Company Recon: Read the Company Before the Interview

How I got here

I was doing this by hand and doing it badly. Somebody sends a role, it looks interesting, and the research I do is thirty minutes of the careers page and a funding headline. Both of those are marketing. Neither tells you whether the design team has a leader, whether the last three seniors left in the same quarter, or how long the money lasts.

The information exists. It is just scattered across review sites, LinkedIn departure patterns, funding filings and forum threads, and assembling it every single time is the kind of tedious that makes people skip it and take the interview anyway.

How I thought about it

Weight the people who already left. Review scores are close to useless taken flat, because current employees write carefully. The signal is in the exit reviews and in the shape of who stopped being there. A leadership team with two unreplaced gaps says more than any rating.

Two verdicts, not one. This is the design decision I think about most. "Should I join" and "should I interview" are different questions with different answers, and collapsing them is why people either over-apply or freeze. A company can be a bad place to work and still worth interviewing at for the reps, or for the leverage a competing offer gives you. So the tool answers both, separately, and says which kind of yes it is giving you.

Score against your constraints, not a generic idea of a good job. On first run it interviews you once about your target band, your stability bar and your hard filters, and saves that profile. Every finding afterwards is measured against your situation. A twelve-month runway is a different fact for someone with savings than for someone who cannot afford a gap.

What I actually did

Built the dossier the way I would if I had an afternoon per company: how long the money lasts, who the founder actually is and what happened at their last company, who left the leadership team in the past eighteen months and whether anyone replaced them, which companies the staff came from and where the alumni went, and what the reviews say once the leavers are weighted properly.

Then it holds that against the profile you gave it and says what it thinks, including when the honest answer is that the evidence is thin. A tool that always produces a confident verdict is producing some verdicts it has not earned.

Where it's heading

I built it for my own search, which is the only reason I know it works. The part I would extend is the alumni trace, because where people go after a company is the least gameable signal there is and it is currently the thinnest section of the dossier.