I stole this from a founder in Palo Alto who used to run it with his executive team quarterly. Five questions, twenty minutes, once every ninety days. I have run it with every direct report I have had since 2019, and it has caught more problems than every engagement survey I have ever seen, combined.
It is not an engagement survey. It is not a performance review. It is a working diagnostic that the operator runs with each direct report every 90 days, and the whole point is to catch the two or three things that are going to become a resignation letter six months from now, while there is still time to do something about them.
Why the engagement survey does not do this work
Engagement surveys are built for HR. This check is built for the operator. That distinction matters, because the two tools measure different things and are used at different cadences by different people for different purposes.
An engagement survey aggregates. It asks fifty people about their satisfaction and rolls up a number. The number tells you the temperature of the room but it does not tell you which person is about to walk. The names are anonymized. The action item, if there is one, arrives in a slide deck three months after the survey closed.
The quarterly check does not aggregate. It is one operator, one direct report, five questions, twenty minutes. The whole point is the specific person. The whole action item is what you and that person do together next week. You do not need an HR budget, a survey vendor, or a slide deck. You need a calendar and the discipline to run it every quarter.
The engagement survey is a temperature gauge. The quarterly check is a stethoscope. Both are useful. Only one of them tells you which chest to listen to.
An engagement survey tells you the room is hot. A quarterly check tells you which person is running the fever.
The five questions
The questions are simple. Their power is that they are the same five questions every quarter, so you can see the drift across a year in the same person's answers. Consistency is what makes the diagnostic work. Change the questions and you have thrown away the baseline.
Fig. 1 · Five questions. Same order. Every quarter.
1. What is one thing you are proud of from the last 90 days?
Start here. You are calibrating on what the person values in their own work, which is often not what you thought you were assigning them. Pay attention when the thing they are proud of is not on your priority list, and pay closer attention when the thing you thought was their biggest win is not on their list at all. That gap is the first drift signal.
2. What is one thing that took longer than it should have, and why?
Not "what went wrong." That question gets defensive answers. "What took longer than it should have" gets diagnostic answers, because everyone can name something that dragged. The "why" is the interesting part. Nine times out of ten the answer is a decision that was pending somewhere else, a tool that was missing, or a hand-off that was unclear. All three of those are fixable, and all three are your job.
3. If you were me, what would you change about how our team runs?
This is the question that scares operators the most and does the most work. You are inviting a subordinate to critique the system you built. Most of the time you will get a hedge the first quarter, a real answer the second quarter, and something valuable by the third quarter. The trust compounds. The first honest answer usually names something you already suspected but had not wanted to look at.
4. Is there anyone on our team you think is being under-used?
This one catches promotions before you would have thought of them. Direct reports see peers with a clarity you do not have from where you sit. Nine times out of ten they will name someone you have been quietly wondering about. Once every twenty times they will name someone who has been in plain sight and who you completely missed. That one time is worth the whole exercise.
5. What would need to be true for you to still be doing this job with joy in a year?
The hardest question and the most important. You are asking, in one sentence, what it would take to keep them. If the answer is "nothing, I am good," believe them and move on. If the answer names something specific, write it down word for word, then treat that specific thing as a working item on your list for the next 90 days. The answer to this question is the one thing that separates an operator who retains their best people from an operator who is constantly surprised by resignations.
The format: async or 20-minute sync
You have two options and they both work. Pick the one that fits the person.
Async: send the five questions in a document on Monday. Ask for a written response by Friday. You read it over the weekend. You bring one specific followup to Monday's one-on-one. Total cost: 30 minutes of your time, 45 minutes of theirs, spread across a week they were going to work anyway.
Sync: block 25 minutes on a Wednesday afternoon. You ask the questions in order. They talk, you take notes, you do not react in the moment. At the end you name the one thing you heard that you want to follow up on. Total cost: 25 minutes of both your times, in one sitting.
I have run both. Async gets more honest answers on question three, because there is no in-the-moment social pressure to hedge. Sync gets more honest answers on question five, because tone of voice matters and you cannot fake tone in email. If I had to pick one, I would run async for the first two quarters of the year and sync for the second two. If you cannot pick, run async. It scales further and it does not require the direct report to compose in real time.
