Preparing content…

One coworker points to a data point while another sketches a decision beside the same report
情绪与沟通

How to Clarify Vague Requirements: Six Questions Using MBTI Sensing and Intuition

Clarify how a report is used today and what it should help people decide. Turn six questions into observable acceptance checks using a refund-report example.

By: FermatMind Editorial

Published: Nov 17, 2026

Updated: Nov 17, 2026

23 min read

Content review completed · Nov 17, 2026

Subscribe to the English blog (Atom) · Add this link to an RSS reader that supports Atom for newly published articles and substantive updates.

When someone says, “Make this report more useful,” ask how they use it today and what decision the new version should help them make. The six questions below use two MBTI perspectives: Sensing for facts and constraints, and Intuition for patterns and possibilities. Put the answers into acceptance criteria someone who missed the meeting can check.

Six questions to ask when requirements are unclear

What happens now?

  1. Who uses this report, and when?
  2. Where did they get stuck the last time?
  3. Which data, deadlines, and existing processes must stay?

What should the report help people do?

  1. What decision should the new version help you make?
  2. Which changes call for action, and which just need watching?
  3. What should a colleague who missed the briefing be able to do independently with one normal sample and one abnormal sample?

Send back a short note covering the use case, the decision, test examples, and open questions with the person who will confirm them. Leave undecided conditions marked as pending.

In the MBTI framework, S (Sensing) favors concrete facts and experience. N (Intuition) favors patterns, connections, and possibilities. This article uses those perspectives to clarify a request. One person can ask both sets of questions without taking a test or knowing anyone’s type. Myers & Briggs Foundation on the preferences

Ask how they used it last time, then what they need to decide next time

This is a fictional example. A regional operations lead asks in a group chat whether an analyst can make the weekly after-sales report more useful. The analyst has orders, refund details, and channel totals, but the request could mean any of three things: make the report easier to read, expose channels with missing data, or show whom to contact. Drawing a chart first makes the first interpretation the default. Instead, ask the lead to retrace the last use and explain what they need to decide next time.

You could start: “Is this still for Monday’s after-sales meeting? Could you walk me through last week’s report and show me the step that took the longest? What would you like to decide as soon as you open it next time?”

QuestionAnswer in this exampleEffect on the work
S1. User and occasion. Who opened it last time, when, and how did they use it?At Monday morning’s regional after-sales meeting, operations flags channels and assigns investigation tickets.The opening view supports that decision. Leave the raw details for follow-up.
S2. Where it gets stuck. Can you point to one row? What else did you have to open?The report has refund counts but no paid-order denominator. The team opens the orders table to gauge the scale.Show refund counts, paid orders, refund rates, and the relevant weeks together.
S3. Scope and constraints. Which fields will be ready when? What must stay?Start with four channels. Take a snapshot before Monday’s meeting. Finance keeps its old export.Retain the finance export, mark the snapshot time and weeks, and keep a route to the details.
N1. Decision. What could you decide from the new opening view that you can’t decide today?Which channel to investigate first and whom to contact.Show the flagged channel and responsible role, not just a trend chart.
N2. Patterns and exceptions. What warrants a ticket? What looks large but only needs watching?Prioritize a rise of at least 2 percentage points if there were at least 100 paid orders this week. Watch smaller samples.Check both the increase and the order count. These cutoffs are fictional team choices.
N3. Verification. What should a colleague who missed the discussion be able to do?From fixed examples, identify which channel to investigate first and explain which cases are watch-only or not comparable.Cover priority, below-threshold changes, small samples, missing data, and zero denominators.

Check your understanding: “What you need is to decide which channel to investigate first, not to turn the refund count into a line chart, right?” If the regional total matters more, the team can correct the priority before debating layout.

Turn the answers into acceptance checks a teammate can run

Use payment time to assign orders to two adjacent full calendar weeks, Monday through Sunday. The numerator counts unique orders with a successful refund by the agreed Monday snapshot. Multiple successful refunds on one order still count once. The denominator is unique paid orders in that channel. Earlier orders have had longer to accumulate refunds, so these rates are cumulative at the same snapshot. Use equal refund observation windows if the team wants to compare channel performance or longer-term trends.

Copy the table into the requirements document and confirm the numbers and rules with the business owner before making them the acceptance baseline.

Acceptance checkAgreement in this exampleHow a teammate checks it
User and decisionMonday after-sales meeting. Decide which channel to investigate first from the opening view.State the channel, reason, and responsible role without another table.
Fields and definitionsChannel, week, refunded and paid order counts, rate, change, status, and responsible role. Mark the snapshot and dates. Use the channel-owner list confirmed by operations. If there is no match, show “Awaiting operations confirmation.”Spot-check order keys and dates. Multiple refunds on one order count once.
Missing data or zero denominatorChannel C’s refund feed has not arrived, or a week has zero paid orders.Show the relevant status, leave the rate and comparison blank, and skip thresholds until the data owner confirms.
PriorityChannel A: 4/200 = 2% last week, 8/200 = 4% this week. Up 2 percentage points, with 200 ≥ 100.Show “Investigate first.” Operations assigns a ticket to the channel owner.
Small sampleChannel B: 1/25 = 4%, then 2/25 = 8%. Up 4 percentage points, but 25 < 100.Show “Small sample: monitor.” Do not add it to this priority queue.
Routine monitoringChannel D: complete data, 4/200 = 2%, then 7/200 = 3.5%. Up 1.5 percentage points, below 2.Label it “Routine monitoring” and keep it for comparison next meeting. Do not open a ticket automatically.
ScopeFour channels, existing finance export, no mobile version or new inputs yet.Use the same dataset across channels and confirm that finance can still get its export.
Sign-offOperations confirms escalation rules. A developer or analyst runs the fixed examples with a colleague who missed the discussion.Compare what that colleague would do with the expected action before delivery.

