Neutrogena waited four days after Hayden Panettiere's death to speak out, and what it said has left them with more to answer for than the silence did.

The statement went up Thursday afternoon as three Instagram slides (comments turned off which tells us something already) and by evening every headline was filed under "backlash"... Today, ABC, The Hollywood Reporter, all of them. After days of declining comment, the narrative frame has hardened into 'Neutrogena punished a mom for talking about postpartum depression'. This is a skincare brand whose customers are women so punishing one for being honest about postpartum depression feels like a betrayal. The people typing "boycott" under its posts aren't activists, they are the customer. So the only role of a statement at that point is to try to change that story. But this one just confirmed it.

The frustrating part is that the bones were right; the outline is close to the one we ran here Wednesday: condolences, credit for her courage, accountability, action. But that's not what this is.

Let's start with the admission:

"We are proud to have partnered with her and understand that we made her feel unsupported during a very difficult time." That's two evasions in one sentence. "Understand that" means we've been told this information so it's at arms length.... "feel" means it was her perception. It's the corporate equivalent of "sorry you feel that way." Put them together and the brand has admitted how Hayden felt and not really taken accountability for the thing it did.

The next sentence, "This is not what Neutrogena stands for or who we want to be" is a crisis comms staple, and a poorly used one at that. It's there to put distance between the organization and the behavior without ever actually addressing what caused it.

Then the action it attempts to take: "a significant investment to a long-standing community health partner," with more details forthcoming "when plans are in place." No charity, no number, no date.

The third slide is the final blow of this whole thing: "Out of respect for Hayden's family and loved ones, we did not want to speak out so soon following her passing." Set aside for a moment that respect for a grieving family covers Sunday to Monday. The silence people are mad about started in May, when she was alive and telling this story everywhere she could. The brand wasn't actually quiet this week; it declined comment to several media outlets. So if respect was the reason, they should have said that when declining further comment on Monday ("we've reached out to the family and we'll have more to say once we've spoken with them"). Instead the explanation showed up only after the delay became the headline. That's using a grieving family as cover.

The obvious lesson is "respond faster," but it's only part right. Nobody outside Neutrogena knows what the last four days have been spent doing but we do know what they didn't produce: a single sentence that says clearly what the company did, what changes, or what it costs them.

Before your next statement goes out, read it one sentence at a time and push on each one: who is this sentence for? Some sentences are for the people you hurt. Some are for the company i.e. protecting its timing, its values, its pride. Cut every single sentence that's for the company. Run Neutrogena's statement through that test and the third slide is gone, "who we want to be" is gone, and "feel unsupported" is gone. What's left has to say three things: what you did, what changes, and what it costs you.

Goldman Sachs ran the tape on Q2 earnings call scripts and found that 65 percent of S&P 500 management teams mentioned AI, down from 70 percent two quarters ago, after three years of more or less straight up. The AI paragraph has become the new safe-harbor statement: everybody reads one at the top of the call, it sounds the same everywhere, and nobody really listens.

The numbers I found interesting in the research are the small ones. Only 11 percent of companies put a real number on what AI did for a specific job (coding and customer support are Goldman's examples) and just two percent put a figure on what it did for earnings, roughly the same as Q1. So of the companies telling investors about AI, about one in thirty is actually showing them what it earned.

And the market isn't paying for the talk either. Goldman's own basket of twenty stocks picked to benefit from AI productivity has trailed the equal-weight S&P since late 2023, 17 percent to 23. Meanwhile the spending keeps climbing, Ramp has median AI spend per employee at $12 a month in July, up from $5 in January. So spend more than doubled but the share that can explain the ROI stayed at 2.

That leaves comms teams writing a strange kind of script. The AI paragraph is close to mandatory (only 35 percent of the index skips it) but it differentiates nothing, moves nothing, and every quarter it ages a little worse because the analysts listening have read this same research. The scarce asset on an earnings call right now is a number included in that AI paragraph. And the number is scarce because the number the Street wants is a labor number. Two-thirds of companies are paying for AI by moving money out of existing budgets, mostly software and labor. So when your CFO says "AI saved us $200 million" your employees, and investors, will also ask where the $200 million came from.

So before the Q3 scripts lock, two questions for your CFO. First: which sentence in the AI section would you put a number? If there is one, build the section around it and cut the adjectives. Second: what does that number tell the people who work here? A number about output, things like cycle time, customers served, revenue per rep travels far better than a number about headcount avoided, and if the only honest number is a labor number, the internal messaging has to be ready to go the same day as the earnings script. Because they are watching.