I used to believe in the perfect job description.

Early in my career, I thought that if I could just get the words right, the title, the responsibilities, the neat list of qualifications, the rest would take care of itself. The pipeline would arrive, the manager would be happy, the role would close. I spent real hours polishing those documents. I was proud of them.

It took me a while, and a few painful searches, to admit something to myself. Almost every role I have ever seen get stuck had a beautiful job description and almost no real agreement about who would actually get hired.

That gap is the thing I want to talk about, because for years I was filling it with the wrong tool.

A job description, when you look at it honestly, is written for the candidate, and it is written to attract. So it drifts. Every skill that might be useful ends up listed as required. Five years quietly becomes seven. A manager who is hedging, not lying, adds one more must have, and then another, until the profile describes a person who does not exist. Then the real candidates show up, and they get rejected for reasons that were never written down, because the written reasons were never the real ones. I would recalibrate by guessing. Two weeks later we were all frustrated and the role was still open, and I would tell myself the market was hard.

Sometimes the market was hard. More often, I had skipped the conversation that actually mattered.

What changed for me was learning to slow down before I sped up. To block thirty minutes with the hiring manager before I sourced a single name, and to treat that conversation as the real work rather than a formality on the way to posting. I started calling it intake, though honestly the name matters less than the discipline of doing it on purpose.

What I ask in that half hour is not complicated, and it is not a friendly chat about team culture. I ask the manager to tell me the two or three things a person genuinely cannot do the job without, and I keep asking until the wish list collapses into something real. When someone tells me everything is a must have, I have learned that just means we have not actually started yet.

Then I do the thing that taught me the most. I bring a few real profiles to the conversation, a clear yes, a clear no, and one deliberate maybe, and I watch the manager react to actual people. The maybe is where the truth lives. You learn more about what a team really wants in ninety seconds of "hmm, I'm not sure about this one, and here's why" than in any list of qualifications. I wish I had understood that sooner.

I ask what success looks like a year from now. Not the responsibilities, the outcomes. What will this person have shipped or owned or changed by then? That single answer reshapes a search more than the title ever does, and it gives every interviewer the same bar to measure against, which is how you turn a pile of opinions into something you can actually compare.

And I ask what bends if we cannot find the unicorn. Seniority, domain, location, comp, something always has to give, and I would rather decide that in a calm room than improvise it later with a live offer on the table and a manager who has fallen in love with a candidate we cannot afford.

Here is what took me longest to see. None of this belongs in the public job description. That used to feel like a contradiction, and now it feels like the whole point. The job description is the marketing. The intake is the strategy. One is what I say to the world, the other is how I actually run the search.

There is a quieter benefit too, one I only appreciated once I started paying attention to the numbers. A real intake is the first data you collect. The must haves become your screening criteria. The trade offs become a fallback plan you defined in advance instead of inventing under pressure. The twelve month picture becomes the scorecard. When I built a generative AI team of forty from close to nothing in seven months, the speed did not come from sourcing harder or longer. It came from intake sharp enough that I rarely sent a profile the manager and I had not already, in effect, agreed on. The pipeline was small and precise instead of large and noisy. I was not working more, I was guessing less.

I think about it this way now. A job description gives you something to post. An intake gives you something to manage.

So if you take one thing from this, take the order of operations. Before you write a word of the posting, have the conversation. Come with your questions written down, bring real profiles to react to, and leave with a one page summary you and the manager both sign off on. That summary is your contract for the search. The job description comes after, and it comes out better, because by then it is the output of a decision instead of a stand in for one.

I am not against a well written job description. I just no longer believe it is where hiring begins. The real decision gets made in the intake, whether you run one on purpose or let it happen by accident across three weeks of rejected candidates and a role that will not close.

These days, I run it on purpose. It is thirty minutes, and it is the difference between filling a seat and helping build a team.