Answering a Journalist When Some of the Evidence Is Missing

A journalist asks three questions in one sentence. The reply answers none of them ("customer service remains our priority"), and the story runs with "declined to say". The team could have confirmed the part that mattered most an hour earlier.

When some evidence is missing, split the request into its separate questions, answer the parts you can confirm, state the limit of the rest precisely, and put a named person on a follow-up time. Never fill a gap with confident language. Reporters are watching for it: in Cision's 2025 State of the Media survey↗, 72% of journalists said they worry about factual errors in AI-generated PR content¹, and in its 2026 survey↗ 53% opposed AI-generated pitches, citing accuracy and personalization².

"Has your delivery service stopped operating in the region, how many customers are affected, and why were they not told?" This is a fictional journalist's question. It asks about service status, scale and communication, while also carrying a premise about notification. A reply that says "customer service remains our priority" leaves the factual questions untouched.

You may have a service notice for one city, an incomplete incident log and a spokesperson waiting for a draft. Begin by establishing what each document can answer. The deadline matters, but it does not turn an estimate into a confirmed count or make a partial service notice representative of an entire region. Newsrooms apply the same priority to their own work: the BBC's editorial guidelines↗ state that in news content, achieving due accuracy is more important than speed³.

Split the request into its separate questions, record what the team holds for each, and answer only the parts the evidence covers.

The discipline is the same as turning a journalist's request into a brief someone can write from. Preserve the original request, the journalist's outlet, the deadline with time zone, the requested format and the intended attribution. Clarify an ambiguous region or time period promptly, while continuing to check the parts that are clear. If background or an embargo is proposed, confirm the terms with the journalist before relying on them. Do not infer an agreement from your own email label.

Split the request into individual questions: Is service stopped? Where? During which period? How many customers meet the definition of affected? What notifications were sent, and to whom? Keep any causal premise visible as a question to investigate. A notification may have been sent but not received, or the record may be incomplete. Those possibilities need evidence before they become an explanation.

For each question, record a provisional answer, source passage, fact owner, evidence gap and public-use status. A task brief or the question itself cannot supply factual evidence. Keep the spokesperson's approved voice examples available for the eventual draft, alongside the organization's current position and disclosure instructions.

Each gap needs different wording: state an absent fact as unconfirmed, hold a conflict until its owner reconciles it, and withhold restricted information without pretending it is unknown.

It answers the confirmed part first, names what is still being checked and commits a named person to a follow-up time.

For the fictional inquiry, imagine that the operations owner confirms a temporary suspension of new bookings in one city. Existing deliveries continue. The notification review and customer count remain incomplete. If those facts and their disclosure are approved, answer the service question directly before explaining the remaining gaps. Avoid importing the journalist's regional premise into your own statement. Prepared lines help only when each is tied to evidence, which is the work of turning talking points into supported answers.

Illustrative draft for approval: "New delivery bookings in [city] are temporarily paused as of [confirmed time and date]. Existing deliveries are continuing. We are checking the number of affected customers and the notification record, and cannot confirm those details yet. [Named contact] will update you by [agreed time], including if those checks remain incomplete." Every bracket needs a confirmed value before sending.

The follow-up sentence promises an update, which the named owner must be able to provide. It does not promise that the investigation will finish by then. If even an update time cannot be agreed internally, say what contact route is available without inventing a deadline. When several outlets are asking the same thing, a holding statement written while the facts are still arriving can answer all of them consistently. A useful response can contain confirmed facts and a clearly bounded gap.

Question by question, in the journalist's order, with the fact owner checking support and the spokesperson approving the attributed wording.

Keep each approved question beside its proposed answer during internal review. Check that the reply addresses the question asked, that its geographic and temporal scope is clear, and that it avoids unsupported assurances. Phrases such as "no one was affected" or "all customers were informed" need their own evidence. A sympathetic tone cannot resolve an unsupported absolute. Apply the claim-by-claim review you would give a press release to each sentence.

Have the fact owner review support and the spokesperson or authorized approver review the attributed wording, which deserves the same care as an executive quote with a clear job in the story. Keep any machine suggestion separate from those human decisions. Check whether the response is a named quote, a company statement or material provided under explicitly agreed terms. Preserve the final version and required release authorization before it leaves the team.

QuoteIt takes the first pass at the support check. A response drafted from the documents you upload has each claim beside the passage it came from, and claims no passage supports are flagged, including absolutes such as "all customers were informed". Wording that gives the spokesperson a feeling, such as regret or concern, is marked as proposed until they approve it. The decision to send stays with the fact owner and the approver. The comparison of AI tools for PR professionals shows which tools check claims at all.

Protect the material used to reach the answer. A private incident log may contain customer details that are unnecessary for the journalist's question. A cleared summary may be shareable even when the source file is restricted. Check permission for the summary itself and retain the underlying evidence where authorized reviewers can inspect it.

The follow-up is part of the job: send the promised update on time, and correct quickly if new evidence contradicts what you said.

Record exactly what was sent, when, to whom and with which follow-up commitment. Put the next action with a named person. If the check remains incomplete at the agreed time, send the promised status update using approved wording. If evidence contradicts the earlier response, alert the communications owner and arrange a correction that clearly identifies the changed fact; a practical correction plan covers who to tell and in what order. Newsrooms hold themselves to a similar rule: the BBC's guidelines say serious factual errors should normally be acknowledged and corrected quickly and clearly³.

An incomplete answer can leave the organization dissatisfied with how much it has explained. That tension is real, especially near a deadline. The practical aim is to answer every part the evidence permits, identify the remaining limits honestly and keep working on the unresolved questions. The follow-up belongs to the assignment even after the first response has been sent.

Once the count and the notification record are confirmed, record them as approved facts with an owner and a date so the next inquiry starts from them. Add the questions that caught the team out to an interview angle bank, so the spokesperson has prepared, supported answers when a similar request arrives.