

Every company wants clinical evidence. Not every project is ready for it. The question worth asking before opening a budget for a human study is more uncomfortable and more useful: do we already have enough evidence to know what we are going to measure?
A human study answers a narrow question: in this population, at this dose, for this duration, does the chosen endpoint move? It does not tell you which endpoint is the right one, it does not suggest the dose, it does not explain the mechanism. All of that has to be decided beforehand, and decided badly it gets paid for later — with a study that costs what was planned and concludes less than what was needed.
Whether the ingredient does something, and on what. Cellular and tissue models tell you whether the hypothesised activity exists and which mechanism it acts on: inflammation, oxidative stress, gut barrier, microbiota, metabolism. Looking at one endpoint rarely tells the whole story.
Which of several candidates deserves to go on. A well designed screening drops early what does not hold, and it is the cheapest way to be wrong.
At what dose and in what form. Behaviour in the final matrix, stability, the share that stays active: all of it changes the design of the study that follows.
What to measure in humans. This is the most underrated contribution: mechanistic evidence points to the biomarkers worth following, and a clinical study with well chosen endpoints needs fewer subjects to say something.
It does not demonstrate a benefit in humans, and that has to be said plainly. It does not account for absorption, metabolism and distribution, which can cut a brilliant cellular activity down to size. And it does not replace a clinical study when strong substantiation is required: it prepares it.
There is an opposite risk too, that of piling up in vitro endpoints without a question behind them. More unvalidated endpoints means noisier data, not stronger evidence.
What decision will this data have to support: continue, reformulate, position, patent? Do you already know which mechanism you want to talk about, or is that exactly what needs establishing? Is the formulation final, or will it change again after the study? And what happens if the result is negative: is that information you need early or late?
Every project starts with a scientific discussion, not with a protocol. You establish what decision the data has to support, choose the model accordingly — cell-based assays, 3D intestinal and mucosa models, prebiotic microbiota tests — and prefer a few robust endpoints to many uncertain ones. When a clinical study is needed, the work done beforehand serves to design it better, not to avoid it.
If you are building evidence around an ingredient, the page on ingredient validation describes the path; if the finished product already exists, supplement valorisation is the right place to start.
Technical content and news from our laboratories, sent when there is something worth saying.