2d • The Weekly Stack
𝗗𝗲𝗲𝗽 𝗗𝗶𝘃𝗲 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆: 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮 𝗧𝗲𝗮𝗺 𝗕𝗲𝗳𝗼𝗿𝗲 𝗬𝗼𝘂 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗡𝗲𝗲𝗱 𝗢𝗻𝗲
There's something that kills more new agencies than bad marketing ever will.
The Scramble.
Not the dramatic kind, where a whale client fires you and the whole thing implodes. That version at least gives you a story to tell. I'm talking about the quiet one. You close your fourth client on a Tuesday, feel great for about six hours, and then lie awake at 1am doing math on whether you can actually deliver what you just sold.
You know the voice. It sounds like this:
"I'll hire when I can afford it."
"I just need to close a couple more clients first."
"Nobody can do this as well as I can anyway."
"I don't have time to train someone right now."
"I'll just work through the weekend on this one."
"I can't hand this off, it's client-facing."
Every one of those sounds responsible. Every one is the sound of a business capping itself.
𝗥𝘂𝗻 𝘁𝗵𝗲 𝗺𝗮𝘁𝗵 𝗼𝗻 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻 𝘄𝗲𝗲𝗸
Pull up your actual numbers. Monthly revenue, divided by the hours you actually worked.
Say three clients at $1,000 a month. That's $3,000. And say you count honestly, 55 hours a week, because you're doing the sales, the fulfillment, the reporting, the invoicing, and the "quick" calls that eat a Thursday.
55 hours a week is about 238 hours a month. That's roughly $12.60 an hour.
Here's the part that should make you pause. The instinctive fix is to close a fourth client. But a fourth client at the same hourly output doesn't fix your rate. It buries you deeper and makes the eventual handoff harder, because now you're carrying four sets of expectations in your head instead of three.
That's not a capacity problem. That's a bench problem.
𝗧𝗵𝗲 𝘀𝗰𝗿𝗮𝗺𝗯𝗹𝗲 𝘀𝘁𝗮𝗿𝘁𝘀 𝘄𝗮𝘆 𝗲𝗮𝗿𝗹𝗶𝗲𝗿 𝘁𝗵𝗮𝗻 𝘆𝗼𝘂 𝘁𝗵𝗶𝗻𝗸
Here's the counterintuitive part. The scramble doesn't begin on the day you get overwhelmed.
It begins months earlier, on the quiet day you decide that finding help is something you do after you need it.
That one decision turns hiring into an emergency. Emergency hiring is how you end up paying someone to produce work you have to redo yourself, then telling everyone in the Facebook groups that "outsourcing doesn't work."
Outsourcing works fine. Outsourcing under a deadline, to a stranger you met four days ago, with no documented process to hand them, does not.
If your first hire blows up, don't blame the hire. Look at when you started looking.
𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗠𝗮𝘁𝘁𝗲𝗿𝘀 𝗳𝗼𝗿 𝗬𝗼𝘂𝗿 𝗣𝗶𝗽𝗲𝗹𝗶𝗻𝗲
This is not a fulfillment topic. This is a sales topic, and most people miss that.
When you have no bench, you sell scared.
You feel it on discovery calls. The prospect mentions they'd want three locations managed instead of one, and something tightens, because you're already doing math on your own weekend. So you steer them smaller. You quote the safe version. You "keep it simple for phase one."
You just discounted your own offer, and the client never asked you to.
Now flip it. Three pre-vetted people sit in a spreadsheet, each of whom has already done a small paid project for you. You know what they cost. You know what they're good at. Same prospect says "three locations," and you say yes without flinching, because capacity is a phone call instead of a crisis.
Same offer. Same you. Completely different close. The bench doesn't just protect delivery, it changes what you're willing to sell.
𝗧𝗵𝗲 𝗕𝗲𝗻𝗰𝗵 𝗙𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸
Five stages. Run them in order. None of this requires you to put a single person on payroll.
𝗦𝘁𝗮𝗴𝗲 𝟭: 𝗡𝗮𝗺𝗲 𝘁𝗵𝗲 𝘀𝗲𝗮𝘁𝘀 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗳𝗶𝗹𝗹 𝘁𝗵𝗲𝗺
You already have a team. Every seat just has your name on it.
