Why this matters
By the time a food trend on TikTok is obvious to you, it has usually been obvious to your shoppers for a month. That gap is the entire commercial question. A category manager who reads the trend when it peaks online orders in time to sell it. One who reads it off their own sales data orders into the back half of the curve, and some of that inventory is still on the shelf when the trend turns.
Food trends on TikTok are not a marketing curiosity for a retailer or a brand. They are a demand signal that arrives before the demand does, and the useful question is not "what is trending" but "what does this do to my orders, and when."
This page covers what viral trends look like once they reach scan data, why the lag exists, and how to convert a trend into a specific order rather than a vague sense that you should stock more of something.
What a viral trend looks like in scan data
Two shapes, and they need different responses.
The sustained shift: cottage cheese
Cottage cheese is the clearest recent case of a social media trend becoming a category fact. According to Circana, U.S. cottage cheese sales rose 20 percent in the 52 weeks to mid-June 2025, following roughly 17 percent growth in each of 2023 and 2024. John Crawford, Circana's SVP of client insights for dairy, made the point that matters for planning:
It is not a fad when you are seeing double-digit growth in both dollars and in volume, quarter over quarter over quarter, for two years.
Note the unit of measurement. Not days, not the week a video posted: quarter over quarter, for two years. Manufacturers felt it as stockouts rather than as a chart. Good Culture told customers that "demand has been WILD" while it worked to get back in stock, and Organic Valley, which reported cottage cheese sales up more than 30 percent in the first half of 2025, described the product as selling faster than it could be made.
The planning response to this shape is a permanent facing change and a higher baseline order, not a one-time buy.
The spike: Dubai chocolate
The other shape is steep and short. Dubai chocolate, the pistachio and kataifi filled bar that spread through food TikTok, produced demand sharp enough to be credited with a global pistachio shortage. As one measure of the concentration involved, Dubai Duty Free reported roughly 22 million dollars of Dubai chocolate sales across specialty brands in the first three months of 2025, which it put at over 1.2 million bars.
That figure is travel retail in one airport operator, not a US grocery number, and the distinction is the point. Published trend figures almost never describe your stores. They tell you a trend is real. They do not tell you what it is worth in your banner, in your region, at your price.
Butter boards belong to the same spike family. Freeze-dried candy sits less cleanly: the same roundup has its market forecast to keep growing, so treat it as an open question rather than a bounded buy. For a true spike the planning response is a bounded buy with an exit, not a permanent reset.
How food trends on TikTok reach your shelf
A trend does not travel from a video to your register in a straight line. It passes through a shopper deciding to look for the item, a store having it, a distributor having it, and a manufacturer having made it. Each step adds time, and the last two are where a small operator loses the window entirely: by the time the item is reliably available through your distributor, the peak of attention has passed.
The chain, in the order it moves:
| Stage | What has to happen | Who controls it |
|---|---|---|
| Attention | The preparation or product spreads online | Nobody |
| Intent | Shoppers start looking for it in stores | Nobody |
| Distributor availability | The item enters the catalogue you buy from | Your distributor |
| Your order | You place it against a specific UPC | You |
| Scan data | The movement shows up in your own numbers | Follows the above |
Only one row in that table is yours. That is why the timing question is worth taking seriously: the single decision you control sits in the middle of a sequence that is otherwise happening to you.
The lag is not a defect. It is the opportunity. A signal that reaches your scan data late is, by definition, visible somewhere else earlier.
Telling the two shapes apart early
The expensive mistake is treating a spike like a sustained shift. The tells show up before the curve does:
- Is there a category behind it, or only a product? Cottage cheese was carried by a durable high-protein preference, which is why it kept compounding. A single confection has nowhere to go once novelty fades.
- Is the supply chain constrained? A trend that depends on a scarce input will be capped and then whipsawed. The pistachio squeeze behind Dubai chocolate is the case in point.
- Does it survive substitution? If shoppers accept a private-label or cheaper version, the demand is real and durable. If only one brand's version will do, you are trading a fashion.
- Is repeat purchase plausible? People buy a novelty bar once. They buy a breakfast staple weekly.
None of these is decisive alone. Together they usually separate the two shapes well before the sales curve does it for you.
Why watching TikTok yourself does not work
Every category manager has been told to spend more time on social media. As a way of sourcing grocery demand signal it does not scale, for three reasons.
- Volume. The number of food trends on TikTok that look like they might matter, per week, is far larger than the number that reach retail at all. Most die.
