.
The Daily Trend Report. Every weekday morning, giovanni gallucci and a rotating cohost (Chloe Dawn on clean label food and beverage, plus Rachel Donovan or Alexis Parker on social media) synthesize the day's cross-LLM trend reports into three operator stories: social media, clean label food and beverage, and content, social, and AI. Tight, specific, no filler. For owners, founders, marketing leads, and operators in clean-label CPG.
.
.
.
| ai-assisted content
[INTRO]
[HOST] Good morning. This is giovanni gallucci, this is the Weekend Trend Report, the Saturday episode of the Daily Trend Report feed, and Rachel Donovan is with me.
[GUEST:happy] Morning, Gio.
[HOST] Four trends this week. People can now tell Facebook, in plain words, what they want less of, and a lot of brand posts are about to land in that pile. The AI answer engines are building their shopping advice out of creator videos and Reddit threads, and almost none of it comes from brands. Protein walked out of the gym and into the coffee cup, the meat stick and the soda can. And the cookie aisle is selling the past, with five SKUs of Nilla Wafers beating a hundred and twenty flavors of Oreo for attention. Let's go.
[STINGER]
[SECTION 1: The Audience Is Writing The Algorithm Now]
[GUEST] So, first one. Eddie had this, and it's been building for a few months. On October thirteenth Facebook rolled out something called Your Algorithm.
[HOST] Tell people where it lives, because where it lives is the whole story.
[GUEST] Three-dot menu on any post. You tap it and there's Your Algorithm, plus Interested and Not interested. And then it can ask you a follow-up question, and you just type what you want more of or less of. In your own words. Meta says the feed updates in real time.
[HOST:thoughtful] And Instagram already had it. Meta tested it on Reels last December and shipped it on Instagram in June. Threads has its own version called Dear Algo. So this is three Meta apps in a row handing the steering wheel to the person holding the phone.
[GUEST] And Meta says it doesn't replace the ranking. It sits on top of it.
[HOST] Right, it's a layer. A language model reads what you typed and weighs it next to your likes, your comments and your watch time. Nobody outside Meta has measured how hard it pulls. But think about what a person types into that box. Nobody types "more brand content, please."
[HOST] What they type is "more cooking videos, less politics, fewer ads." Or "more Texas barbecue." Which means a brand that sits inside a topic people ask for gets pulled in by name, and a brand that posts a little of everything gets sorted out by the very people who used to tolerate it.
[GUEST] Okay, but there's a second piece. Eddie also had Threads.
[HOST] Give me that one.
[GUEST] Connor Hayes, who runs Threads now, said they made a recent change so people see more posts from accounts they actually follow. Mixed with some stuff you haven't found yet, but tilted back toward the follow.
[HOST] And LinkedIn is testing a Network tab that only shows your connections and the people you follow. Put those three next to each other and you get the trend. For two years the platforms said the follow was dead and the recommendation engine was everything. Now they're adding a manual override and turning the follow back up.
[GUEST:surprised] So the follower count is worth something again?
[HOST] It's worth more than it was in the spring, and the reason is the same on all three apps. People got tired of feeds that guessed wrong.
[GUEST] So what does a small food brand do with a dial it doesn't control?
[HOST] You make yourself easy to ask for. When somebody types "more of this," the model has to know what "this" is and decide whether you're part of it. A hot sauce account that posts tacos on Monday, a meme about Mondays on Tuesday, a hiring post on Wednesday and a giveaway on Thursday is four different topics. The model can't file it, and the person who said "less memes" just filed it for them.
[HOST] And on Threads, the follow you earn this fall pays you back more than it did in the spring. So the job changes a little. You still want reach from strangers, but you want the follow, and you want the follower to actually like what shows up. Three things.
[HOST] One. Pick two topics your customer would type into that box, in their words, and post nothing outside them for thirty days. Breakfast and camping. Game day and tailgates. Write them on a sticky note on the monitor.
[HOST] Two. Say the topic out loud in the first line of every caption and the first three seconds of every video. "Breakfast taco, ten minutes, one pan." The model reads it, and so does the person who scrolls past with the sound off.
