Published in Playbook
What a hiring plan should actually contain
Most "hiring plans" are a job description with a salary band stapled to it. Then the interviews start, and everyone discovers they were hiring for a different role: one interviewer probes system design, another wants leadership stories, and the debrief is a negotiation between opinions formed on incompatible criteria.
A real hiring plan is the document that prevents that meeting. It's short — a page, not a deck — but it settles the three questions that actually determine hiring quality before the first application arrives.
Start with the decision, not the ad
A job ad is marketing; a hiring plan is a decision spec. Write the ad last. The plan starts with what this hire must accomplish in the first year — the two or three outcomes that made you open the req. "Own checkout reliability" is a plan. "Strong backend engineer" is a vibe.
Deriving requirements from outcomes keeps the wish-list honest. If an outcome doesn't require a skill, the skill doesn't belong in the plan. This is where most reqs bloat: every stakeholder adds a "must-have," nobody subtracts, and the resulting unicorn spec screens out every real candidate while committing to none of the actual work.
The screening bar
The bar is the heart of the plan: the must-haves a candidate needs to clear to be worth an interview, and the nice-to-haves that break ties. Each must-have should be evidence-checkable — phrased so a screener (human or agent) can point at something concrete in an application and say met or not met. "Has shipped a consumer mobile app to production" is checkable. "Thrives in ambiguity" is not; it's an interview question wearing a requirement's clothes.
Keep must-haves brutally short — three to five. Every additional must-have multiplies false rejections, because real candidates are jagged: strong ones routinely miss one criterion on paper. The nice-to-haves are where taste lives, and they should influence ranking, never rejection.
The loop
The plan assigns every interview a job. Each round covers a named subset of the requirements, and no two rounds cover the same one — redundant interviews feel rigorous but just measure the same signal twice while doubling candidate fatigue. Four focused conversations beat seven overlapping ones in both signal and close rate.
Write down what "pass" looks like for each round before anyone interviews. This is the difference between a debrief that reads evidence against criteria and one that trades impressions. It's also what makes the loop fair: candidates are compared against the plan, not against whichever candidate happened to interview the same week.
The artifact worth sharing
Before the job goes live, the one-page plan — outcomes, bar, loop — goes to everyone who'll touch the hire: interviewers, the recruiter, the skip-level who'll approve the offer. Not for FYI; for sign-off. Every objection surfaced now is a debrief argument you don't have later, when there's a real candidate waiting and the pressure is to fudge.
In Tandem, this document isn't shelf-ware — it's the operating config. The plan you write is the rubric the screening agent scores against, the structure the interview rounds follow, and the criteria the pipeline reports on. When you change the plan, the whole machine updates. That's the test of a real hiring plan: it doesn't describe the process, it runs it.