The co-manufacturer that scaled you may not be the one that scales you next
Prebiotic soda reached about $820 million in roughly five years — around 2% of the $42.4 billion U.S. carbonated soft drink market, by Citi's estimate. PepsiCo bought Poppi for $1.95 billion. Olipop reached about $400 million in revenue in 2024. Coca-Cola entered with Simply Pop.
Then the category slowed. Beverage-Digest reported Poppi down in both dollar sales and volume in the first half of 2026, following a distribution transition and the arrival of larger competitors.
Almost every brand in that category scaled without owning a plant, which was the right decision. It is how you reach national distribution in five years instead of fifteen.
What is worth examining is a quieter consequence. Those brands selected a manufacturing partner based on one product, at one volume, with one set of requirements. The category has since moved toward smaller portions, higher protein and fiber, and faster reformulation cycles. The partner that was a good fit for product one is not automatically a good fit for products two through five.
This article is about how to tell the difference, and what to do if the answer is no. It is not an assessment of how any of those companies are run.
What demand is asking for now
Measured data, not projection:
- RTD high-protein shakes grew 71% in four years, from $4.7 billion to $8.1 billion (Circana, through 30 November 2025)
- 21% of U.S. households now include a GLP-1 user, up from 9% in January 2025 (PwC, with Numerator data)
- Among those households, 35% buy more packaged protein and 34% more fresh protein
- Greek yogurt +6.8%, jerky +7.9%, nutrition bars +3.4%, energy drinks +7.0%
- GLP-1 use more than doubled in 18 months, from 5.8% of U.S. adults in February 2024 to 12.4% in 2025 (Gallup)
And the counterweight: protein snacks reached a ceiling of consumer interest in 2026, according to Bakery&Snacks. Formats that fit an existing routine keep growing. Standalone novelty does not.
General Mills described the pattern from the other side: consumers trade down on staples and trade up selectively on products with a clear functional benefit.
What that changes about the plant you need
Translate the demand shift into production requirements and four things change.
Smaller portions mean more pack configurations. A single product in three sizes is three sets of change parts, three validation runs, and three separate minimums. A line optimized for one format handles this badly.
Higher protein and fiber content changes the process. Protein affects viscosity, suspension stability, and heat treatment tolerance. Fiber adds a dispersion requirement. A co-packer that runs conventional beverages well may not hold spec on either.
Faster reformulation cycles mean more changeovers. A schedule built for long runs of few SKUs treats every changeover as an exception. That works when you launch once a year and breaks when you launch quarterly.
Functional claims raise the documentation bar. Certification scope matters more than certification existence. A plant certified for one category is not automatically certified for yours.
None of these show up in a tolling rate comparison. All of them decide whether the next launch happens on time.
The screening gap
Most brands qualify a manufacturing partner once, at the moment of highest urgency — when the first retail commitment is signed and production has to exist. The criteria used are the criteria of that product.
Then the product line evolves and the qualification is never revisited. The relationship continues on inertia, and the constraint only becomes visible when a launch slips.
The practical test is simple. Take the three products most likely to be on your roadmap in the next 18 months. For each one, can you answer these without asking your current partner:
- Can their line physically run it, at what batch size?
- Do they have documented experience with that format or protein type?
- Is their certification scope valid for that product category?
- What is the lead time from formulation lock to first commercial run?
If you cannot answer three of the four, you do not have a partner problem yet. You have an information problem, and it becomes a partner problem at the worst possible moment.
Re-screening is not switching
Widening the qualified list is not the same as leaving your current partner, and most of the time it should not end that way.
A second qualified partner does three things: it gives you a place to run a format your primary cannot, it removes the scheduling single point of failure, and it changes the tone of every subsequent conversation with your primary.
Companies that treat external manufacturing as a core capability build that redundancy before they need it. Our CPG Leaders 100 data shows the pattern: three of the ten highest-ranked companies have no significant plant capacity of their own, and their Network scores are among the highest in the ranking. The gap between the average score of the top 10 (79.9) and the 51–100 group (47.1) is 32.8 points, and it reflects years of accumulated decisions about which partners to qualify and how consistently to use them.
The same question at enterprise scale
For a large operator this is a network question rather than a partner question, but the logic is identical.
If the innovation pipeline is moving toward functional benefit and smaller pack sizes, the useful question is not whether the network has capacity. It is whether the network has the right kind of capacity, in the categories where the pipeline is actually moving.
Most enterprise networks were built to optimize cost per unit at volume, which is correct for a stable portfolio and wrong for a portfolio in reformulation.
What to do this quarter
- List the three products most likely to launch in the next 18 months.
- Score your current partner against the four questions above. Note which you cannot answer.
- For any gap, identify two alternative plants with documented experience in that specific format and a certification scope that covers it.
- Run a pilot with one of them before you need it. A pilot run when you have time costs a fraction of a pilot run when you have a retail date.
Before that first conversation, we wrote down the twelve questions we would ask a co-manufacturer, in the order we would ask them, with the red flag that should end each conversation early.