The role that shows up every spring
Every growing service company reaches the same moment. The founder or the office manager is drowning, the team is dropping follow-ups, and someone says the sentence that costs $75,000 a year: "We need to hire an admin." Sometimes they're right. More often, nobody has looked at what the admin would actually do all day.
When we map the role before the hire, the pattern is consistent across law firms, home-services companies, agencies and advisory firms: roughly two-thirds of the proposed job is work with a clear trigger, a known set of rules and a predictable output. That is the definition of work a connected system runs. The remaining third is judgment, relationships and exceptions, which is the part worth paying a person for, and usually not a full-time person.
The map: what absorbs, what stays human
This is the table we build in the first week of an operations engagement. Yours will have different rows; the shape won't change.
| Task | Absorbed by a system? | How | What stays human |
|---|---|---|---|
| New inquiry intake (form, email, voicemail) | Yes | Every inquiry lands in the CRM with the fields filled, the source tagged and a confirmation sent in under a minute, including after hours | Deciding whether to take the work |
| Follow-up on quotes and proposals | Yes | Sequenced follow-ups triggered by behavior (opened, didn't reply, visited the pricing page) until booked or dead | The call when the prospect has a real objection |
| Scheduling and reminders | Yes | Booking links tied to real availability, reminders by text and email, reschedules handled without a human | Nothing, usually |
| Data entry between tools | Yes | The CRM, the project tool and the invoicing system share records; a change in one updates the others | Auditing the exceptions once a month |
| Weekly and monthly reporting | Yes | Reports assembled from the systems of record and delivered on a schedule; dashboards that are live instead of a Friday project | Reading them and deciding |
| Meeting prep and post-meeting notes | Mostly | Research pack before the meeting, summary and action items after it, logged to the record | Reviewing the summary before it goes to a client |
| Document collection and onboarding | Yes | Checklists that chase themselves, with reminders and a status the team can see | Judgment calls on incomplete or unusual files |
| Client questions and escalations | Partly | First-line answers from your own policies and FAQs; anything outside them routed to a person with context attached | The answer itself, when it matters |
| Vendor and partner relationships | No | All of it | |
| Noticing something is wrong | No | Alerts help; judgment doesn't transfer | All of it |
Add the hours. In most of the maps we build, the "yes" rows account for 20 to 30 hours of the proposed 40, which is the same number FrontPipe clients see come back when the Operations stack goes in. The job that remains is eight to fifteen hours a week of real judgment. That is a redeployed senior person, a part-time role, or a half of someone you already have, not a new full-time hire.
The 30-day test before you post the job
Don't take our word for the split; run the test. It costs a month and no salary.
- Week 1: write the real job. Not the posting. The list of tasks the person would do, with the hours each takes now and who does it today. Mark each row absorbable or human using the table above.
- Week 2: build the top three absorbable rows. Intake, follow-up and reporting are almost always the three with the most hours. Build them on the tools you already run; don't buy a platform to find out.
- Weeks 3–4: measure. Hours the owner and the team spent on those three rows, before and after. Follow-ups sent within an hour, before and after. Inquiries that got a same-day response, before and after.
- Day 30: decide what the role actually is. If 25 hours came back, you are hiring for a 15-hour job, and you may already have that person. If they didn't, you learned something about the work that the job posting would have hidden.
The honest outcome of this test is sometimes "make the hire." A role that is mostly exceptions, relationships and judgment is a real role, and a system will not do it. The test tells you that, too, before you've spent the first quarter finding out.
What it costs, against the hire
A coordinator or admin at $55,000 is a $70,000 to $80,000 line once taxes, benefits, software and management time are in, and it ramps for a quarter. The Operations stack is $2,500 a month, live in 14 days, month to month after the 90-day kickoff, with the software included. That is $30,000 a year for a system that doesn't take vacation, doesn't quit in August and gets better every month, against a line that only gets more expensive.
The comparison that matters more is revenue per employee. Grow 15% next year and hire the admin, and the number barely moves. Grow 15% and absorb the admin's work into a system, and it moves by exactly that hire. Owners who track one number through an AI program track that one.
Where systems fail, so you can plan for it
Three places, every time. Exceptions: the client whose situation doesn't fit the form. Route those to a person with the context attached; don't try to automate the edge. Tone: a follow-up sequence that reads like a machine loses the deal a person would have saved. Write them in the voice of whoever would have sent them, and read the first fifty before they go out. Ownership: if nobody owns the system, it drifts, and in six months the founder is the system admin again. Someone, inside or outside, owns the monthly review.
Plan for those three and the map above holds. Skip them and you'll be back to the job posting by spring, having spent on tools what you could have spent on the system.
If you'd rather have it mapped for you: the free Revenue-per-Employee Audit takes the hires you're planning, shows what a system absorbs from each, and prices the 90-day build. Written, within three business days.