Why a Food Creator’s Reach Was Only Part of Nutrinaut’s Story
When your app can estimate a meal’s nutritional content from a photograph, a food creator gives you an obvious demonstration. The meal is already on screen. The viewer understands what the camera is pointing at. The app can appear inside the video without a long explanation.
That was part of the appeal of working with Donnie Otero, known as @thatcrazysubguy. His sandwich content gave Nutrinaut a natural setting. I remember the TikTok and Instagram posts accumulating roughly three million combined views. That figure is my recollection of the campaign, rather than a newly reconstructed post-by-post total. Donnie on TikTok, Donnie on Instagram.
Our food-creator experiments also included @baebyy.e. I remember that collaboration too, although I have not reconstructed its exact published posts and results. I keep it in the history without attaching Donnie’s reach figure to it.
The remembered reach was substantial. To understand what we learned, I have to keep several different questions beside it.
The first question was whether the product made sense on screen
Food scanning gave us a simple interaction to show: a meal, the phone, and an estimate that could become a saved entry. It was easier to demonstrate than a long list of app features.
The setting also made the product’s limits matter. The video had to describe an estimate, with a portion and ingredients someone could review. A compelling demonstration should not turn a food photograph into a promise of exact measurement or guaranteed weight loss.
This is where a brief does useful work. It identifies the actual interaction we want shown and checks that the words match what the app does. The creator can then fit that moment into their own style.
The next question was what the viewer noticed
There is a small, concrete example from another part of our creator work. Under BB Kickers’ “FOOTBALLERS FULL DAY OF EATING,” published on 9 June 2025, a viewer wrote:
Love the vid and calories tracking app!
That is a response to BB Kickers’ video, not Donnie’s campaign. I keep the examples separate. The comment shows that at least one viewer noticed the app as part of the food content. It does not establish whether that person installed it, continued using it or paid.
This is the distinction I want a campaign record to preserve. A public response can be useful evidence at its own level without being promoted into a sale.
Combined views still describe a particular thing
Adding TikTok and Instagram counters can describe recorded plays across the posts. It does not remove repeat plays or people who watched on both platforms. It also does not establish that every viewer reached the app segment.
For the remembered three-million figure, the proper next step is to reconstruct the exact posts, their dates and the counters at a stated time. A current profile total would answer a different question. So would the creator’s following today.
I would keep five stages in the record:
| Stage | Evidence worth keeping |
|---|---|
| Delivery | The exact published post, date and agreed placement |
| Attention | A counter with its platform, definition and observation time |
| Recognition | A relevant comment or a recorded answer naming the creator |
| Product use | Reliably attributed visits, installs or activity, each labelled correctly |
| Business result | Confirmed payments and other relevant outcomes compared with the collaboration’s total cost |
Our own source records add a different piece of the story
On 3 November 2026, we checked the retained Nutrinaut source-questionnaire logs. The records cover 15 August 2025 onward. A conservative match for Donnie’s name and established handle found four creator-name responses, associated with four distinct device-and-app identifier pairs. One response accompanied an Instagram choice and three accompanied TikTok choices; the matched responses run from 14 November 2025 to 19 February 2026. Two additional Instagram answers used shorthand variants that we kept in a separate probable-match group, rather than silently adding them to the strict count.
That is a small, specific observation beside a large remembered reach figure. It is tempting to divide one by the other. I would not do that: the view total has not been reconstructed, the dates do not establish matching campaign windows, and a creator-name question is optional. The logs also have incomplete client coverage. Four recorded answers cannot stand in for every person who encountered the campaign.
What the records do add is a trace of recognition inside the product. Someone submitted a source answer naming the creator. That is further along the path than a public view counter, while still being different from a verified purchase. It gives us a better question to investigate rather than a conversion rate to announce.
A delivered asset and a published campaign are different records
Our earlier UGC work also taught us a practical production lesson. In an October 2024 delivery, the file available through the sharing route was only 320?568. We asked for the original 1080?1920 export and kept separate plain and music versions. The higher-resolution re-delivery was checked before we moved on.
That was an asset-delivery problem, rather than a public view-counter problem. Keeping the two kinds of work separate makes the campaign history more useful: a finished file establishes that an asset was delivered; a published post establishes where an audience could encounter it.
What I would change in the next food-creator brief
I would give more attention to the transition after the demonstration.
The creator should show one interaction accurately. The viewer should be able to recognize the app’s name and find the right destination for their phone. The destination should explain the same product they just saw. We should preserve the published artifact and its links before moving on to the next campaign.
I would also decide in advance what question the collaboration is meant to answer. Testing whether a feature is understandable is different from testing whether the channel can support repeat spending. Both can be worthwhile, but they require different evidence.
Donnie’s campaign remains one of the clearest examples I remember of Nutrinaut attracting substantial short-form attention. What I want to carry forward is a better record of the path from that attention to the product. That would let us discuss the next collaboration with more than a large number and an impression of how it went.
Source note: I work on Nutrinaut. The collaborations with Donnie and @baebyy.e, and the approximate three-million reach for Donnie, are my account of the work. Questionnaire counts come from our 3 November 2026 snapshot and use exact normalized matches to Donnie’s confirmed public identity. Device-and-app pairs are a technical deduplication measure, not a count of people, installs or customers. The BB Kickers example is a separate public post.