Common Mistakes in Interviewing Whistleblowers—and How to Avoid Them

A whistleblower interview rarely collapses in one dramatic moment. It goes wrong earlier and more quietly: a call made from the wrong phone, a promise nobody wrote down, a source who assumed the story would run within a fortnight and instead waited five months. By the time the trouble becomes visible, the damage usually sits in the record — a name in an email chain, a document that can be traced back to one desk, a relationship curdled into suspicion.

These are the errors that come up most often in investigative work, and the habits that keep them from happening.

Protection is a plan, not a checkbox

The common mistake is treating anonymity as a yes-or-no question asked at the top of the first call. Protection is not a status you grant. It is a chain of small practical decisions, and it breaks at the weakest link.

Start by asking what the person is actually protecting themselves from. A regulator? A manager who will read the file? A small industry where everyone knows everyone? That answer shapes everything that follows. Then work through the mechanics together:

• How you contact each other, from which number or account, at what hours. One agreed channel, never a work phone, never a work laptop.
• What appears in your notes. Coded identifiers rather than names, and no identifying details in shared folders, calendars or messaging apps you use for everything else.
• Who inside the newsroom knows, and on what basis. A protective editor needs the facts. A curious colleague does not.
• Where recordings and transcripts live, who can open them, and whether the backups are as protected as the original.
• Whether anything in the published story could identify them by inference — job title, department, location, dates, a turn of phrase only one person would use.

Spell out what you cannot protect them from. You are not their lawyer and you cannot stop an employer from guessing. Protections for whistleblowers differ by country and by sector, and they change; before you promise anything on that front, speak to a lawyer or a specialist organisation. A short, honest sentence early on is worth more than a comforting one you cannot stand behind.

Consent is a conversation you keep having

A single yes at the beginning is not consent for everything that follows. Recording. Quoting by name. Publishing a document in full rather than summarised. Naming their department. Appearing on camera. Being identified to a second editor. Each is a separate decision and deserves its own clear question.

Two failures recur. The first is taking consent from someone frightened and hurried, who says yes to get off the phone. That is not consent so much as compliance; revisit it when things are calmer. The second is taking a yes and then escalating quietly — using the real name in a note, or passing the documents to a colleague without saying so.

Keep a short written record of what was agreed and when. Not a contract, just a line or two in your own notes. Circumstances change, and so does consent. Someone who was relaxed about being named in March may have a different view in June, and they are entitled to.

Expectations drift when nobody names them

Whistleblowers usually arrive with a story and an outcome in mind: an investigation, a dismissal, a policy quietly rewritten. The mismatch between that and your actual job is where most resentment is born.

You cannot promise outcomes, and you should say so plainly. What you can offer is narrower and more useful: you will report thoroughly, you will keep to the protection plan you agreed, you will tell them when publication is close, and you will put the allegations to the organisation before publication. What you cannot offer is control over timing, editing, legal review, or what anyone decides to do afterwards.

Explain the machinery too. Their account will be tested rather than transcribed. The organisation will get a right of reply. Lawyers may read the piece. Publication may take months, and there is a real possibility the story will not run at all. If you do not know when it will appear, say that instead of guessing. A source who was warned about the slow parts is far less likely to feel abandoned during them.

Taking the story instead of testing it

The craft error hides in plain sight: accepting the narrative as delivered. A compelling account, told with conviction, is exactly the moment to slow down.

Ask what records exist and who holds them. Ask who else was in the room, what the minutes say, what was said in the version sent to the board. Ask what the other side would say if you called them tomorrow — and watch how the answer is delivered, not just what it contains. Save your challenges for a second pass so the first telling stays intact and uncontaminated by your own theories.

Resist the urge to share what other sources have told you. It flatters the conversation and corrupts the evidence: memories adjust to fit what they think you want, and you can no longer tell which account was independent.

The vanishing act

The most frequent complaint from whistleblowers is not about accuracy. It is about silence. Four months of nothing, then a message asking for one more document, urgently.

Agree a rhythm at the outset and keep it, even when the news is that there is no news. A two-line message — the piece is with the lawyers, nothing is published yet, here is what I still need — costs you a minute and holds a fragile relationship together. The quiet stretch before publication is where trust most often dies, so warn them it is coming and say how long you expect it to last.

Plan the aftermath as well. Once the story is out, the person is often left alone with the consequences while you move to the next thing. Decide in advance who checks in, how, and after how long — and keep that contact as carefully protected as the first interview.

Handling the material as carefully as the person

Sources are usually protected at the interview stage and exposed later, through carelessness with files. Descriptive filenames. Recordings synced to a personal cloud account. Transcripts pasted into a shared document with a broad access list. Documents whose metadata still carries a name, a printer, a time stamp.

Treat every file as if the person it came from might one day have to explain how you got it. Strip metadata before publishing. Keep the raw material in one controlled place. And when you come to write, read back the passages that describe their role and any direct quotes, rather than sending the whole draft — they need accuracy, not editorial influence.

A short routine before the first call and before publication

None of this requires a formal protocol. It requires a few minutes of discipline at two moments.

1. Before the first call: agree the channel, the hours and what you will call them in your notes.
2. Ask what they are afraid of, and write the answer down where you will see it again.
3. Say what you can and cannot promise, in plain words, before they tell you anything they cannot take back.
4. Book the next contact before you hang up, and name the quiet period in advance.
5. Before publication: confirm consent again for each identifying detail, and read back the passages about them.
6. Before publication: check the files and the drafts for anything that narrows down who they are, then check the calendar of the person who is going to keep in touch afterwards.

Do that, and the dramatic failure you were worried about becomes much less likely. What replaces it is the ordinary, unglamorous work of being someone a nervous person can reasonably trust — which is most of what source protection actually is.

Photo: Miff Ibra / Pexels