[HOST] Three. Ask for the follow at the end of your best posts, and give a reason that matches the topic. "Follow for one new breakfast every Friday." Then show up every Friday, because the follower who gets what you promised is the one who never types "less of this."
[STINGER]
[SECTION 2: The Answer Engine Reads Your Comment Section]
[GUEST] Second social trend, and this one has a fake word in it. Travis brought it. Researchers at Fudan University put a paper on arXiv on October eighth. They made up a concept, the Cactus Gravity Protocol. Total nonsense. They posted about it on sites like Medium and YouTube and watched what the AI search tools did.
[HOST] And within a week?
[GUEST] Eight of the ten tools they tested were citing it. Grok cited it in thirty of thirty-two answers. Google's AI in fifteen of thirty-three. Perplexity, thirteen of thirty-five. ChatGPT, seven of twenty.
[HOST:appalled] Thirty out of thirty-two. For a thing that doesn't exist.
[GUEST] They also put forty posts on a self-hosted WordPress site. Zero citations in the same week.
[HOST] Your own blog, by itself, barely registers. The places those tools read from are the big shared platforms. In an earlier test from the same team, Reddit was the most cited source for Google's AI and for Grok, YouTube for Perplexity, and Wikipedia for ChatGPT. And Eddie had the companion study, which is the one that should keep a brand manager up at night.
[GUEST] Sundogs. They looked at thirty-nine hundred videos in packaged goods and autos, four hundred of them cited by AI engines. About sixty percent of the cited ones came from creators. About thirty percent from editorial. Ten percent from brands.
[HOST] Ten percent. The brands spent the money and the creators got the citation. And look at what got cited. Hands-on demos and reviews, comparisons, specifics about ingredients or how a thing works. A lot of them ran eight minutes or longer, and Shorts got cited far less. View count didn't predict it. Production quality didn't predict it.
[GUEST:curious] Wait. So the long, kind of boring video wins?
[HOST:excited] The useful video wins. A woman in her kitchen spending nine minutes comparing three jarred salsas, reading the labels out loud, beats the brand's thirty-second hero spot with the slow-motion chip dip. The machine wants an answer, so it goes where the answers are.
[HOST] One caveat. Sundogs said food was more mixed, and it's two categories from one vendor, so treat the percentages as a direction. The direction lines up with the third piece, which Kendall brought.
[GUEST] Feedvisor surveyed shoppers. Sixteen percent start looking for products in ChatGPT, nine percent buy there, and only five percent buy an AI pick on the spot. And PartnerCentric asked what people do after the AI suggests something. Thirty-three percent go check reviews. Twenty-two percent compare prices.
[HOST] So the machine starts the trip and people finish it by reading other people. Reviews, comments, a creator who actually used it. Both ends of the trip run on what customers say about you, and almost none of it runs on what you say about yourself.
[GUEST:annoyed] Okay, but if a fake protocol can get cited in a week, can't a brand just fake it? Plant a bunch of Reddit posts?
[HOST:sarcastic] You can, the same way you can put a sign on the highway that says "best brisket in Texas." It works right up until somebody checks. Google already said its spam rules cover trying to manipulate AI answers, and Reddit users can smell a brand account wearing a fake mustache from across the state. When it comes out, and it does come out, the thread about you becomes the thread about you lying. That's the citation you get for a year.
[HOST] So show up where people already talk, as yourself, with a name on the account, and with the specific stuff a model can quote. Here's the three.
[HOST] One. Make one long video a month, eight to twelve minutes, where the founder answers the question your customer types before they buy. "What's actually in this and why does it cost more." Label on screen, ingredients read out loud, posted to YouTube with the question as the title.
[HOST] Two. Find the three Reddit threads and the five creator reviews where people already ask about your category, and show up in them as yourself. Founder account, real name, answer the question, link nothing unless somebody asks. Do that every week.
[HOST] Three. Turn reviews into content on purpose. Ask every new customer for one sentence about how they use you, then put the best ones on screen in your own videos with their permission, because that's the stuff people go read after the AI hands them your name.
[STINGER]
[SECTION 3: Protein Left The Gym]
[GUEST] Okay, to the shelf. Kendall, Travis and Eddie all had protein this week, and it's everywhere. Kendall had General Mills taking Epic into meat sticks.
