Imagine a shake with 30 grams of protein. Someone drinks half, screws the cap back on and leaves the rest in the fridge.
So far, they have consumed 15 grams, assuming an evenly mixed product.
The formulation has met its target. Has the product done its job?
That is the question I would put at the top of a GLP-1 food brief. Before choosing another ingredient. Before designing a badge for the front of the pack. Before deciding that this needs to be a new product at all.
The half-finished shake is a hypothetical example. The measurement gap is real: what a product contains and what someone eats are different quantities. Three studies published in 2026 make that distinction worth a closer look.
A better shopping basket can still leave a nutrition gap
A Danish study published in January 2026 examined supermarket receipts from 293 people starting GLP-1 receptor agonist treatment and 884 matched comparisons. Among initiators, protein density in purchases rose from 6.6 to 6.9 grams per 100 grams of food. Purchases were tracked for a year before and after initiation, with prescriptions identified in 2019-2022.
A useful finding. But a receipt cannot tell you who ate the food, how much they ate or what was left on the plate.
In a small Japanese study published in August 2026, 25 people with type 2 diabetes were followed for three months after starting oral semaglutide. Reported daily energy intake fell from 2,032.5 to 1,618.2 kcal. Protein intake fell from 76.6 to 63.6 grams. The proportions of macronutrients did not change significantly.
That is the distinction a product team needs to understand. The share of protein in a diet can stay much the same while the number of grams eaten falls.
The US CRAVE study followed adults starting semaglutide or tirzepatide for 24 weeks without a structured nutrition intervention. In its analytic sample of 28, overall diet quality did not significantly improve; energy and several micronutrient intakes declined.
These are different people, medicines and study designs. The Danish receipt sample was self-selected and lacked BMI data. The Japanese study was small and uncontrolled. CRAVE was exploratory, with a small sample and attrition. We cannot join them into one clinical story. We can use them to stop asking three different measurements to answer the same question.
| Measurement | What it tells you | What it leaves unanswered |
|---|---|---|
| Food purchased | What entered the shopping basket. | Who ate it, and how much? |
| Nutrients consumed | The contribution of the amount actually eaten. | Was the overall diet adequate? |
| Diet quality | How the dietary pattern scores on the measure used. | Did a particular product cause a benefit? |
My reading for food R&D is straightforward: put actual consumption in the brief. A stronger nutrition panel is worth investigating. It is not the end of the investigation.
A portion makes sense inside an eating occasion. Find out what is chosen, eaten and left behind.
Dani Rendina / Unsplash. Illustrative photograph.The serving is written. The portion is chosen.
Consider two hypothetical products. One offers more protein per pack. The other contributes more in the amount consumed.
Which number would your brief record?
- Product A: 30 g per pack, half consumed, 15 g consumed.
- Product B: 20 g per pack, 90% consumed, 18 g consumed.
This does not make B the better product. It makes consumption the missing number.
A smaller portion might fit an occasion better. It might also cost more per gram, leave someone dissatisfied or fail to justify another purchase. A resealable pack might help someone return to the rest later. It might simply become another unfinished item in the fridge.
Those are useful research questions. A universal rule that every GLP-1 customer wants a smaller pack is not.
Nor should finishing the pack become the target. Someone may stop because they have had enough. The job is to offer food that fits, not persuade them to eat past comfortable fullness.
What happens after the first few bites?
A first-sip liking score tells you about the first sip. Yet the development decision concerns a portion someone will choose again.
I would want to know when the experience changes. Does the drink become tiring? Does the mouth-coating build? Does the aftertaste linger? Was the amount sufficient, or did the eating occasion end? An unfinished portion needs an explanation before it becomes a reformulation brief.
Measure the amount offered, consumed and left. Follow liking and the relevant sensory attributes across the portion. For a product with separate components, record which components remain. A spoonful of cereal left behind is not necessarily equivalent to a spoonful of yoghurt.
Then return to the commercial question. Does the format remain useful on later occasions? Will people choose it at the proposed price? Purchase intent, an observed purchase and repeat purchase need separate reporting.
The unfinished half belongs in the product brief, alongside the protein number.
Define the population and use occasion before recruiting. Treatment, duration and dietary context matter; a convenience sample cannot stand for everyone taking these medicines. Use an appropriate research protocol and nutrition expertise. The 2025 joint clinical advisory emphasises individual assessment and nutrition support. It does not certify a snack or shake.
Follow texture, aftertaste and liking through the chosen portion. Then find out whether people choose it again.
Ellie Ellien / Unsplash. Illustrative photograph.Does the business need another product?
For an ingredient supplier, this makes the application brief sharper. Can the protein reach the intended concentration without spoiling flavour or texture? Will the result survive processing and storage? Is there a measured problem that the ingredient solves?
The same discipline applies to fibre and packaging. A gut-health association does not establish tolerance during treatment. A smaller or resealable pack is a design option, with freshness, food safety, waste and cost to work through. Neither earns a place merely by sounding relevant to GLP-1.
Before funding a new line, I would put four options on the table:
- Retain. The existing product may already fit. Find out before changing it.
- Reposition. A clearer occasion may be more useful than another formulation. Check the evidence and claims before changing the message.
- Reformulate. Fix a measured problem with portion, texture, composition or stability. Define success before selecting an ingredient.
- Create. Build something new when an unmet need survives that review, and demand, feasibility and economics justify the work.
These are editorial choices to investigate, not a validated scoring system. A fast-growing treatment category does not establish demand for your particular food, let alone willingness to pay a premium for it.
Write the missing question into the brief
Before commissioning the next formulation, write down the occasion, the intended portion and the comparison that would change your decision. Measure what gets eaten. Ask why people stop. Find out whether they return.
A protein target tells the team what to put in the pack. The next question is what makes that pack useful enough to choose again.
Sources
- Sørensen and colleagues. Consumer Food Purchases After Glucagon-Like Peptide-1 Receptor Agonist Initiation. JAMA Network Open, 26 January 2026. DOI: 10.1001/jamanetworkopen.2025.55449.
- Sawada and colleagues. Effects of Oral Semaglutide on Dietary Intake and Body Composition in Japanese People With Type 2 Diabetes. Endocrinology, Diabetes & Metabolism, 4 August 2026. DOI: 10.1002/edm2.70295.
- Babazadeh and colleagues. The CRAVE study. Obesity Pillars, online 29 June 2026, September collection. DOI: 10.1016/j.obpill.2026.100292.
- Mozaffarian and colleagues. Nutritional priorities to support GLP-1 therapy for obesity. Joint advisory, 2025. DOI: 10.1016/j.ajcnut.2025.04.023.
For the next development review: what belongs in a sensory data room?
