[_fool]

I wish I'd had "a real plan" in jobs past as an individual interviewer in a series, instead of a vague goal like "talk about his technical chops" or "see if she can work across teams". I turned out not to be very good or consistent at either freestyling questions completely OR reporting reliably whether a candidate was "successful" in my area without planning and consistency.

I'm a former developer, turned hiring manager for a very technical support team and employ several things I feel are fair and useful and I believe would also have been in that past life.

0. repeatable interview process. When comparing candidates, the comparison should be apples:apples. "check for culture fit" isn't; "ask questions 1,2,3 of each candidate, with goals of finding out x,y,z and help to N degree if requested" is. Leave room for the candidate to demonstrate personality and style of course, as well as (short) tangential conversations that come up during the Q&A process. The goal isn't to make it robotic/automatable.

1. There is a take home exercise. It has a clear scope:

- accomplish goal X (create a website)

- with tools like Y (your choice of static site generator)

- in Z fashion (public github repo)

- in less than 4 hours.

Encourage questions before candidates begin the exercise, so that people work on the right problem rather than guessing wrong. Don't force a timeline ("you have until the job closes which won't be sooner than a week from now"), and use an objective grading rubric for each section of the task.

Here's an example of the objective rubric: 1 point for grammar, 1 point for each of two distinct points of content, and 1 point for providing the context as to why those are important points. Few of the questions have only one correct answer and it is not always obvious to candidates what we're looking for in an answer; but it is known in advance to the grading committee what is being sought, e.g. "perspective over plot".

2. interviews are in slack. We're a remote team who communicate 95% via writing, and it doesn't matter if you stutter or what you look like or how you present. Get comfortable in your PJ's with a cup of tea, or use your standing desk if that helps you feel sharper. It only matters how you communicate in writing.

Once again, the interviews are structured - "interviewer 2 will ask these 5 questions" and the criteria for success aren't "this woman was well spoken" but instead "she showed ability to think on the fly about something unexpected, demonstrated empathy in communication, and made a compelling argument for her proposed solution". Not quite as objective as the take home test, but still "gradeable". And, in case of a hard choice between two candidates, other stakeholders in the hiring process can review the transcript of the chat, providing their grade considering the pre-specified criteria, to help decide.

It's not ideal, but it is practical, allows for cooperative interviewing (anyone can help a candidate start the well-documented interview process; anyone can do any stage of the interview and know that we cover all the questions we want asked before the end), and encourages consistent judgment.

It's worked well for us so far and people have appreciated it; it also lets us give concrete answers to "what could I have done better?" to candidates we reject who ask for a follow-up. I feel pretty strongly that you spent your time working with us (particularly on the take home exercise) so I can spend time explaining how you could improve (often it is useful to other applications for other jobs: "your code should have comments to explain confusing things or highlight clever things", "consider using git for revision control rather than only committing your completed project", "you'd benefit from using a grammar checker"). We've had many repeat applicants who answer better with each iteration.