Plan and run a stakeholder interview that uncovers real needs, then confirm what you heard.
Why it matters
Requirements are rarely sitting in someone's head ready to be written down. People describe symptoms, jump to solutions and forget the steps they do automatically. Elicitation is the skill of drawing out what people actually need. The word is deliberate: you elicit requirements, you do not just gather them like apples.
Elicitation techniques
| Technique | Good for | Watch out for |
|---|---|---|
| Interview | Depth with one person or a pair | One view; schedule several |
| Workshop | Agreement across a group | Loud voices dominating |
| Observation (job shadowing) | Seeing the real process and workarounds | People behave differently when watched |
| Document analysis | Rules, forms, policies, old specs | Documents are often out of date |
| Survey | Many people, simple questions | Shallow answers, low response |
| Prototype | Reactions to something concrete | People fix on colours, not flow |
The interview is the workhorse. Almost every analysis starts with a handful of them.
Planning an interview
- Pick the right people. For Riverside Library: the head librarian (sponsor), two front-desk librarians (users), one member who books often, and whoever looks after the library's IT.
- Set a goal. "Understand how a room booking happens today, from first call to the meeting itself, and where it goes wrong."
- Do your homework. Read the paper diary, any booking rules and complaint emails. Do not waste the interview on facts you could look up.
- Write a question guide, not a script. Group questions by theme and keep it to about ten.
- Book 30 to 45 minutes and say why you are there and how the notes will be used.
Asking good questions
Start broad and open, then narrow down.
- Open questions get stories: "Walk me through the last booking you took."
- Probing questions dig into the story: "What happened next?", "How often does that happen?", "What do you do when the diary is not at the desk?"
- Closed questions confirm facts: "Is the maximum booking two hours?"
- The magic wand question surfaces goals: "If you could change one thing about bookings, what would it be?"
- The final question catches what you missed: "What should I have asked you that I did not?"
Ask about real recent examples, not general opinions. "Tell me about Tuesday's double booking" beats "Are there problems with bookings?"
A worked fragment
Analyst: Walk me through the last booking you took. Librarian: A lady from the knitting group rang. I checked the diary, Room B was free on Thursday, so I pencilled it in. Analyst: What happens after you pencil it in? Librarian: I am supposed to ring back to confirm, but if it is busy I forget. Analyst: How often would you say that happens? Librarian: Maybe once a week. And then they turn up and someone else is in there, because my colleague took a call for the same room.
In two minutes you have found two problems (no confirmation, two people writing in one diary) and a rough frequency. Notice the analyst never suggested a solution.
Listening and noting
- Listen more than you talk. Aim for the stakeholder speaking 80 percent of the time.
- Capture their words. "I forget to ring back" is better evidence than your paraphrase.
- Separate facts, opinions and solutions. "We need text reminders" is a solution; the need underneath is "members must know their booking is confirmed".
- Note rules and numbers as you hear them: opening hours, room capacities, how far ahead people can book.
After the interview
Within a day, write up the notes and send a short summary back for confirmation: the process as you understood it, the problems, the rules and any open questions. This catches misunderstandings early and shows the stakeholder their time mattered. Then compare interviews: where two librarians describe the same step differently, you have found something to investigate.
Common mistakes
- Leading questions. "Would an app make this easier?" invites yes.
- Interviewing only managers. They describe the process as designed, not as performed.
- Taking solutions at face value. Always ask "what would that let you do?"
- No confirmation step. Unconfirmed notes become requirements nobody agreed to.
Quick recap
- Elicitation draws out needs; use interviews, workshops, observation and documents together.
- Prepare a goal and a short question guide; ask about real recent examples.
- Move from open to probing to closed questions; end with "what did I miss?"
- Write up and confirm within a day.
Try it
Open the Do tab and run a real 20-minute interview with someone about a process they use.