Write out every distinct job your agency does. Not tasks. Jobs:
- Prospecting and list building
- Outreach and follow-up
- Sales calls and closing
- Onboarding and intake
- Core fulfillment (whatever you actually sell)
- Reporting and client comms
- Admin, invoicing, collections
- Content and marketing for your own agency
Mark each one with how many hours a week it eats and how much you'd pay to never do it again.
The rule: you cannot delegate a job you have never named. Most first hires fail because the owner hired "a VA" instead of hiring a seat.
𝗦𝘁𝗮𝗴𝗲 𝟮: 𝗣𝘂𝘁 𝗔𝗜 𝗶𝗻 𝗲𝘃𝗲𝗿𝘆 𝘀𝗲𝗮𝘁 𝗮𝘀 𝘁𝗵𝗲 𝘁𝗲𝗺𝗽 𝘄𝗼𝗿𝗸𝗲𝗿
AI is not your team. AI is your worker bees. Bees are excellent at volume and repetition and terrible at deciding what the hive should do next.
So put AI in the high-volume, low-judgment seats first:
- Research and list enrichment
- First-draft outreach copy
- Meeting notes, call summaries, and follow-up drafts
- Reporting summaries from data you feed it
- Repurposing one asset into many
Keep AI out of the seats where a wrong answer costs you the client. Strategy, pricing, the hard conversation when something slipped, the final read on anything a client will see. Those stay human, and right now that human is you.
The rule: AI fills a seat until the seat outgrows it. Every seat AI holds is a seat you are not panic-hiring for, and every hour it buys you is an hour you can spend recruiting on your own calendar instead of your client's.
One warning, because this is the part that bites people. AI holding a seat does not mean the seat is documented. It means you wrote a prompt. Those are not the same asset, and Stage 5 is where that difference gets expensive.
𝗦𝘁𝗮𝗴𝗲 𝟯: 𝗕𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝗿𝗼𝗹𝗼𝗱𝗲𝘅 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗵𝗮𝘃𝗲 𝗮 𝗿𝗲𝗾
This is the actual work of this post, and it costs you a few hours a month.
Where to look:
Other agency owners. The best source, and nobody uses it. Agency owners lose clients and have to cut good people. Ask right here in this community: "Who's the best contractor you've worked with that you don't have enough work for right now?" That's not poaching. That's picking up slack capacity from someone who wishes they had more to give that person.
OnlineJobs.ph. Built for Filipino talent in long-term roles. The free tier won't let you contact anyone. Pro is $69/month or $299/year, Premium $99/month or $349/year. Worth knowing: they publish no official rate card, so anybody quoting you "the typical Filipino VA rate" is quoting their own anecdote, not a benchmark. Verify pricing before you buy, this space moves fast.
Upwork. Better for project specialists than long-term seats. Free to join on Basic, with a published client fee of 5% on payments plus a contract initiation fee between $0.99 and $14.99. Verify pricing before you buy, this space moves fast.
legiit.com an online marketplace where businesses and individuals can buy and sell digital marketing, SEO, AI, web, and other freelance services.
Your client base and local network. Whoever does the books at your plumbing client's office knows three people looking for remote work. Ask.
Communities where the work gets discussed, not advertised. Skool groups, Discords, Slack channels for the specific skill. People solving problems in public are far easier to evaluate than people writing proposals.
The rule: recruit on a schedule, not on a need. Two hours a month. Put it in the calendar next to your outreach block.
𝗦𝘁𝗮𝗴𝗲 𝟰: 𝗩𝗲𝘁 𝗰𝗼𝗹𝗱, 𝗯𝗲𝗳𝗼𝗿𝗲 𝘁𝗵𝗲𝗿𝗲'𝘀 𝗺𝗼𝗻𝗲𝘆 𝗼𝗻 𝘁𝗵𝗲 𝗹𝗶𝗻𝗲
Here's the sequence. Don't skip steps, and don't compress them because you're in a hurry. Being in a hurry is the exact condition this whole system exists to prevent.
