives through a demo form. The pragmatic rule: if an account was touched by outbound within the preceding 90 days and then converted through any channel, record it as outbound-influenced and report it separately from outbound-sourced. Two lines, both honest, no arguing about which team gets the credit — and critically, the influenced number is never added to the sourced number in a total.

The one qualitative input

Numbers do not tell you why replies are declining. Once a month, read twenty replies end to end — ten positive, ten negative. Patterns show up in the language that no dashboard surfaces: the objection that keeps recurring, the confusion about what you actually do, the competitor named repeatedly. That reading is the input that makes the next month's numbers move.