SELF-GUIDED BA PRACTICE Packages available · Access after confirmed paymentSee packages ↗

Requirements

Requirements elicitation techniques: 8 methods and when to use each

Requirements · Published · Updated · 8 min read · By BA Mentorship Editorial

Part of our guide to business analyst roles and responsibilities.

Why elicitation is harder than it looks

Elicitation sounds like simply asking people what they want. In practice, that rarely works well on its own. People forget routine steps because they are so familiar. They describe solutions ("I need a button") rather than needs. They tell you the official procedure rather than the workaround they actually use. Different people give conflicting answers, and some important people never volunteer anything.

That is why we use the word elicit, which means to draw out, rather than gather, as if requirements were lying around waiting to be collected. A good analyst chooses methods that fit the situation and uses more than one to cross-check the picture. This article describes eight common techniques, when each shines, and how to combine them. It builds on the earlier steps in stakeholder analysis, because who you talk to determines what you learn.

Eight techniques at a glance

TechniqueBest forMain risk
1. InterviewsIndividual detail, sensitive topics, expert knowledgeOne person's view; leading questions
2. WorkshopsResolving disagreements, agreeing a process, prioritisingDominant voices; poor preparation
3. ObservationSeeing how work really happens, finding workaroundsPeople change behaviour when watched; time-consuming
4. Document and system reviewLearning rules, policies and the current system quicklyOutdated or aspirational documents
5. Data analysisTesting claims with numbers, finding patternsPoor data quality; misreading
6. Surveys and questionnairesMany people, simple questions, geographically spreadLow response; shallow answers
7. PrototypingMaking abstract ideas concrete, testing screensMistaking a sketch for an agreed design
8. Focus groups and brainstormingGenerating ideas, exploring user attitudesGroupthink; ideas without evidence

The next sections describe the ones analysts use most, and then show how to combine them.

Three overlapping sources: what people say, what they do and what the records showThree overlapping circles. What people say: interviews, workshops and surveys. What people do: observation. What the records show: documents and data. Where all three agree, in the centre, is what you can trust.Whatpeople sayinterviews,workshops,surveysWhatpeople doobservationWhat records showdocuments and dataTrust
Cross-check every important claim. Combine at least two techniques so what people say can be checked against what they do and what the records show.
Diagram in words
  • What people say: interviews, workshops, surveys and focus groups.
  • What people do: observation of real work.
  • What the records show: document and system review, and data analysis.
  • The centre, where they agree, is what you can rely on.

Interviews: the workhorse

A one-to-one interview is the technique used most. It is flexible and lets you follow interesting threads. The quality depends on preparation and questioning.

Before the interview

  • Know the person's role and what you hope to learn from them.
  • Prepare five to eight open questions, but be ready to depart from them.
  • Send a short note about the topic and how long it will take, so they can think.

During the interview

  1. Start broad. "Tell me about your role and how this process fits into your day."
  2. Ask for examples. "Walk me through the last time that happened." Examples reveal reality; generalisations reveal the official story.
  3. Use open questions. "How do you...?" "What happens when...?" "What is the hardest part?"
  4. Probe gently. "Can you say more?" "Why is that?" A short silence often brings out more than another question.
  5. Avoid leading questions. "Wouldn't a barcode scanner help?" invites a polite yes. Ask "What slows you down?" instead.
  6. Summarise. Repeat back the key points and ask "Have I got that right?"

After the interview

Write up notes within a day, mark open questions and send a short summary for confirmation. Confirmation builds trust and catches misunderstandings while they are cheap to fix.

Ask

  • “Tell me about your role and how this process fits into your day.”
  • “Walk me through the last time that happened.”
  • “What slows you down?”
  • “Have I got that right?”

Avoid

  • “Wouldn’t a barcode scanner help?”
  • “You always do it this way, right?”
  • “So you need a portal?”
  • Filling a silence with another question
Open questions reveal reality; leading questions invite a polite yes.

Workshops: getting people into one room

A workshop brings several stakeholders together to achieve a defined result, such as agreeing a process or ranking features. It is efficient when stakeholders disagree, because the disagreement is resolved in the room.

How to run a good one

  1. Define the outcome. "By the end we will have an agreed process map and a list of the top five problems" is better than "discuss the process".
  2. Choose the right people. Include those who do the work, those who decide and those who are affected. Keep the group small enough that everyone can speak.
  3. Prepare an agenda and materials. A draft diagram or a set of cards gives people something to react to.
  4. Facilitate, do not dominate. Use turn-taking, parking-lot lists for off-topic ideas, and ask quieter people directly.
  5. Record decisions and actions. Capture them visibly during the session.
  6. Send minutes quickly. The same day if you can. See meeting minutes.

The most common failure is turning up with no preparation and hoping the group will self-organise. The second is letting the most senior or loudest person decide everything. A good facilitator protects the quiet expert who knows the real process.