[HOST] Epic, the Austin meat bar brand General Mills bought about a decade ago. What did they launch?
[GUEST] Four sticks. Original beef, Wagyu beef, rotisserie chicken and hot honey chicken. Grass-fed beef, free-range chicken, up to ten grams of protein. On Amazon and at Sprouts now, nationwide shelves in 2027. And Food Dive lined it up against Chomps, Archer, Jack Link's and PepsiCo's Good Warrior.
[HOST] And the number PepsiCo keeps quoting is that eighty-six percent of Americans say they're adding protein, and more than half put it first when they snack.
[GUEST] And it's not just sticks. Travis had a Beverage Industry piece from the twenty-second about protein moving past sports nutrition.
[HOST:curious] Read me the examples, because the examples are the trend.
[GUEST] Protein cold foam on coffee. Bottled protein coffee. Sparkling protein drinks, clear protein water. There's a GHOST Yoo-hoo protein can. And a couple of the people quoted said GLP-1 users are a big part of who's buying. And Travis had the other side of it. Whey.
[HOST] Whey protein concentrate more than doubled in price in a year, according to Expana's numbers. Everybody wants protein at the same time, and there is only so much whey coming out of cheese plants.
[GUEST:thoughtful] So what is this, actually? Because "everybody wants protein" sounds like a fad.
[HOST] A fad burns out. This one has a reason under it. For twenty years protein was a gym word. It meant a tub with a lion on it and a guy named Brad. Now it's a grocery word. It's the thing a mom checks on the yogurt and the thing a sixty-year-old on a GLP-1 checks on the coffee. When a claim moves from a niche to the whole store, it stops being special. Every bag has a number on it now.
[GUEST] Which is bad news for a small brand that built everything on "twelve grams."
[HOST:angry] Brutal news. If your whole personality is a gram count, General Mills can print a bigger gram count by Tuesday, with a trade budget that buys the end cap. That's the Big Food play every single time. Founders spend five years teaching people that meat sticks don't have to be gas station garbage, and then the conglomerates show up with a line extension and the same claim in a bigger font.
[GUEST] So where does the small one win?
[HOST] Everywhere the number can't go. Where the protein comes from. Whose cows. Which county. How many ingredients. Who's holding the stick. A rancher in Kerrville with forty head and a dehydrator has something Epic will never have, which is a face and a pasture you can film at six in the morning. And nobody saves a post that says "twelve grams." They save the post that tells them when to eat it. That's the other half of protein leaving the gym. It moved into normal days. The morning coffee, the school pickup, the drive between job sites, the afternoon slump at a desk. Those are occasions, and occasions are what people search, what they type into that Your Algorithm box from the first segment, and what creators film without being asked.
[GUEST:excited] So you stop filming the workout and you film the commute.
[HOST] You film the commute. The lifter already knows about you. The new buyer is somebody who would never walk into a supplement store and doesn't think of themselves as a protein person. They think of themselves as hungry at three o'clock. Three things.
[HOST] One. Retire the gym from your feed for a month. Every protein post goes into a normal day: the cooler on the jobsite, the cup holder at pickup, the desk drawer at three p.m. Same product, a life your new buyer actually lives.
[HOST] Two. Put the source on camera instead of the gram count. Film the ranch, the dairy, the farmer, the ingredient list read out loud in under ten seconds. Let the bigger brand print the bigger number while you show the cows.
[HOST] Three. Seed creators who don't do fitness. A mom with a carpool account, a long-haul trucker, a nurse on night shift. Send them product, ask them to show where it fits in their day, and repost the ones that feel like a Tuesday.
[STINGER]
[SECTION 4: The Pantry Went Back In Time]
[GUEST:happy] Last one. And it's the cookie thing.
[HOST] The cookie thing. Eddie had it out of Food Dive.
[GUEST] Mondelez has a group it calls its Americana brands. Nutter Butter, Nilla Wafers, Honey Maid, Teddy Grahams, Mallomars, Barnum's Animal Crackers. More than a billion dollars in sales a year, and after two years of declines, their brand manager, Caroline Suppiger, called this a breakout year.
