I hate fishing for work.
When I worked at Domo, there was always this push-pull between the sales team and the post-sales team. We all knew sales was grueling — the cold calls, the research for the introductory call, the lead gen process. And then in post-sales, we had to deliver miracles based on whatever the sales team sold. I love building miracles. I do not enjoy fishing for work.
So when I went out on my own as a consultant, I had to solve the fishing problem. I couldn't just ignore it — no pipeline, no revenue. But I also didn't want to spend my days doing research and cold outreach when I could be building things.
I built a runbook to automate the grueling parts. Here's how it works.
The Problem: Pre-Sales Is Repetitive
Customer outreach for a Domo consultant looks like this every time:
- Discover — find companies that use Domo
- Research — deep dive on a target company (Slack conversations, forum posts, LinkedIn)
- Dossier — create a Google Doc with company overview, people, org chart, outreach strategy
- Contacts — sync discovered people to Google Contacts
- Track — update a prospect tracker spreadsheet
Four of those steps are mechanical — gather data, format it, write it somewhere. One step — writing the outreach strategy — requires judgment. It needs to sound like me, reference the right angle, and know what to say to whom.
Sound familiar? This is the same pattern I wrote about in my last article on automating daily SOPs — a repeatable workflow where most steps are mechanical, but one or two need a human's touch.
The Data Flow
My outreach runbook pulls from four sources and outputs to three destinations:
flowchart LR
subgraph sources["DATA SOURCES"]
WEB["🌐 Web Search<br/>(Exa)"]
SLACK["💬 DUG Slack<br/>(1,609 members)"]
FORUM["📋 Domo Forums<br/>(Vanilla Forums)"]
LINKEDIN["🔗 LinkedIn<br/>(Exa search)"]
HIST["📊 Historical Data<br/>(Connections Tour speakers)"]
end
subgraph pipeline["PIPELINE"]
DISC["Discover<br/>(bulk)"]
RES["Research<br/>(single)"]
SCAN["Scan<br/>(channel)"]
end
subgraph outputs["OUTPUTS"]
DOC["📄 Google Doc<br/>Dossiers"]
CONT["👤 Google<br/>Contacts"]
TRACK["📊 Prospect<br/>Tracker (Sheets)"]
ALL["📋 All Contacts<br/>Tab"]
end
WEB --> DISC
SLACK --> RES
SLACK --> SCAN
FORUM --> RES
LINKEDIN --> RES
HIST --> TRACK
DISC --> DOC
RES --> DOC
RES --> CONT
SCAN --> TRACK
DISC --> TRACK
DOC --> ALL
CONT --> ALLThe Domo User Group Slack is my richest source — 1,609 members, many of whom work at companies that use Domo. The forums give me active practitioners. LinkedIn fills in the gaps for people who aren't in the community.
How I Built It
I broke the pipeline into scripts. Each script does one thing. Then I stitched them together with a runbook.
Discovery scripts — web search for Domo customers, enrichment with company profiles, LinkedIn cross-referencing. These run in bulk and output JSON files.
Research scripts — deep dive on a single company. Pulls Slack conversations (using the search API), forum posts (Vanilla Forums scrape), and LinkedIn profiles. The synthesis step calls Claude to turn raw data into structured personas and outreach strategies.
Output scripts — creates Google Doc dossiers (4 tabs: Overview, People, Org Chart, Outreach Strategy), syncs contacts to Google Contacts, and updates the Prospect Tracker spreadsheet.
Pro tip: ask Claude to "write a parameterized CLI" for each step. Each script takes --company "Name" and --from-research path/to/research.json so they're composable. The runbook just calls them in order.
The Three Phases
Phase 1: Discover (bulk) — I run the discovery pipeline to find Domo customer companies. It searches the web, enriches the results with company profiles, and creates a raw list. This is the "fishing net" — cast wide, sort later.
Phase 2: Research (single company) — When I find a company I want to approach, I run the research pipeline. It scrapes Slack for conversations mentioning the company, pulls forum activity, and searches LinkedIn. Then Claude synthesizes everything into structured personas — who works there, what they care about, what angle to take.
Phase 3: Output — The research output goes three places: a Google Doc dossier (the main artifact), Google Contacts (so I can find them later), and the Prospect Tracker (so I don't lose track of where I am in the process).
What I Learned About Myself
Here's the thing — yes, AI is getting smarter. Yes, AI empowers us to get more done. But the real win isn't "AI does everything." It's identifying the processes you want to delegate so you can focus on the real magic of what makes your job fun.
For me, the magic is building solutions for clients. The grueling part is the pre-sales research and outreach. I don't need AI to do my job — I need AI to do the parts of my job I don't enjoy, so I can spend more time on the parts I do.
When I worked at Domo, I saw this pattern everywhere. Sales teams hated the cold calling. Post-sales teams hated the cleanup. Everyone had a part of their job they'd delegate if they could. The question isn't "can AI do this?" — it's "which of my recurring tasks can I describe well enough to hand off?"
Your Turn
Everyone has a process like this. Something you do repeatedly that takes too long and feels like busywork. It could be:
- Pre-sales research and outreach (like me)
- Onboarding new clients with the same checklist every time
- Writing weekly status reports across projects
- Preparing for quarterly business reviews
- Building prospect lists from conference attendee lists
Ask yourself:
- Is this recurring? Do you do it more than twice?
- Is it multi-step? Does it involve 3+ actions across different systems?
- Is it partially mechanical? Could you script most of it, but one step needs your judgment?
If yes, you have a runbook candidate. I wrote more about this pattern in my tutorial on automating daily SOPs and how I teach my agent skills and runbooks.
Want Help Building This?
I work with teams to identify which processes deserve automation, build the scripts, teach agents your voice, and upskill your team to maintain and extend the system on their own.
Whether you're a Domo customer looking to automate your analytics pipeline, or a team that wants to build AI-assisted workflows around your existing tools, I can help you figure out where AI actually earns its keep — and where a script is all you need.
I walk through the full outreach pipeline in this video on my YouTube channel:
