~/jordan-green · session started 2026 · all systems normal

Marketing that ships like software.

I run growth like an operator: campaigns with a real point of view on the front, and automation, pipelines, and AI doing the heavy lifting underneath. Every project below has two faces — the story, and the machine. Flip between them.

ops-console — live feed · tail -f growth.log

## 01 · selected_work cron: 0 6 * * 1-5

Five projects, ten faces.

01

Lifecycle email overhaul

DTC skincare · retention

A skincare brand with great first orders and a leaky bucket. We rebuilt the lifecycle program around how people actually use the product — skin cycles and bottle life, not send schedules.

  • Replenishment emails timed off real usage windows, not a flat 45-day guess.
  • A post-purchase arc that teaches routine-building instead of pushing the next SKU.
  • Win-back copy that admits you left and gives one honest reason to come back.
+38%repeat purchase rate
14flows rebuilt
02

AI lead-scoring pipeline

B2B services · n8n + Claude

Sales was drowning in form fills and calling the list top to bottom. We built a pipeline that reads every lead the way a good SDR would — company, role, intent signals, the free-text answer everyone ignores — and routes only the ones worth a human's morning.

  • Every lead gets a score and a one-line reason a rep can actually use on the call.
  • Hot leads hit Slack in under a minute, with context attached.
  • The rubric is versioned and audited monthly against closed-won data.
4,200leads scored / month
+22%more SQLs, same spend
03

Content engine, human in the loop

B2B SaaS · organic growth

Ninety posts a quarter without the content-farm smell. The machine owns research, outlines and first drafts; two editors own voice, claims and the final word. Nothing publishes without a human byline that means it.

  • Topic clusters built from sales-call transcripts, not keyword tools alone.
  • Editors reject or rewrite roughly a third of drafts — that's the feature, not the bug.
  • An internal-link pass runs on every publish, so old posts keep earning.
3.1×organic traffic, 12 months
90posts / quarter
04

Paid creative testing framework

Consumer app · performance creative

We stopped debating ads in meetings and let a framework decide. Creative got modular — hooks × formats × proof points — so every opinion became a testable variant with a name, a budget and a kill threshold.

  • 214 variants shipped in six months without the account turning to soup.
  • Losers die automatically at statistical thresholds; winners get budget the same day.
  • The creative team gets a weekly readout in plain English, not a pivot table.
−27%customer acquisition cost
214ad variants tested
05

Marketing ops warehouse + attribution

Multi-channel retailer · data infrastructure

Every team had its own spreadsheet and its own truth. We built the one that wins arguments: six channels reconciled nightly into a single warehouse, with an attribution model honest enough to admit what it can't see.

  • One dashboard the CMO, the media buyer and finance all read the same way.
  • Attribution cross-checked against holdouts and geo tests — not blind faith in last click.
  • Monday's budget meeting shrank from ninety minutes to twenty.
6channels, one source of truth
20 minweekly budget meeting

## 02 · capabilities_matrix cron: continuous

Craft on one axis, machinery on the other.

discipline strategy craft machinery
acquisition Channel mix & CAC math. Where the next dollar goes and why. Performance creative direction, offer testing, landing pages that read like a human wrote them. Auto-UTM specs, spend rules engines, kill/scale thresholds, API-fed reporting.
lifecycle Retention economics. Cohorts, replenishment windows, when to shut up. Email & SMS that people don't resent — voice, arc, timing. Event-triggered flows, usage models, holdout design, deliverability hygiene.
content Topic authority maps built from real customer language. Editing for voice and claims. The taste layer AI doesn't have. Human-in-the-loop draft pipelines, CMS APIs, internal-link automation.
data & ai Measurement philosophy. What to trust, what to sanity-check, what to ignore. Dashboards people actually open. Plain-English readouts. n8n orchestration, Claude API workflows, dbt models, warehouse-first stacks.

// the point isn't either column — it's that both are one person's job description.

## 03 · ops_log cron: 0 2 * * *

A normal Tuesday, unattended.

system.journal — last 24h● all jobs green
02:00CRON warehouse sync · 6 sources · OK
02:22dbt run · 38 models · 214 tests passed
06:00CRON lifecycle usage model refresh · 41k customers
06:15UTM audit · 312 links clean · 3 auto-fixed
07:30content clusters · 12 new questions ingested
09:14lead #48213 scored 87 HOT · routed in 54s
12:00spend rules · 1 kill · 1 scale · report posted
17:00anomaly watch · nothing to report · sleep

// orchestration

n8nGitHub Actionscronwebhooks

// ai layer

Claude APIprompt rubricseval setshuman-in-the-loop

// data

BigQuerydbtSegmentGA4Fivetran

// channels

KlaviyoMetaGoogle AdsLinkedInWebflow
Hands typing on an amber-backlit mechanical keyboard in a dark room
most of this job is done between 6 and 9, before anyone asks for anything.

## 04 · about whoami

Operator first. Marketer always.

Overhead view of a dot-grid notebook with a hand-sketched flowchart, brass pen and espresso cup under warm lamp light
every system here started as a pencil sketch.

I've spent ten years in growth roles at DTC brands and B2B companies, and somewhere along the way I noticed the pattern: the best marketing teams weren't the most creative or the most technical — they were the ones where those two things lived in the same head.

So that's what I built. I write the positioning and then I write the pipeline. I direct the creative and then I wire up the system that tests it honestly. AI made this a superpower: one person can now run what used to take a pod of five, if they know where the machine is brilliant and where it's confidently wrong.

I work with teams who want that leverage without the snake oil — clear measurement, human judgment where it counts, and automations boring enough to trust.

rule_01if it runs more than twice, it gets automated
rule_02if it can't be measured against a holdout, it's a story
rule_03the machine drafts, the human decides
rule_04boring infrastructure, interesting marketing — never the reverse

## 05 · contact 0 9 * * 1-5 — replies within one business day

Let's build your growth machine.

Full-time roles, fractional engagements, or a one-off "wire our marketing stack together" project — the first conversation is free and I'll tell you honestly if you don't need me.

© 2026 Jordan Green built by hand · zero frameworks · 100/100 vibes