Who Corrects It

Why waiting until you can fix a client's problem can leave you with more than the problem to explain.

A spokesperson in round glasses and a cardigan stands at a lectern with one flat palm raised, beside the words no cyanide. An examiner in a tie at a desk holds out an open hand beside the blue words was wrong, and a blue lab door stands open on a small bottle.

We say the client should hear bad news from us first. Then we wait until we can explain the cause. Or until the fix is ready. Or until we are sure the problem is worth mentioning.

The client still hears it from us. But first has come to mean first with a complete answer.

I understand the appeal. A problem and its solution belong together. You want to arrive with something useful, not hand over a worry. You don't want to alarm the client unnecessarily, and you're already working on a fix.

You haven't decided to hide anything. You expect to have solved the problem by the time you explain it.

From your side, waiting feels like taking responsibility: the work is moving, and the message can wait. You can put off the explanation without feeling you've withheld information.

But the fix that should take a day takes a week. The client notices something. They ask a question you weren't ready to answer.

Now you have more to explain than the original problem. You also have to explain why you didn't mention it.

The time you spent trying to spare them concern becomes time they spent without information you already held. Your intention hasn't changed. What they can see has.

Your intention hasn't changed, but what they can see has.

Your intention hasn't changed, but what they can see has. Dots, in order: the information you hold, your intention, the gap, what they can see.

Click a numbered dot to see what it stands for. Start with 1.

Before anyone knew the scope

The Tylenol recall seems to offer a cleaner model: danger discovered, 31 million bottles pulled, trust preserved through one decisive act.

The record is less tidy.

Between September 29 and October 1, 1982, seven people in the Chicago area died after taking cyanide-laced Extra-Strength Tylenol capsules. Johnson & Johnson had much at stake. Fortune estimated that Tylenol supplied 15 to 20 percent of its 1981 profit on about 7 percent of sales.

On September 30, the company recalled lot MC2880, covering 93,400 bottles. On October 1, it added a second lot, bringing the total above 264,000 bottles. It also recalled every Extra-Strength capsule in the Chicago area.

That day, the FDA warned consumers nationwide against using Extra-Strength capsules. J&J set up a consumer hotline and offered a $100,000 reward. Television ads came down on October 2.

These actions preceded the national withdrawal. They also preceded a settled account of what had happened.

McNeil, which made Tylenol for J&J, first said it used no cyanide on its premises. When the Cook County medical examiner asked on Friday, McNeil said its quality-control lab used it.

The national capsule withdrawal came on Tuesday, October 5, six days after the first death. Strychnine-laced capsules had turned up in Oroville, California. The FDA then agreed with J&J on the national action. Tablets stayed on sale.

The period record counted eight cyanide-tainted bottles, all in the Chicago area. That count was not available to settle the earlier decisions.

The company acted, learned more and changed its response. Its first account of itself was wrong, and a question from outside fixed it.

Speaking early has a cost

Wasn't that the risk of speaking before you knew?

Yes. Partial information can cause needless alarm. A wider recall could frighten people without making them safer.

According to James Burke's account in Fortune, both the FBI and FDA counseled against a precipitous recall on Monday, October 4. He said the FBI feared encouraging the poisoner. The FDA feared creating more anxiety than the action would relieve.

Those are Burke's accounts of their advice, not statements from the agencies themselves.

The cost was real. The recall cost about $100 million before tax, roughly $50 million after tax. Tylenol's market share fell from 35 to 37 percent to about 7 percent.

A year later, it stood near 29 to 30 percent, more than four-fifths of its old share. That recovery came with help from 80 million $2.50 coupons and a thinner profit margin. Warning people did not spare the business from damage.

The October 1 warning went out before anyone knew the scope. That does not prove speaking first caused the recovery.

The research offers narrower support. In 1993, Williams, Bourgeois and Croyle ran two mock-trial studies with 257 and 148 students. Damaging evidence did less harm when the side it hurt disclosed it first.

Fennis and Stroebe's 2014 experiments found a benefit for companies with poor starting reputations, but none for those with good ones. This was lab work with students, not field proof that early disclosure always builds trust.

I still tell clients before they find the problem themselves. I can't promise that doing so will keep the account. I can make sure they don't have to discover what I already know.

Within 24 hours

My rule is to communicate within 24 hours. The repair may take longer. The client should not have to wait for it to hear that something went wrong.

That means sending a message while some answers are missing. State what happened and what it means so far. Say what you are doing and when they will hear next.

As a template, rather than an account of an event: "We found [problem]. The confirmed effect is [impact]; we're still checking [unknown]. We're taking [action]. I'll update you at [time], whether or not the fix is finished."

The unknown belongs in the message. Leaving it out would make the update sound more complete than your knowledge allows. Giving a next update time lets the client know when the missing answers will be checked again.

A handoff is not made while the client lacks information you hold and they need. Working hard on the fix does not complete it.

A handoff is not made while the client lacks information you hold and they need.

A handoff is not made while the client lacks information you hold and they need. Dots, in order: the fix, the information you hold, the client.

Click a numbered dot to see what it stands for. Start with 1.

The same rule applies before a scheduled meeting or report. The client should already know. A meeting gives you time to decide what to do together, rather than making them absorb a surprise before they can think.

The update gives them something they can rely on while the work remains uncertain. The fix still has to make the work perform. Neither makes the other unnecessary.

First used to mean having the answer ready before the client asked.

Now it means telling them before the fix, while the cause is still unclear, before I am sure. It also means returning when something I told them no longer holds.

McNeil's first answer about cyanide on its premises was wrong, and the medical examiner's question corrected it a day later.

A first message can be wrong; someone will correct it, and the only choice is who.

I send a first message with what I know, then an update with what I've learned.