A move from 2% to 4% is a rise of 2 percentage points and a 100% relative increase. Multiplying last week’s rate by 1.02 sets a very different threshold. Test the inclusive boundary too: exactly 2 percentage points and 100 orders should qualify. Compare unrounded rates so values that both display as 2.00 percentage points do not get treated alike if they fall on opposite sides.

What you miss by asking only one set

Only current use and constraints. You may know the meeting time, missing field, and export requirement, yet produce two red rows that leave the chair unsure whom to call. Ask what requires immediate action and how to handle small samples.

Only purpose and possibilities. A striking dashboard can flag an anomaly that is really a delayed extract or a disagreement about which date defines a week. Ask to see a real row and agree on the snapshot.

Someone suggests the goal but nobody has approved the threshold. Make a reversible demo, mark the priority rule as pending, and name who will confirm it and when. Until then, keep the example’s 2 percentage points and 100 orders out of automated ticketing. Conflicting departmental targets need an owner who can decide whom this version serves, not a guess about anyone’s type.

What if the people making the request disagree?

Operations wants a priority list before Monday’s meeting, but the data team says one channel often sends refunds later. Ask first what is actually ready: are paid orders complete, and is the refund feed missing or complete with zero refunds? Then ask whether the meeting must decide on that channel or whether operations can open a ticket in the afternoon. Waiting for every channel delays the whole report. Acting on missing data risks a false flag.

Record three things: confirmed, other eligible channels can be compared; pending, show the missing-data status for the affected channel, leave it out of ranking and schedule a check; decision owner, data confirms arrival and operations decides whether to wait and who takes over. If completeness is still uncertain in the afternoon, notify operations and leave the channel pending. Do not turn blank into zero.

When the request includes “more useful” and “half the work,” ask whether the priority is time spent screening at the meeting or time spent investigating orders afterward. Agree which one determines acceptance. For meeting triage, test whether the reader can find the channel and its owner. For investigation time, agree in advance how to record it across comparable tasks. One short meeting is not evidence of less follow-up work.

Ask someone who was not in the meeting to try it

Show only the mock screen. Ask what they would check first, what cannot yet be compared, and whom they would contact. Don’t prompt them with cutoffs or colors. If they interpret “Awaiting operations confirmation” as already assigned, clarify the status or show where to report the gap. If they identify the right channel without being able to tell sample size from change threshold, put the reason next to the action.

Change one condition on each of three examples:

  1. Remove A’s owner. It should still be a priority. Operations fills in the owner before assigning it.
  2. Give C paid orders but no refund feed. Show missing data and skip the comparison.
  3. Give C complete data and 100 paid orders in both weeks, with one successful refund last week and zero this week. The rate is 0%, down 1 percentage point, and belongs in routine monitoring.

The last two examples both show no current refund records, but only the third establishes a true zero. Have the tester point to the row and column that support their action. If you need to explain a label aloud, put that explanation on the screen or in the acceptance notes and ask a new tester. This checks the requirements, not anyone’s information-processing preference. The same questions work for an approval form, a roster, or a support handoff: start with the last use and the next decision, then write the information and exceptions down.

If you only have two minutes, ask where the last attempt got stuck and what decision the reader wants to make next time. When answers differ, record who said what, which day’s data they mean, and who can decide. An existing report or lightweight mockup often helps more than debating the word “useful.” Use fictional or authorized examples whenever the data, permissions, or key rules remain unconfirmed.

If you want to observe your own habits, note the first question you naturally ask and which one you tend to miss. The FermatMind MBTI test offers 93- and 144-question versions with free completion and basic results; what appears depends on the version. If you already have results, see the guide to the four preferences and the full report. If disagreement arises after delivery, see how to check the facts and agree on the next step when giving feedback.

If you like clarifying a meeting, assigning follow-up, and tracking it to completion, the project management specialist profile covers confirming goals and scope with stakeholders, tracking schedules, and coordinating delivery. The 94-question FermatMind career assessment can help you reflect on the activities and working conditions you prefer. Check what you learn against a job posting or a small trial task.

Frequently asked questions

How do I turn a vague request for a chart into acceptance criteria? Ask what someone should be able to decide after seeing it. Pick a real example and write the expected action. If an order needs follow-up, specify which record must be visible and who should act. If the goal is spotting a multiweek pattern, specify the dates and what comparison the chart must make clear.

What if two stakeholders disagree on a report requirement? Ask them to run the same sample and explain what each would do. If after-sales needs payment-date grouping while finance needs refund-date reconciliation, name both time bases and have each group approve one example for its own purpose. If there can only be one default view, record who decides which audience it serves.

Related tags

情绪与沟通

Go to topic hubs

Continue from the article into a more structured topic entry surface.

Take the test

If you want to turn reading into self-measurement, continue into an assessment.

About FermatMind

FermatMind helps young people everywhere understand themselves and explore careers. We started in the Chinese-speaking world. Explore your preferences with MBTI and your interests with the Holland assessment, then use more than 1,000 occupation profiles to compare tasks, skills, and entry paths and find a next step worth trying.