The one thing you must not do is run the check inside a normal one-on-one. It has to be labeled as its own thing, with its own calendar block, its own document, and its own protected time. If you fold it into your weekly, it becomes background noise and it stops working.
What to do with the answers
The check is worthless if the answers sit in a document. Here is the sequence I use in the week after each round.
Fig. 2 · Read, sort, act, close. Ten days end to end.
Read all answers by Sunday of the same week. Not next week. This week. If you cannot read five short documents in a weekend, you are running too many direct reports and the check is not the problem you have.
Sort what you heard into three buckets. Bucket one: things I can fix this quarter. Bucket two: things I need to raise up the chain. Bucket three: things I need to sit with. Then act on bucket one inside two weeks, escalate bucket two inside a month, and diarize a followup on bucket three for next quarter's check.
Close the loop with each person by naming one thing you heard from them that you are going to act on. Do this in person, not by email, and do it inside 10 days of the check. The whole diagnostic dies if the direct report cannot see a line between what they told you and what you did about it. Once you close the loop three times in a row, the answers get more honest every quarter.
A story from Zareen's
The first time I ran this at Zareen's, six months into the fractional operations role at the $30M Michelin-recognized Bay Area group, I got an answer to question five from a sous chef that I want to reproduce here without the name. Question: what would need to be true for you to still be doing this job with joy in a year? Answer: "I would need someone to ask me if I want to run a kitchen someday, and mean it."
Nobody had asked her. She had been holding the pastry program together across two locations for eighteen months and was already interviewing at a competitor when we sat down. We spent the next 90 days building her a real path to a chef de cuisine role. She stayed. She is still there. She is running her own kitchen now. The whole thing cost me one 20-minute conversation and the willingness to take the answer seriously.
That is the ROI on the quarterly check when it works. One question, one direct answer, one 20-minute conversation, and a person you were about to lose stays and grows into a role you needed filled anyway.
The direct report who is about to resign is almost never the one you thought. The quarterly check tells you who it actually is, in time to do something.
What breaks the check
I have seen this go sideways three ways, and I have caused two of the three myself.
First: the operator argues with the answers. The moment you push back on what someone told you in question three, you have killed the diagnostic for a year. The correct response to anything you hear is "thank you, tell me more." Debate the answer next quarter if you want. In the room, listen only.
Second: the operator promises action they cannot deliver. If a direct report tells you they are frustrated by a decision that is out of your control, do not promise to fix it. Say what you can do, name what you cannot, and be specific about which is which. Overpromising in a check that is designed to measure trust destroys the trust the check is designed to build.
Third: the operator skips a quarter. This one is the most common and the most fatal. Miss one quarter and the pattern feels optional. Miss two and it is over. If the quarter gets away from you, run the check late rather than skipping. Ninety days plus two weeks is fine. Ninety days plus ninety days is a different problem, and the person you were trying to retain is already interviewing.
The mistake I made for years
For most of my career I ran engagement surveys and called it employee listening. It was not listening. It was measurement. The two are different jobs. Measurement tells you the size of the problem. Listening tells you the shape.
The quarterly check is listening. It is one operator, one person, five questions, twenty minutes, every 90 days, forever. It does not replace the engagement survey. It is the tool you actually use to keep the people the engagement survey told you were at risk.
Once I started running this with every direct report at Zareen's across the 215-person workforce, the trailing twelve month retention on my senior team went from something I was quietly worried about to something I stopped tracking as a risk. Not because the check retained them, but because the check told me, in time, what to change to retain them.
The point
You do not need a new tool. You need a 20-minute calendar block, five questions, and the discipline to run it four times a year with every person who reports to you. That is the entire program. It is not sexy. It does not require software. It is not going to show up in a leadership book.
It is going to catch the resignation letters six months before they get written, and it is going to tell you which of the people you thought were fine are actually running out of patience, and it is going to give you a chance to do something about it while there is still something to do.
Cadence beats charisma. Run the check.