Screen on a filter, not a resume. Three yes/no gates before you spend any time: Can they show three pieces of work in the specific thing you need? Is their writing clean in their first unprompted message? Did they answer the actual question in your job post? That third gate alone cuts most of the pile.
Interview for process, not personality. Everyone is charming for 30 minutes. Ask questions that require them to have actually done the work:
- "Walk me through the last time you did this. What was the first thing you did, and what was the last?"
- "Tell me about a project that went badly. What did you do when you realized?"
- "What would you need from me on day one to do this well?"
- "What part of this work do you not like doing?"
A bad answer is vague and resume-shaped, all outcome and no sequence. "I'm very detail-oriented and I always hit deadlines" is not an answer. A good answer is boring and specific, full of tool names and small annoyances. The person who tells you which part of the job they hate is telling you the truth.
Pay for a small test. Always. A real, scoped, paid piece of work. Small enough to throw away, real enough to mirror the job. One GBP audit. One outreach sequence. One month of reporting for a single account.
You are not testing whether they can do it. You are testing four things a call cannot show you: Did they ask a clarifying question before starting? Did they hit the deadline they agreed to? Did they follow the format you gave them? Did they flag what was unclear, or guess?
Free test projects are a mistake. They select for desperate people and give you zero right to hold anyone to a standard. Pay them, even a little, and you get to be a client instead of a favor.
The rule: nobody enters the bench who hasn't been paid once.
𝗦𝘁𝗮𝗴𝗲 𝟱: 𝗞𝗲𝗲𝗽 𝘁𝗵𝗲 𝗯𝗲𝗻𝗰𝗵 𝘄𝗮𝗿𝗺, 𝗮𝗻𝗱 𝗯𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝗮𝘀𝘀𝗲𝘁 𝘀𝗵𝗲𝗹𝗳
A rolodex of people you tested eight months ago and never spoke to again is not a bench. It's a list.
Two things keep it alive.
Contact. Message them quarterly. Not a pitch. Two minutes that keeps you current on availability and rates, and keeps you a real person instead of a stranger when you need capacity on a Monday.
Assets. This is the part people skip, and it's the part that makes the whole thing work. Every time you do a job yourself, record the screen. Every time you write a prompt that gets AI to do a job well, save it in a file named after the seat. Every time you answer a question about how something gets done, that answer goes in a document.
A bench with no documented process is just a list of strangers you're about to disappoint.
𝗔 𝘄𝗼𝗿𝗸𝗲𝗱 𝗲𝘅𝗮𝗺𝗽𝗹𝗲
Meet Marcus. Nine months in, three clients, local SEO and GBP management for home services, about $3,400 a month. He's the entire company.
A referral comes in. HVAC company, four locations, wants GBP management on all four plus monthly reporting. Roughly triples his revenue.
Without a bench. He says yes because he's scared to say no, spends five weeks underwater, misses two reporting deadlines on his existing clients, then posts a job on Upwork at 11pm on week six and hires the first person who replies. He hands over a login and a voice memo and gets back work he has to redo. By month three he's lost one of his original clients to neglect, and he tells himself he's just "not ready for a team yet."
With a bench. Three names already sit in a spreadsheet: a GBP contractor who did a paid single-location audit in March, a reporting VA who built one month of dashboards in May, and a writer he tested in July. He knows what each charges and how each communicates. He also has screen recordings of his GBP process and a prompt file for the reporting summaries.
He messages the contractor Tuesday. Paid, scoped, starts Monday, recordings attached. He takes the deal at full price, and the only thing that changed in his week is that he now has one person to manage.
Same referral. Same Marcus. The difference is six hours of unglamorous work spread across the prior four months.
𝗖𝗮𝗽𝗮𝗰𝗶𝘁𝘆 𝗶𝘀 𝗻𝗼𝘁 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝘆
These two get confused constantly, and confusing them is expensive.
Capacity is how much output you can produce. AI gives you capacity. Ten outreach drafts in the time it took you to write one. A VA gives you capacity too, on whatever process you hand them.
Capability is whether the output is right. It's knowing this client's GBP category is wrong, and that fixing it will do more than the ten posts the AI just drafted. That's judgment, and neither AI nor a brand new VA has it.
