The industry has been measuring the wrong thing with the wrong dataset

In a NECTAR panel room somewhere in North America, a consumer sits alone in a booth and scores coded, unbranded samples zero to nine for acceptability. The panel is the most serious consumer sensory work in alt-protein: 122 products, 14 categories, 2,684 omnivores. The methodology is rigorous. The data is the industry standard. The 2025 dataset is publicly cited by every major plant-based brand and every major investor following the category.

And yet the NECTAR panel measures acceptability in a single bite, in a controlled environment. It does not measure whether the same consumer walks back into a supermarket three weeks later and buys the product again. That distance, between the booth and the till, is the whole story the industry has been misreading for a decade.

For ten years the conventional wisdom in alt-protein has been that taste is the barrier. Product roadmaps, investor pitch decks, R&D hires, and reformulation budgets have all been organised around closing the taste gap. It is not unreasonable. Consumers say taste matters. Press coverage reinforces the taste frame.

But the NECTAR 2023-to-2025 year-over-year data says something different. Taste scores have moved meaningfully closer to animal-protein benchmarks across fourteen categories in two years. Texture scores have not.

Start with the mechanism. Taste is a formulation and flavour-chemistry problem. Fat-system replacement, Maillard management, umami loading, aroma-compound engineering - the science is mature and the toolkits exist. Texture is a structural engineering problem. A plant-based protein is not meat fibre. Simulating muscle architecture at commercial scale is an unsolved problem in most categories.

Look category by category. Plant-based bacon has approached taste parity in premium products, but the pan-fry render, the crisp, the specific browning pattern - those are texture failures that no amount of flavouring fixes. Plant-based cheese melts are where the texture gap is most visible because texture IS the product (melt, stretch, elasticity). Plant-based nuggets close the gap faster than burgers because the breaded coating is itself the primary sensory experience; texture failures underneath are masked. Plant-based dairy has improved on taste more than on texture in solid formats.

Texture is the gate the category has not cleared. Taste gets consumers to try. Texture is what decides whether they come back.

Now the measurement. Blind taste panels measure whether a consumer will accept the product once. Repeat purchase measures whether they will accept it again. One is noise. The other is revenue. The scan-panel providers - Circana, Kantar, NielsenIQ, Numerator - instrument SKU-level repeat behaviour across hundreds of retailers. The data exists. What it shows for alt-protein is an uncomfortable pattern: most plant-based brands post first-trial rates of 25-40 percent and repeat rates well below conventional protein benchmarks. A brand with 35 percent trial and 12 percent repeat has a declining customer base, not a growing one. The industry has been reporting trial and calling it traction.

And the connection to texture: consumer research on lapsed trialists - those who tried once and did not buy again - consistently finds texture cited more often than taste as the primary reason for non-repeat. The repeat-purchase curve and the texture gap are the same problem, measured from opposite sides.

The industry has been measuring the wrong thing with the wrong dataset. The data that matters has been sitting in sensory panels and store loyalty feeds all along.

The most misunderstood company in Food Biotech right now is the premium-priced, low-volume, texture-obsessed brand.

The investor consensus has said for five years that premium niche will not scale. Mass-market volume is the stated path. But the NECTAR repeat-purchase data suggests premium texture-obsessed brands cluster in the upper tier of repeat rates. They have smaller absolute volumes but their customers come back. The mass-market low-price brands are the ones stalling - strong first trial, weak repeat, steep occasion compression.

Names to watch: Juicy Marbles (Slovenia, whole-cut plant steaks), Redefine Meat (Israel, 3D-printed texture cuts), MyForest Foods (US, mycelium bacon), Heura (Spain, premium-tier category dominance). All of them are betting the company on texture.

This episode continues a four-piece FoodEdge arc. Issue 20 (rent before build), Issue 21 (downstream processing moat), Issue 23 (platforms should start as ingredient companies), and now S02E06 (taste, texture, repeat purchase). Each one is about commercial discipline coming before optionality. Read the prior posts on FoodEdge before or after listening.