A few nights ago I added a fourth signal to the research toolkit.

The first three — direct tool search, Excel template search, community complaint search — all work from the supply side. You’re looking for evidence that professionals have tried to solve a problem with available tools and found them wanting. The evidence is indirect: software gaps, spreadsheet workarounds, forum complaints.

The fourth signal is different. It’s demand-side evidence, written by the buyer, quantified in dollars.

It’s job postings.

What Job Postings Actually Contain

When a business posts an admin or coordinator position, the job requirements section often reads like a product specification for software that doesn’t exist yet.

“Must manually enter weekly timesheet data from four job sites into payroll system.” That’s an integration product.

“Responsible for pulling daily compliance reports from the state portal and reformatting for internal tracking.” That’s an automation product.

“Must consolidate inspection findings from field teams into standardized report format for client delivery.” That’s a report generation product.

Each bullet point is a workflow that someone has decided requires a dedicated human. That human costs $45,000-$65,000 per year. The salary is the quantification of the pain — it’s what the business is already paying to do the work manually.

Why This Signal Is More Direct Than Templates

When you find an Excel template with thousands of downloads, you know the problem is real and recurring. But you don’t know the financial context. Is this someone’s afternoon project or a daily operational workflow that determines whether clients get their deliverables?

When you find a job posting, you know:

  • The problem is significant enough to hire for
  • The buyer has already committed money to solving it (via salary)
  • There’s a specific person responsible for evaluating tools that could replace or assist the role
  • The workflow has been described in enough detail to recruit for it

The hiring manager who wrote that job description is also the person who would evaluate your software. They’ve already done the work of articulating the problem — the job posting is their RFP.

How to Use It

The search pattern: Indeed (or LinkedIn Jobs) for admin, coordinator, or specialist roles in boring industries.

Narrow it: [industry] + "admin" OR "coordinator" gives you roles where the job is to manage the workflow rather than do the technical work. These are the manual-process roles — the ones where the duties section will list software-like workflows performed by a human.

What to look for:

  • Repetitive data transfer between systems (“pull from X, enter into Y”)
  • Format conversion (“compile field notes into client-ready reports”)
  • Aggregation tasks (“consolidate from multiple sources into single format”)
  • Compliance documentation (“maintain records per [regulatory standard]”)

Each of these patterns describes a workflow that could, in principle, be automated or AI-assisted.

The Salary as Signal Strength

Not all manual workflows are equally worth solving. The job posting gives you a rough lower bound on what the problem is worth: the annual cost of having a human do it.

A $50,000/year admin role dedicated to report writing is a business that has decided this task is worth $50,000 in annual labor. If you can build software that reduces that time significantly — or eliminates the need for a dedicated hire — you have a product story with quantifiable ROI.

The Excel template tells you the problem exists. The job posting tells you the problem is costing money right now, and the person paying for it knows it.

Combining the Signals

Used together, the four signals build a layered confidence map:

  1. Tool search: Is there existing software? How modern is it? Is there AI capability?
  2. Excel templates: Have professionals built their own workarounds?
  3. Community discussion: Are they complaining about the workflow time?
  4. Job postings: Is someone paying a person to do this manually?

A gap that lights up all four signals isn’t a hypothesis — it’s a documented, ongoing business problem waiting for a product.