Here's where it bites. Adding capacity to a business with weak capability produces more wrong output, faster, with your logo on it. If your process is bad, AI makes it bad at scale. If you've never defined what good looks like, a new hire can't hit it, and you'll call that a hiring problem when it's a documentation problem.
Build capability first. It lives in your recordings, checklists, and prompt files. Then stack capacity on top.
𝗧𝗵𝗿𝗲𝗲 𝘀𝗰𝗿𝗶𝗽𝘁𝘀 𝘆𝗼𝘂 𝗰𝗮𝗻 𝘂𝘀𝗲 𝘁𝗵𝗶𝘀 𝘄𝗲𝗲𝗸
Adding someone to the bench with no work to offer yet.
"Hey [Name], I run a small marketing agency and I came across your work on [specific thing]. I don't have a role open today, but I expect to need help with [specific seat] in the next few months and I'd rather know good people before I'm in a rush. Would you be open to a paid test project, small scope, so we both know what working together looks like? Happy to share details."
Scoping the paid test.
"Here's the test project: [one specific deliverable]. Budget is $[X], flat. Deadline is [date]. Here's the format I need it in: [link or template]. If anything in the brief is unclear, ask me before you start rather than guessing. That's part of what I'm evaluating."
The stay-warm quarterly check-in.
"Hey [Name], just checking in. How's your workload looking this quarter, and have your rates changed since we last worked together? No ask on my end today, I just like to keep my list current."
𝗧𝗵𝗲 𝗺𝗼𝘀𝘁 𝗱𝗮𝗻𝗴𝗲𝗿𝗼𝘂𝘀 𝗽𝗵𝗿𝗮𝘀𝗲 𝗶𝗻 𝘁𝗵𝗶𝘀 𝘄𝗵𝗼𝗹𝗲 𝘁𝗼𝗽𝗶𝗰
"I'll figure that out when I get there."
It feels like pragmatism. It's actually a decision to handle your biggest operational risk under maximum stress with minimum information.
You'd never run your pipeline that way. You don't start prospecting the day a client cancels. You build a pipeline so that losing one client is an inconvenience instead of an event.
A bench is a pipeline, just for labor. Same logic, same discipline, and almost nobody does it.
𝗢𝘄𝗻𝗲𝗿'𝘀 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗶𝗯𝗶𝗹𝗶𝘁𝘆 𝗰𝗵𝗲𝗰𝗸
Answer these out loud. Yes or no, no explaining.
- Can you name every distinct job your agency does, on paper, today?
- Is there at least one person, not you, who has been paid to do work for your agency in the last six months?
- Do you know, right now, what three specific people would charge you for help this month?
- Is there a recording or written process for the thing you do most often?
- Have you spoken to anyone on your bench in the last 90 days?
- Are there AI worker bees actually holding seats, or do you just have ChatGPT open in a tab?
- When did you last spend two hours recruiting with no open role?
- If you closed a deal tomorrow that tripled your workload, would your first feeling be excitement or dread?
Number 8 is the whole post. Your gut already answered it before you finished reading the sentence.
𝗬𝗼𝘂𝗿 𝗻𝗲𝘄 𝗿𝘂𝗹𝗲
Recruit on your calendar, not on your client's.
Two hours a month. One paid test a quarter. One process recorded a week.
That's the entire system, and it costs less than the time you'll spend rewriting one bad hire's work.
The Scramble never announces itself. It shows up as a great Tuesday and a terrible Wednesday, and by then all you can do is survive it.
Build the bench while the Tuesdays are still quiet.
So: what's the one seat in your agency you'd hand off tomorrow if the right person were already sitting in it? Name the seat in the comments, and tell me what's actually stopping you from testing someone for it this month. Go!
2
2 comments
𝗗𝗲𝗲𝗽 𝗗𝗶𝘃𝗲 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆: 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮 𝗧𝗲𝗮𝗺 𝗕𝗲𝗳𝗼𝗿𝗲 𝗬𝗼𝘂 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗡𝗲𝗲𝗱 𝗢𝗻𝗲
powered by
skool.com/digital-edge-5127
Designed for people looking to start or grow a digital agency, come network with like-minded people who are building success on their own terms.
Suggested communities
Powered by