Check-In Bot Prompts That Get Honest Answers, Not "All Good"

When your team's async check-in returns "all good" every morning, you don't have a communication problem. You have a visibility problem. The team is filtering what reaches you, and you're making decisions on sanitized data.

Why "How are things going?" gets you "all good"

Open-ended check-in prompts reward performative answers. If a person types "all good," the check-in is complete. There is no friction to honesty, and there is social friction to admitting a stumble. Your bot is training the team to hide friction.

The fix is not a better culture talk. It is a better prompt.

What to ask instead: prompt patterns that work

Use questions that make a generic positive answer impossible or obviously false.

1. Forced trade-off Bad: "What are your priorities today?" Good: "If you could only finish one thing today, what is it — and what are you deliberately not doing?"

2. Yesterday fact-check Bad: "What did you do yesterday?" Good: "What did you plan to finish yesterday that is still not done? Why?"

3. Blocker-first framing Bad: "Are you blocked?" Good: "Name one thing that is slower than it should be right now, even if you are still working on it."

4. Dependency check Bad: "Is anything blocking you?" Good: "Who are you waiting on today, and what happens if they don't reply by 3 PM?"

5. Energy signal Bad: "How do you feel?" Good: "On a scale of 1–5, how confident are you that your current task will ship this week? What would move it to a 5?"

Copy/paste check-in bot template

Good morning. Please answer in one short message:

1. What is the one thing you must finish today?
2. What did you plan yesterday that is still open? (one line)
3. Name one thing that is slower than expected, even slightly.
4. Who are you waiting on, and by when do you need them?
5. Confidence 1-5 that this ships this week: __

This takes under three minutes. The key is question 3: it assumes slowness exists, so reporting it is normal, not a confession.

Tool tip (AiAdvisoryBoard.me): Run this check-in for one week and compare the answers to what your project tracker shows. The gap between what people report and what tasks actually moved is your Plan → Fact → Gap signal. See how the 7-day diagnostic surfaces that automatically. https://aiadvisoryboard.me/?lang=en

Manager scan (2-minute digest example)

As the owner, you should not read every individual check-in. You should read a rolled-up scan that highlights exceptions.

• Team members who answered "confidence 1–2" without naming a blocker.
• Repeated "still open" items from yesterday (3+ days in a row).
• Names appearing in "waiting on" across multiple people (a hidden dependency).
• Anyone who skipped the check-in without out-of-office notice.

If your scan is empty every day, your prompts are too soft. Tighten them.

How to introduce new prompts without mutiny

Announce the change before you ship it. Say exactly why: you want fewer status meetings, not more surveillance. Frame the bot as a tool that protects deep work by surfacing blockers early.

Start with two questions for one week. Add a third only when the team is answering honestly. If you dump five new fields on a Friday, you will get "all good" by Monday.

Internal links

If your team works in shifts or on the floor, pair this with a written check-in for frontline staff: /blog/written-shift-check-in-frontline-field-staff

For teams that need to coordinate handoffs across departments, see the waiting-on list framework: /blog/waiting-on-list-for-handoffs-between-departments

Micro-case (what changes after 7–14 days)

A 120-person logistics company replaced their open-ended standup bot with the forced-choice template above. Within two weeks, the owner saw three recurring "waiting on" entries pointing to the same procurement contact. That dependency had been invisible for months because no one volunteered it in a meeting. The founder called the vendor, unblocked two shipments, and stopped a Friday escalation before it reached the client.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30–500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

FAQ

Will people game forced-choice prompts? Yes, if you punish honest answers. The prompt only works if the manager responds to blockers with help, not blame.

How often should the bot run? Daily for execution teams, twice a week for strategic or creative roles. Cadence should match how fast work actually moves.

What if someone still writes "all good"? Reply publicly with a follow-up: "What is one thing that could go wrong with that task today?" Do it once or twice, and generic answers usually stop.

Should the owner read every check-in? No. The owner reads the exception scan. Individual check-ins are for the team lead or manager.

Honest async check-ins are not about catching people slipping. They are about giving the founder real signal before a small delay becomes a missed deadline. If you want that signal surfaced automatically across the company every day, see how the 7-day diagnostic works. https://aiadvisoryboard.me/?lang=en