Most leaders meet "media handling" for the first time on the worst day of their year, a journalist's email landing at 4.55pm, a phone buzzing with a story already half-written. By then the instinct is to clam up, run it past legal, and say as little as possible. That instinct is usually wrong. The research on reputation is surprisingly consistent: the organisations that fare best are the ones that get ahead of the story rather than chasing it, and that treat the public as a party to talk with, not a crowd to talk at.

The quick version

  • Tell it yourself, first. Disclosing your own bad news ("stealing thunder") raises how credible people find you and softens how severe the problem looks, a finding replicated across studies since 2005.
  • Match the response to the blame. Coombs' crisis model sorts events into victim, accidental and preventable, and the more you're seen to be at fault, the more your response has to shift from explaining to apologising and fixing.
  • Treat PR as a two-way conversation, not a megaphone. Grunig's research links the most effective communication to listening and adjusting, not just broadcasting.
  • In an interview, bridge, don't dodge. Acknowledge the question, then steer to your two or three key messages in plain language.

The idea in depth

PR has a credibility problem inside leadership teams: it sounds like spin, and spin is the opposite of trust. But the academic field has spent forty years arguing the reverse, that the durable version of the discipline is about relationships and honesty under scrutiny, not manipulation. Three strands of that research are worth a leader's time.

1. Tell your own bad news first, the "stealing thunder" effect

The single most counter-intuitive finding in crisis communication is also one of the best-supported. In a 2005 experiment, Laura Arpan and David Roskos-Ewoldsen had participants read about an organisational crisis disclosed either by the organisation itself or by an outside party. When the organisation broke the news about its own problem first, a move the authors called stealing thunder, people rated it significantly more credible and judged the crisis as less severe than when a third party broke it (Arpan & Roskos-Ewoldsen, Public Relations Review, 2005). Later studies have repeatedly reproduced the effect, and it has since been folded into mainstream crisis-communication practice.

Why would confessing help? The likeliest mechanism is that volunteering bad news signals you have nothing to hide, which buys you credibility, and a more credible source makes the problem itself look more contained. So the move is: when something has gone wrong and will surface, be the one who surfaces it. Draft the holding statement before the journalist calls, name the problem plainly, and say what you're doing about it, in that order.

An honest limitation: stealing thunder is not a licence to over-confess, and the lab evidence is exactly that, controlled experiments, often with student samples. It works best for problems that are real, bounded and already likely to come out. It is not a strategy for blurting out every internal wobble, nor a substitute for actually fixing the thing.

2. Match the response to the blame, situational crisis communication

Not every bad day is the same kind of bad day, and the response that calms one can inflame another. W. Timothy Coombs' Situational Crisis Communication Theory (SCCT), set out in 2007 and developed across editions of his book Ongoing Crisis Communication, makes the point precisely: crisis managers should match the response to the level of responsibility the public will attribute to the organisation (SCCT overview; Coombs, Corporate Reputation Review, 2007).

Coombs sorts crises into three clusters by how much blame attaches to you. In the victim cluster (natural disaster, rumour, sabotage, product tampering) the organisation is itself a casualty, and blame is weak. In the accidental cluster (a technical failure, an equipment fault) the harm was unintended, and blame is moderate. In the preventable cluster (cutting a known corner, misconduct, management deceit) people see the harm as avoidable, and blame is severe. The response strategies form a ladder: deny (it wasn't us), diminish (it wasn't that bad, or out of our control), and rebuild (apology, compensation, repair), with bolstering reminders of past good behaviour layered on top.

flowchart TD
  A(["Something has gone wrong"]) --> B{"Who will the public
blame?"}
  B -->|"Victim, not your fault"| C(["Lead with sympathy + facts;
light defensive response"])
  B -->|"Accidental, unintended"| D(["Explain + diminish;
show controls, take some ownership"])
  B -->|"Preventable, avoidable"| E(["Apologise + rebuild;
compensate, fix, prove change"])
  C --> F(["Reinforce with a true reminder
of your track record"])
  D --> F
  E --> F
Coombs' SCCT, simplified: the more blame the public attaches, the further your response moves from explaining toward apologising and repairing. Leaders Loop

So the move is: before you write a single line, answer one question, who will the public hold responsible? If the answer is "us, and we could have prevented it," denial and excuse-making will read as evasion and make things worse; lead with the apology and the fix. If the answer is "no one, we were hit too," you can lead with sympathy and facts and keep the defensive register light.

An honest limitation: SCCT tells you which category of response fits, not the exact words, and blame attribution is slippery, the public can reclassify a "victim" crisis as "preventable" the moment a missed warning sign surfaces. Treat the clusters as a starting read, not a verdict, and revisit them as facts change.

