A segment without a process. A quota without a team.
I joined the company as an Enterprise Account Executive. After Enterprise was up and running, I took over the SMB segment — an area that had never been actively worked on before. There was no trial. No associated SDRs. SMB revenue came through inbound and referrals when it came. The strategic decision had been made: activate SMB. The operational reality was clear: without new hires.
I was full cycle on the segment. Prospecting, approaching, qualification, demo, negotiation, closing — one person, each step. The standard answer would have been: build lists manually, send out 50 templates a day, hope. This bill cannot last twelve months and no quota. And it doesn't stand up to the personalization that this segment responds to at all.
Why I built instead of buying or hiring.
Three options were on the table. Set SDRs. License a tool stack like Apollo or Outreach.io. Or build the workflow yourself. The first option did not exist: no headcount budget. And even if there had been, no single SDR writes 2,000 cleanly personalized messages a month with the depth that this segment needs. The second option was tempting, but too shallow: standard tools often reduce personalization to first names and industry variables. This is exactly what is being ignored. So I built.
Framework conditions that forced the design
- Lean. No unnecessary tool tax. There is no provider that cannot be replaced within a week.
- Specific. Personalization had to be tied to real people: posts, companies, recommendations, context — not template variables.
- Ownable. The process had to be owned by the company. No black box agency, no seat logic. Workflow in n8n, data in sheets, message creation using your own prompts, sending via PhantomBuster.
- Transferable. The architecture had to survive a single person. Whoever takes over later should inherit an ongoing process, not a personality show.
Four building blocks. One process. Built lean.
The architecture was deliberately minimal. Four building blocks, each with a clear task, connected via a workflow. The same basic form is now documented step by step in the Outreach Agent Playbook. This case was the origin.
01 · Pull signals (n8n)
The scraping logic ran in n8n. For each target contact, the workflow pulled publicly visible LinkedIn information: profile, role, company, recent posts, activities, recommendations, shared contacts, and profile descriptions. If the company had a public website, mission, news and visible tech stack signals were added. Everything ended up in a Google Sheet, one line per prospect.
02 · Prepare messages (Claude API)
Each line went to the Claude API with a prompt. The prompt combined three concrete reference points: a current activity, a common signal and an industry-specific pain point. The result was an initial message that read like person to person — not like a template. A manual fine-tuning before shipping caught anything that didn't land cleanly.
03 Shipping (PhantomBuster)
PhantomBuster adopted the LinkedIn actions: contact requests with personalized message, sending sequences within platform limits and response recognition. The shipping mechanism was the only part I didn't want to build myself. PhantomBuster solved it well enough, so I bought this layer instead of developing it.
04 · The sheet as truth
Every prospect, every promotion, every answer lived in a Google Sheet. The status changed with each step: pulled, prepared, sent, accepted, replied, qualified, demo booked. The sheet was both a cue and proof. Nothing important was just in one person's head or in a tool that wasn't exportable.
A process that started. And a pipeline that kept running.
The first qualified leads came within the first week after go-live. After four to six weeks, the pipeline was running at a rhythm that I could work on without breaking demos and completions. The numbers about the collaboration: 2,198 personalized contact requests sent, 903 accepted, 40% reply rate in the connected cohort, around 55 hours of time saved per outreach wave compared to manual work.
A system observed from a leadership perspective.
The work did not happen in a vacuum. The RVP Sales who hired me and was my direct manager during that time later described the segment structure in his own words. Names are anonymized at the company's request:
This case is the prototype of the Outreach Agent. Before he had a name.
Everything documented in the Outreach Agent Playbook today — the four-brick form, the personalization logic, the lean stack, the idea of “people, not titles” — was first built in practice here. Not conceived in the workshop. Not theoretically designed. Built because otherwise the bill wouldn't have worked out.
This is the core of the MSA logic: first build as an operator, then document - never the other way around. The architecture must survive contact with a real segment, real quota and real responsibility before it can be sold as a system.
