Instrumenting an outbound engine

CONTEXT

Outbound at ClickPost was manual and unmeasured. Nobody could say which campaigns worked because nothing was instrumented. Reply rates sat at 0.1%.

QUESTION

Can outbound be run like a product: instrumented, experimented on, and improved campaign over campaign?

WORK

I built the pipeline, then the dashboard that made every campaign comparable. Each of the 10 campaigns tested one variable against the last. Select a campaign below.

FIG 4.1 — CAMPAIGN DASHBOARD

CAMPAIGNS

POSITIVE REPLY

47%

VOLUME

1,930

DELIVERABILITY

Healthy

WEEKLY POSITIVE REPLY TREND — CAMPAIGN 10

Positive reply rate climbed from 0.1% on Campaign 01 to 47% on Campaign 10.

Reply rates mean nothing if the mail never lands. Sending domains were monitored as infrastructure, rotated and warmed on a schedule.

FIG 4.2 — DELIVERABILITY HEALTH

DOMAIN GROUP A

● HEALTHY

DOMAIN GROUP B

● HEALTHY

DOMAIN GROUP C

● WARMING

DOMAIN GROUP D

● HEALTHY

DOMAIN GROUP E

● ATTENTION

DOMAIN GROUP F

● HEALTHY

Domains rotate through warming before joining sends. One group flagged is one group paused.

The pipeline itself was the biggest lever. Everything that repeated got automated.

FIG 4.3 — WORKFLOW, BEFORE / AFTER

BEFORE — MANUAL

Find leads by hand, one search at a time

Research each company in a browser tab

Write and send each email individually

Track replies in a spreadsheet, when remembered

No per-campaign measurement at all

AFTER — INSTRUMENTED

PHANTOMBUSTERLead sourcing on schedule

↓

CLAYEnrichment and qualification

↓

N8NSequenced, personalized sends

↓

ZAPIER + SQLReplies land in the dashboard

~90% OF MANUAL EFFORT AUTOMATED

The analyst's time moved from sending mail to reading results and designing the next test.

DECISION

Treat outbound as an instrumented product. One variable per campaign, measured on positive replies, deliverability guarded as infrastructure.

RESULT

Positive reply rate rose from 0.1% to 47% across 10 campaigns, with roughly 90% of the manual effort automated away.

WHAT I'D DO DIFFERENTLY

Define "positive reply" more strictly from campaign one. The early definition was generous, which made the first few campaigns look worse in hindsight than they were.

METHODOLOGY ▸

Positive reply rate = replies expressing interest / delivered emails. Campaign names are anonymized to Campaign 01–10; the trend shapes are reconstructed to match reported per-campaign outcomes.

Each campaign changed exactly one variable against its predecessor: audience, angle, sequence length, or send window.