Observation, documents and data: checking what people say

Observation

Watching work happen is the best way to find the gap between the official procedure and reality. Sit with a user for an hour, ask permission to watch and take notes without interrupting. You will see workarounds, shortcuts, sticky notes, duplicated data entry and interruptions that nobody mentioned. Afterward, confirm what you saw: "I noticed you retype the order number in the second system. Is that always necessary?" Observation takes time, so use it where the process matters most.

Document and system review

Policies, procedure manuals, forms, previous project documents and screenshots of existing systems give you a quick start. They tell you official rules and vocabulary. Always ask: "Is this still how it works?" Documents age, and the real process often differs.

Data analysis

Numbers turn opinions into testable claims. If someone says "most customers cancel late", look at cancellation times. Use simple queries and counts; see SQL for business analysts. Data also reveals patterns nobody described, such as one region creating most of the errors. Remember that data has limits; check data quality before relying on it, and treat what you find as evidence to discuss, not as the final truth.

A worked example: returns at Brightpath Logistics

Brightpath Logistics is a fictional delivery company. The operations manager says, "Our returns process is slow and customers complain. We need a returns portal." The analyst, Leila, plans her elicitation rather than starting with the portal.

The plan

Question to answerTechniqueWho or what
How does a return actually flow today?Observation, then a process workshopTwo depot shifts; depot staff, support agents
Where is time lost?Data analysisThree months of return records
What do customers find frustrating?Review of complaint logs; short surveySupport complaint records; recent returners
What are the rules for refunds?Document review; interviewRefund policy; finance lead
What would a better flow look like?Prototype and workshopSketched screens; depot, support, finance

What she learns

  1. Observation shows that depot staff photograph the return label with a phone and email it to support, because the system has no field for the return reason. Nobody had mentioned this.
  2. The data shows that the longest delay is between "parcel received at depot" and "refund approved", not the courier time that the manager assumed.
  3. The complaint log shows that many complaints are about not knowing the status, not about the refund amount itself.
  4. The finance lead explains that refunds above a set amount need a second approval, which explains some delays.
  5. In the workshop, the group agrees that a status message and a return-reason field would fix most problems, and that a full portal is not necessary yet.

The result is a smaller, better-targeted project than the one first requested. Notice that no single technique would have found all of this. Observation found the photo workaround, data located the real delay, documents explained the approval rule and the workshop created agreement.

The request

  • “We need a returns portal.”
  • Assumed: the courier is the slow part

What elicitation found

  • Observation: label photos emailed to support
  • Data: the delay is between depot receipt and refund approval
  • Complaint log: customers want to know the status
  • Finance: a second approval above a set amount

What was agreed

  • A status message
  • A return-reason field
  • No full portal yet
A smaller, better-targeted project. No single technique found all of this.

Combining techniques and avoiding common traps

  1. ReadDocuments and data prepare informed questions.
  2. ObserveSee how the work really happens.
  3. InterviewFill in detail and reasons.
  4. WorkshopResolve what remains and agree.
A reliable pattern. Use prototypes when talk becomes abstract, and surveys when you need breadth.

A reliable pattern is: read first, observe second, interview third, workshop last. Documents and data prepare you to ask informed questions; observation shows reality; interviews fill in detail and reasons; a workshop resolves what remains and builds agreement. Use prototypes when talk becomes abstract, and surveys when you need breadth.

Traps to avoid

  • Taking the first answer as the requirement. Ask why until you reach the underlying need.
  • Talking only to managers. People who do the work know details managers do not.
  • Confirmation bias. Do not look only for evidence that supports your favourite solution.
  • Skipping the write-up. Unwritten conversations are forgotten or remembered differently.
  • Treating prototypes as designs. Label them as illustrations to prompt feedback.
  • Ignoring the quiet stakeholder. Regulators, support teams and night-shift staff are easy to miss.

Once you have gathered information, turn it into structured requirements: user stories with acceptance criteria, or the documents described in BRD, FRD and SRS. To practise the skill, interview a friend about how they plan their week, then observe what they actually do. The BA Lab has an elicitation stage where you prepare questions and review meeting notes against a rubric, and the free worked project guide shows a stakeholder claim being tested against evidence.

Frequently asked questions

What is the best requirements elicitation technique?

There is no single best method. Interviews are the most common, but combining at least two, such as interviews with observation or data review, gives a more reliable picture of what is needed.

What is the difference between elicitation and requirements gathering?

They are often used interchangeably. Elicitation emphasises drawing out needs that people may not state clearly, rather than just collecting what is offered.

How many interviews do I need?

Enough to cover each stakeholder group and until new interviews stop revealing new information. For a small process, that may be a handful.

How do I run a workshop with conflicting stakeholders?

Define a clear outcome, prepare evidence, use a neutral facilitator, record disagreements openly and escalate unresolved points to the person accountable for the decision.

Should I record interviews?

Only with explicit permission, and in line with your organisation's policies. Many analysts take notes and send a written summary for confirmation instead.

This article is educational. All companies, people and numbers in the examples are fictional. BA Mentorship does not issue professional certifications and cannot guarantee any job or interview outcome.

Key terms in this article

Browse the full glossary

CHECK BEFORE CONTINUING

Keep your work safe