3. Make it two-way, Grunig's excellence research

The deepest shift the research asks for is in posture. James Grunig's Excellence Theory describes four historical models of PR, press agentry (publicity at any cost), public information (one-way but truthful), two-way asymmetrical (research used to persuade), and two-way symmetrical (research used to reach genuine mutual understanding), and argues the symmetrical model is the one associated with effective, ethical communication (Excellence Theory overview). The headline isn't "talk more." It's listen, then adjust what you do, because the channel that carries your message also carries the early-warning signal that you have a problem at all.

So the move is: build a relationship with the two or three journalists and stakeholder groups who cover you before you need anything from them. Answer their routine questions promptly, off the back of no crisis. When the hard day comes, you are a known quantity returning a call, not a stranger issuing a statement.

And when you do face the questions, the practical skill is bridging, acknowledging what was asked, then steering back to the two or three things you came to say, without dodging. Media trainers teach it as a simple A-B-C loop:

flowchart LR
  Q(["Hostile or off-topic
question"]) --> A(["Acknowledge
'That's a fair concern…'"])
  A --> B(["Bridge
'…and what matters here is…'"])
  B --> C(["Communicate
your key message"])
  C -.->|"next question"| Q
The bridging loop used in media training: never leave a question unacknowledged, but always return to your key messages. Leaders Loop

An honest limitation: the "symmetrical" ideal is contested, critics note that organisations and their publics rarely hold equal power, so true symmetry is often aspirational. Read it as a direction of travel (toward listening and adjustment) rather than a claim that every conversation is a level playing field.

A worked example

Picture a mid-sized logistics firm, call it Northbridge. A software update mislabels a batch of refrigerated shipments, and for eleven hours the temperature logs are unreliable. No one is harmed, but a regional newspaper has a tip and a reporter wants comment by morning. (Illustrative scenario; figures below are invented to show the method.)

The reflex is silence: "no comment until we know more." Run it through the toolkit instead. Blame cluster? Accidental sliding toward preventable, the bug was theirs, and a prior release had carried an ignored warning. That rules out flat denial. Stealing thunder? Yes: Northbridge's head of operations issues a short statement that evening, before the paper runs, naming the fault, the eleven-hour window, the ~1,400 shipments checked, and the recall of anything that can't be verified. Two-way? The reporter, who has dealt with this operations lead before on dull supply-chain stories, gets a direct line and a straight answer.

The published story still isn't flattering. But the framing is "company spots fault, pulls stock, briefs customers" rather than "newspaper uncovers temperature scandal firm tried to hide." Same facts; a different reputation. That gap, between the story you tell and the story told about you, is the whole game.

The reporter isn't your enemy or your friend. They are a channel, and the only question that matters is whether your version of the facts is the one that travels.

Frequently asked questions

Isn't "stealing thunder" just admitting fault and inviting a lawsuit?

It's disclosing what is true and will surface, not assigning legal liability. The two can be separated, and they should be: you can state the facts of what happened and what you're doing about it without conceding fault in the legal sense. Where liability is genuinely live, this is a conversation to have with a qualified lawyer in your jurisdiction, not instead of one. The research says the reputational cost of being exposed is higher than the cost of disclosing; it doesn't waive your legal advice.

What do I do when I genuinely can't comment yet?

Say that, specifically. "No comment" reads as guilt; "We became aware of this two hours ago, we're verifying the facts, and I'll have a substantive update by 9am" reads as control. Give the reporter a real next step and a real time, and then hit it. Silence cedes the narrative; a credible holding line keeps your hand in.

How is this different from spin?

Spin distorts the facts to flatter you; handling distorts nothing and instead controls timing, framing and channel. The line is honesty. Choosing to announce your own recall is handling. Claiming the recall was "a proactive quality initiative" when it was a defect is spin, and the modern media environment punishes it quickly, because someone always knows the real story.

I'm a manager, not a CEO. Why does this apply to me?

"Media" now includes a viral internal Slack message, a Glassdoor review, a customer's social post. The same logic, surface it yourself, match your response to how blameworthy you'll seem, talk with people rather than at them, governs how you handle a team-level reputation hit long before any newspaper is involved. The skill scales down as well as up; see managing up, down & across for the everyday version.

Should I just hire a PR agency and stop worrying about it?

An agency drafts and advises; it does not face the camera or the board for you. The credibility a stealing-thunder statement buys comes from a named, accountable human standing behind it, usually you. Outsource the craft if you like; you can't outsource the presence.

Related in the Toolkit

Media handling sits inside the wider work of reading and managing the people around a decision. A few neighbours worth pairing it with:

Where to go next