Cracker Barrel reported third-quarter earnings yesterday that beat a very low bar by a very wide margin. Analysts expected a loss of around 42 cents per share but the company delivered an adjusted profit of 29 cents. Revenue came in at $797 million down about 3 percent from a year ago, but well ahead of estimates. The loyalty program now stands at 12 million members, with more than 40 percent of sales tracked to a member. The stock jumped on the news.

Worth being honest about what this is and isn't.

Guest traffic is still down 6.7 percent so fewer people are walking through the door, and the sales decline was cushioned by people spending more per visit. So this isn't a finished recovery, more like a stabilization. But for a company that lost nearly $100 million in market value in a single week last August after its logo redesign created massive backlash, stabilization is a great story.

The more interesting part is how they got here, and CEO Julie Masino spelled it out on the earnings call: "We put the remodel program on pause this year. Given everything that happened to us, we wanted to really listen to our guests." Guests, she said, didn't want the company to move forward with it... so they didn't.

What I noticed though is that "we're listening" didn't work the first time they said it. Masino was thanking guests for their feedback within weeks of the reversal, way back in September and traffic kept falling by double digits anyway. The words were the same ones every company uses after a backlash, and they did what those words usually do: nothing.

What actually changed the trajectory was that Cracker Barrel spent the next three quarters turning listening into machinery. They built a feedback system they called "Front Porch Feedback" wired into the loyalty program, so guest comments connect to an actual transaction at an actual store (not just some sentiment dashboard) a traceable record of what 12 million regulars experienced and said. Then they let it overrule them. Dishes guests missed went back on the menu. A remodel program that anchored a $700 million strategy got suspended because guests said no. Every cook in the system got retrained on the core recipes. By yesterday's call, "we wanted to really listen to our guests" was a legit description of how the company had operated for nine months and the numbers, imperfect as they still are, were the receipt.

The next time "we hear you" goes into a statement you're drafting, ask the harder question first which is, hear them how? Where does customer feedback physically enter your company, who reads it, and what decision has it ever changed? If you can't explain the system or point to something the company did differently because of it, the line is just a slogan and your audience can tell, because Cracker Barrel said the same words in September and the traffic data ignored them. Listening only became credible when it became operational.

Reuters reported today that companies across China in tech, entertainment, and advertising are quietly cutting jobs as AI takes over more work, deliberately keeping the reductions small and unannounced. Nine workers across those industries described the same pattern of contractors being let go in waves, graduate hiring being scaled back, teams and teams being thinned through attrition. An engineer at Alibaba Group's cloud division said AI-driven headcount reductions have started there too, and will likely roll out gradually rather than in one big round. There will be no restructuring narrative, no CEO memo, no employee FAQ... and that's the point.

Two forces make this different from anything in the Western playbook.

The first is legal. Under Chinese labor law, any company cutting more than 10 percent of its workforce needs government approval, and Chinese courts have already ruled against companies at least three times for firing workers to replace them with AI. So staying quiet and staying small is how you stay under the legal threshold.

The second is political. Beijing is pushing AI adoption hard as a national priority, with targets of 70 percent adoption in key sectors by 2027. The government wants companies to adopt AI fast, just not so fast or so visibly that job losses threaten social stability. So companies are caught holding both ends: adopt the technology but hide the consequences.

The infrastructure for Western-style workforce comms doesn't exist there anyway. LinkedIn shut down its social platform in China in 2021, there's no functioning Glassdoor equivalent. WeChat, Weibo, and Douyin are monitored and censored. Workers coined their own euphemism for being cycled out years ago (毕业 (bìyè), "graduation") because using the word 'layoff' would be flagged. And the pressure now reaches inside the company too because Reuters found some employers ranking employees by how much they use AI tools and tying those rankings to performance reviews and promotions. One engineer said he can't shake the feeling he's training his own replacement. None of this means workers don't know what's happening - the story circulates in private channels whether or not anyone writes it down. Citi estimates roughly 70 million Chinese jobs or about one in ten are at high risk of AI displacement.

What makes this a problem worth your attention isn't that China requires a different communications plan... any competent team localizes. The problem is that silence in one market doesn't stay in that market. The workers being quietly cycled out are talking, to each other, to reporters, to anyone who asks. Today's Reuters story exists because nine of them told it and that story doesn't even get read in China, where Reuters is blocked unless you have a VPN. But it does get read in New York and London, by your investors, your journalists, and your own employees, the same people hearing your CEO talk about AI productivity gains on the next earnings call. Alibaba's contribution to today's coverage was "did not respond to a request for comment," which means the only people telling the story of what's happening inside the company were anonymous employees and a consultant.

That's the real test of a global AI narrative; whether what you're saying in your loudest market survives being read next to what you're doing in your quietest one. Your US employees will read the Reuters story about hush-hush cuts in the China office and ask the obvious question which is, is that happening here, under a different name? Your CEO's productivity story will get fact-checked against your headcount reality, and the people doing the checking don't care which market the facts came from.

If you've been waiting for the data to call a turn in the economy, this wasn't the week, and there probably won't be one anytime soon.

The National Federation of Independent Business (NFIB) reported yesterday that small business optimism fell to 95.3 in May, below its 52-year average for the second straight month. The uncertainty index sits at 91, against a historical average of 68, and it's been elevated like that for months, not weeks. Plans to invest in equipment and facilities hit their lowest level since March 2009. The share of owners planning to raise prices is the highest since July 2022, driven, NFIB's economists say plainly, by the Iran war and fuel costs that small businesses can't pass along. And the wider backdrop is reinforcing the mood rather than relieving it: US inflation data lands today amid open speculation that the Federal Reserve's next move could be a rate increase, the European Central Bank is expected to raise rates Thursday, and the Bank of Japan is expected to follow next week.

So, in plain terms, the cost of money is more likely to go up than down across three continents because the war keeps pushing energy prices up and nobody can say when that ends. NFIB Chief Economist Bill Dunkelberg captured the whole condition in one line: "Despite the enthusiasm around AI, the overall picture is divided."

Divided, and durably so. The AI economy is generating real excitement at the same time that the businesses your company sells to, buys from, and employs people alongside are absorbing a cost shock with no visible end date. You can hear both in a single news cycle: Cracker Barrel's earnings call yesterday flagged fuel inflation and softness among lower-income customers as the conditions they're operating in, not a storm they're waiting out, just the weather now.

Here's where this becomes a comms problem.

Messaging written inside a long uncertain stretch tends to smuggle in a promise: as conditions normalize, once the environment stabilizes, positioned for the recovery. Each of those phrases bets that the fog lifts soon but your audiences have probably been hearing versions of that bet for months (from your company and everyone else's) while their own numbers keep saying otherwise. The longer the gap runs, the more those phrases stop reading as optimism and start reading as a company that isn't describing the world its customers actually live in.

Search your current messaging for the subtle promises - every "as conditions improve," every "when the environment normalizes." Replace them with what you can stand behind in month nine of this, or month eighteen - what the company is doing now, for the people feeling the squeeze now. Audiences forgive uncertainty but they don't forgive being told the corner is near by people who can see the same data they can.