- No magnitude. A video with ten million views and a video with ten million views tell you nothing different about whether the item sells in your stores. Attention is not units.
- No translation. Even a correct read leaves the hard part undone. "Chili crisp is trending" is not an order. An order names a UPC, a quantity, a distributor and a date.
The third reason is where the money is. The gap between knowing about a trend and acting on it is where the value sits, and it is a data problem rather than a cultural-awareness problem.
From trend to order
The chain has four links, and each has to be explicit or the whole thing stays a hunch.
- The trend. A specific product or preparation gaining attention, dated.
- The catalogue match. Which items you can actually buy that serve it, by UPC, from the distributors you already use.
- The baseline. What those items do in your stores today, so the size of the move is measurable rather than asserted.
- The order. A quantity, against a specific distributor, with a review date.
Scout monitors food trends on TikTok and runs this chain automatically. A trend is matched against the KeHE, UNFI and retailer catalogues connected to your account, checked against your own movement for the items involved, and surfaced as an order recommendation with the quantity and the distributor already attached.
Worked example, illustrative
Sunrise Market is a fictional eleven-store operator used throughout this site. The numbers below are illustrative and are not drawn from any real customer's data.
A chili-crisp preparation starts moving on food TikTok. Sunrise carries two chili crisp SKUs, both slow, both bought through KeHE, both sitting at roughly 0.4 units per store per week.
Scout matches the trend to those two UPCs plus four more available through the same KeHE catalogue that Sunrise does not carry. It reads the current baseline, flags that the category is under-assorted relative to the signal, and produces a recommendation: raise the two existing items, add two of the four, with a quantity per store and a four-week review date attached.
The review date is the part that separates this from a guess. A spike-shaped trend needs an exit, and the way an operator gets hurt is by treating a spike like a sustained shift and holding inventory into the decline. Four weeks later, the same movement data says which shape this one was, and the order either becomes a permanent facing change or unwinds.
Reading the four-week review
The review is only worth setting if you know what you are looking for when it arrives. Three outcomes, three responses.
The new items moved and the existing two also lifted. The category grew rather than cannibalising itself, which is the signature of a real preference shift. Raise the baseline, keep the added items, and treat the facing change as permanent.
The new items moved and the existing two fell by roughly the same amount. Shoppers switched rather than added. The trend brought no new demand to the category in your stores, so the correct response is to keep the better items and drop the weaker ones, at a flat total order.
Nothing moved. The trend did not reach your trade area, or it reached it and your shoppers declined. Unwind, and record it. A trend that failed in your stores once is evidence about your shoppers, and the next similar signal should be weighted accordingly.
That third outcome is what keeps the approach affordable. Bounded buys with a review date mean a wrong call costs one order cycle rather than a season of dead stock.
What this does not tell you
Being straight about the limits:
- Attention is not demand. A trend Scout surfaces is a candidate, not a forecast. The baseline read is what makes it actionable, and the baseline comes from your stores.
- Availability governs everything. If the item is not in your distributor's catalogue, the trend is not addressable for you at any speed, and the recommendation will say so.
- Your stores are not the country. Regional and banner-level differences on trend items are large. A national trend figure is a reason to look, never a reason to order.
- Scan data is transactions, not people. It supports basket, velocity and category claims. It does not tell you the demographics of who bought, which needs loyalty-linked data or a panel.
Doing this in Scout
Trend recommendations arrive against the distributor feeds you already have connected, so the output is an order you can place rather than a report you have to interpret. The baseline read, the catalogue match and the review date travel with each recommendation.
Scout recommends and flags. It does not transmit the purchase order: the ordering system you already use does that, and the recommendation is an input to it.
Summary and further reading
- Viral trends reach scan data in one of two shapes. Sustained shifts, like cottage cheese, justify a permanent facing change. Spikes, like Dubai chocolate, need a bounded buy with an exit date.
- The lag between online peak and shelf demand is the window worth trading against.
- Published trend figures prove a trend is real. Only your own movement data says what it is worth in your stores.
- A trend becomes actionable when it names a UPC, a quantity, a distributor and a review date. Anything short of that is awareness, not a plan.
Further reading: responding to a category shift in assortment, where to get help with assortment, and measuring brand equity with scan data.
Sources: Circana cottage cheese figures and the Crawford quote via Entrepreneur; Dubai chocolate travel-retail sales via Newsweek; trend roundup via FoodNavigator.