[HOST] Give me the numbers.
[GUEST] Specialty cookies up four percent through September twelfth. The cookie category overall, flat. And then the SKU counts. Nilla has about five. Honey Maid has nine. Nutter Butter has ten. Oreo has about a hundred and twenty in the U.S. alone.
[HOST:surprised] Five. Nilla grew on five SKUs while Oreo needed a hundred and twenty flavors to stand still.
[GUEST] And people are using them as ingredients. Graham crackers for baking, Nilla for desserts and dips.
[GUEST] Kendall and Travis had the drink side, too. DD Beverage, out of Vancouver and Fort Worth.
[HOST] The Bear 'n Beaver people.
[GUEST] Bear 'n Beaver craft soda and GRIZZLY Energy. Their CEO, Andrew Drayson, said root beer and the cream flavors are their best sellers on the soda side. So they put root beer and cream soda into the energy drink. And he said zero sugar is becoming an expectation.
[HOST] Which is the smart version of nostalgia. The flavor people knew as kids, in a format where they've never seen it, with the sugar people don't want anymore taken out.
[GUEST:mischievous] And then Pepsi showed up.
[HOST] Of course Pepsi showed up.
[GUEST] Pepsi Vanilla and Cream, plus a Zero Sugar version, nationwide on October nineteenth. Limited time, holiday release. And Sporked says fans are already calling it a knockoff of Coca-Cola's Creamy Holiday Vanilla from last year.
[HOST:excited] So the biggest soda company on the planet ran the same play as a craft soda company from Fort Worth and still got accused of copying. That's how strong the pull is. When the giants start borrowing flavors your grandmother would recognize, the people who made those flavors the whole time get a free education campaign.
[GUEST:thoughtful] But how does a small brand make nostalgia without it feeling like a costume? Like the fake old-timey font on a brand that started in 2023.
[HOST] That's the trap, and I see it every week. Sepia filter, a typewriter font, "since forever" on a label from last spring. People can tell. Nostalgia only works on social when it's documented. The real recipe card with the grease stain on it. Your grandfather's handwriting. The photo of the first delivery truck. The year on the sign. If you don't have a history, borrow the occasion instead of faking a past: root beer floats on a hot porch, graham cracker crust at Thanksgiving.
[GUEST:happy] I've got a photo of my grandpa's gas station in Missouri that would get more likes than anything I've ever posted.
[HOST] Because it's true and it's specific. That's the whole trick, and Big Food can't buy it. Mondelez can sell the memory of Nilla Wafers. It can't sell the memory of your mom's banana pudding at the church potluck in 1987, and that's the post that gets shared, because people tag their own mom in it.
[GUEST] So the move is the old stuff, but shown as it really was.
[HOST] Shown as it really was, and then put to work in a kitchen today. Three things.
[HOST] One. Go find the oldest true thing your business has. A recipe card, a first label, a photo, an invoice. Post it with the year and one sentence about who made it. If you're two years old, post the first batch.
[HOST] Two. Run your product as an ingredient in a dish people already have a memory about. Banana pudding, a float, a graham cracker crust, a casserole somebody's aunt makes. Film it start to finish with your product on screen, and let the comments fill up with people's own versions.
[HOST] Three. Ask your audience for their memory, and build the next month on the answers. "What did your grandmother always have in the pantry?" Then make the top three answers with your product in them, and tag the people who named them.
[STINGER]
[OUTRO]
[HOST] That's the Weekend Trend Report. If you run a food or beverage brand and you know your feed needs more and better content, but there's nobody to shoot it, edit it and post it every week, that's the problem I fix. I produce content for food and beverage brands. Short-form video, photography, carousels, the founder on camera, the ranch at six in the morning, all of it made for the way your customers scroll. Fifteen-plus years running organic social media for food and beverage brands. gallucci dot net, or DM at gallucciNET on Instagram or LinkedIn.
[GUEST] And tomorrow?
[HOST] Tomorrow in this same feed is The Week Ahead, what's coming this week and what to do before it gets here. Thanks, Rachel.
[GUEST] Bye, Gio. I'm going to go find that gas station photo.
[HOST] Now go make something worth following.