Asking your business a question
Four questions a business asks on a Monday. None of them has a screen in any CRM, and all four are answered in a sentence each.
A manager's Monday morning is a handful of questions that no dashboard was ever built for, because each one is asked once. Building a report for them is not worth it; not answering them is worse. This is the gap the conversation fills.
Everything below is a real exchange with your own org, captured on 2026-08-19. The question is typed in English; what comes back is what came back.
“How is the pipeline looking?”
You: How is the pipeline looking — how much is sitting in each stage?
| Stage | Enquiries | Value |
|---|---|---|
| Contacted | 22 | ₹5,33,26,664 |
| New Enquiry | 18 | ₹4,83,05,696 |
| Appointment Booked | 14 | ₹3,08,89,481 |
Contacted
- Enquiries
22- Value
₹5,33,26,664
New Enquiry
- Enquiries
18- Value
₹4,83,05,696
Appointment Booked
- Enquiries
14- Value
₹3,08,89,481
That is the whole interaction. No chart, no saved query, no clicking through a filter builder — and nothing left behind afterwards, which is the point. A question asked once should not leave an artefact somebody has to maintain.
“Which of the open ones are big, and came from a referral?”
You: Which of the OPEN ones are worth more than fifty lakh and came from a referral?
One enquiry came back — Loan Against Property — Kabir Sethi, ₹77,64,592, sitting at Documents Received. That is a precise question with three conditions in it, answered without anybody constructing a filter.
Compare that with doing it on the board. Open the list, add a value filter, add a source filter, remember to exclude the closed stages — four interactions, and you will not do it again next week because it was tedious the first time.
“What has gone quiet?”
You: How many enquiries have gone quiet — nothing logged for a fortnight?
63 open enquiries have had nothing logged on them for a fortnight, carrying ₹15,72,69,968 between them.
Read that number again, because it is the one that changes a client meeting. It is not a projection or a health score — it is a count of cases in their own system that nobody has touched. Ask it of a client's org in a first meeting and you have said something about their business that they did not know, in under a minute.
You: Show me the five worst.
| Enquiry | Stage | Silent for |
|---|---|---|
| Home Loan — Nandini Oberoi | New Enquiry | 90 days |
| Loan Against Property — Rohit Kaur | Offer Sent | 88 days |
| Education Loan — Swati Agarwal | Documents Received | 88 days |
Home Loan — Nandini Oberoi
- Stage
New Enquiry- Silent for
90 days
Loan Against Property — Rohit Kaur
- Stage
Offer Sent- Silent for
88 days
Education Loan — Swati Agarwal
- Stage
Documents Received- Silent for
88 days
Same definition of “quiet” as the workflow in The morning sweep: what has gone quiet uses, for the same reason: a machine touching a record is not somebody working it.
“What do we tell someone who asks…?”
You: What do we tell someone who asks whether applying will hurt their credit score?
The top result is Will applying for a loan hurt my credit score? — out of the customer knowledge base, not out of the model. It is the same retrieval you read in What a knowledge base is, and how search finds things, reached from your own tools rather than from an agent.
Which makes it useful for something the product does not have a screen for: checking what the assistant would say. Before a client goes live, ask their knowledge base the ten questions their customers actually ask, and read what comes back. That is a QA pass, and it takes ten minutes.
What this is good for, and what it is not
| Reach for the conversation | Reach for a dashboard |
|---|---|
| A question asked once | A number somebody watches every week |
| An odd shape nobody filtered for | Something a client's team opens themselves |
| Checking something before a meeting | Anything that has to look the same next month |
A question asked once
- Reach for a dashboard
- A number somebody watches every week
An odd shape nobody filtered for
- Reach for a dashboard
- Something a client's team opens themselves
Checking something before a meeting
- Reach for a dashboard
- Anything that has to look the same next month
The rule, and it holds for the rest of the module: conversation for the thing you do once. If you find yourself asking the same question three weeks running, that is the signal to build it properly — and Building a dashboard by asking is how, by asking for that too.
Change one thing — ask five
- Ask your org the four questions above, in your own words rather than mine.
- Then ask a fifth that this course has not covered, about something you are curious about.
- Count how many of the five you could have answered from an existing screen. It will be one or two.
- Write down the one you would ask a real client in a first meeting.
Try it
- Ask a question you expect to have no good answer, and see how it tells you so.
- Work out what a read-only key cannot do, and check by asking it to change something.
Next: Changing things by asking — the same connection, with the scopes that let it write.
Related lessons
Base rates — what a piece of evidence is actually worth
A face-recognition system that is 99.9% accurate and almost entirely wrong, and a number that sent an innocent woman to prison. Both are the same arithmetic, and it is the arithmetic that decides what any piece of evidence is worth.
ReadConfirmation and survivorship — what you never looked for
Two questions about evidence you did not go looking for. One is a rule you have to discover, and one is a pattern in five famous people — and in both, the thing that would have told you the truth is the thing nobody checks.
ReadLoss aversion, sunk cost and regression — what it costs you
Four questions you answer about yourself rather than about a scenario, and your own answers are the finding. Then the pattern that makes praise look useless and criticism look like it works, whatever you actually do.
Read