My second hire only gets one idea at a time

The second role on my one-person team is the research bench, and what defines it is a limit. Each time it runs, it can take at most one idea off the scout's list. It has to source every claim in that idea, say who published each number, and then make the strongest case it can against it.

If you're new here: every Monday this newsletter takes one role from that team and shows how it works, something it actually produced, what broke, and a template you can copy. Last week was the scout, which works for my show, The Operator Economy, looking for problems someone might pay to have solved. The bench is the scout's checker.

It runs Monday and Thursday mornings. It doesn't go looking for ideas; that's the scout's job. It reads the shortlist and the scout's latest run, rechecks that a lead's sources still say what the scout said they did, and checks the lead against everything I've already covered or turned down. If one holds up, it writes it up properly, with every claim tied to its source, a pass on how much each piece of evidence can actually carry, and a review whose only job is to attack the idea. If nothing holds up, it writes that down and stops instead of inventing something to fill the run.

The limit is there because finding ideas is the cheap part now. My scout turns up plenty. The expensive part is reading one closely enough to know whether it's real, and a checker that takes a handful at a time reads all of them badly.

The other rule I care about is who's doing the talking. I've sat through enough vendor pitches to know the best number on the slide usually comes from the company selling the thing. So a seller's price list or case study can go in the bench's file as context. It can never be the reason an idea moves forward.

And it can't decide anything. It can't approve an idea, give it an episode, write a word of script or contact anyone, and its only spending is a small, capped amount for reading the pages it checks. Its verdict is advice. Whether an idea goes forward is my call.

What it actually did

On Thursday, October 15, it took one lead off the list: a service that follows up on unpaid invoices for small firms that bill other businesses.

The case for it was better than most. A survey of small businesses says more of them are waiting on overdue invoices than a year ago, though the survey comes from a company that sells invoicing software, and the bench marked it as one vendor's survey. Employers are posting part-time jobs for exactly this work, so someone already pays to have it done. And a company already sells the done-for-you version to small professional firms, at monthly prices you can look up. The bench wrote those prices down as context for what an operator might charge, then added the line I'd want from any analyst: that isn't proof anyone pays it.

Then it made the case against. Whoever runs this gets squeezed from both sides, by software the owner could run alone and by that human service at a known price. In some states, chasing another company's invoices can count as running a collection agency, which means registering or getting a license. And the gap might not be labor at all. Owners may skip the second reminder because they're afraid of losing the customer, and a stranger sending it could make that worse.

It also took the most attractive number out of the file and asked whether the idea still stood without it. It did.

The verdict was to keep researching, with one pass before anything else: which large states would treat this as collection work, and what the software version costs a small firm. It parks the idea only if most of the big states require a license one person couldn't reasonably get. Whether owners would actually pay an outsider for this is written down as the first thing a viewer would have to test.

What broke

The bench was too good at saying no, for the wrong reason.

For most of September, its reviews kept parking ideas until I'd done field work myself: interview buyers, ask someone for a paid pilot, send outreach. On paper that's rigor. In practice it made me the blocker for every idea, and it misread what the show is. An episode is research a viewer can take and go test. It isn't proof that I ran the business.

On September 28 I wrote that down as a rule. Now the bench can never make my field work a condition for moving an idea. Anything only field work could answer becomes the viewer's first test, labeled as one. Nothing else got looser, which is why the October 15 verdict lists paying customers as an unknown and still won't count the seller's prices as evidence.

Copy this

ROLE: Checker (works on another role's list, never its own)

JOB (one sentence):
[e.g. Twice a week, take at most one item from the shortlist and make the strongest case against it.]

INPUTS (the only places it may read):
- The shortlist it checks, and that list's latest run
- The sources each item cites, rechecked

SCHEDULE:
[e.g. Mondays and Thursdays, 9:00am]

OUTPUT (same shape every time):
- The item, with every claim sourced and who published each number
- The question most likely to sink it, and the answer so far
- Keep, park or drop, with one line of reason

LIMITS (it may never):
- Take more than one item per run
- Treat a seller's own marketing as evidence
- Approve, schedule or spend anything
- Make my extra work a condition for moving forward (turn it into a test instead)

APPROVALS:
- Its verdict is advice. Moving an item forward is my decision.

If you change everything else in that template, keep the first two limits. One item per run is what makes it read closely, and the seller rule is what stops it from handing you a sales pitch dressed up as research.

Shipped this week

• This newsletter comes out every week now.
• Canairy, my early-warning dashboard for families, now puts each area it watches, like jobs, costs and the weather, at a household stage from Baseline to Brace, with a plan for what to do at each one.

Next issue, next Monday: the writing room, where an approved idea becomes a script. If someone passed this along, you can get it every Monday at mpthaker.xyz.