STRATEGIC SCIENTIFIC EVALUATION · EVALUATION NOTE 01
How do I evaluate a novel hair growth ingredient before licensing it?Seven tests, in the order I would actually run them.
The honest version of this question is not the one people type. What a founder is actually asking, at eleven at night, with a term sheet open in another window, is: am I about to spend money I cannot get back on something that was never going to work?
I have been on both sides of that sentence. I have developed and patented actives, published the mechanism work, and taken formulations from bench through clinical evaluation and onto pharmacy shelves internationally. I have also written the memo telling a client to walk away from something they had already emotionally purchased. The second is harder and worth considerably more.
An ingredient does not fail because the biology was wrong. It fails because nobody asked, early enough, which of seven separate things had to be true simultaneously.
What follows is the sequence I actually use. It is deliberately ordered by cost of discovery: the cheapest disqualifying questions come first, because the purpose of an evaluation is to fail fast on the things that cannot be fixed, not to build a beautiful document.
The seven tests, in the order I run them
Test 01Is the mechanism causal, or merely correlated?
Most ingredient dossiers open with a pathway diagram. A pathway diagram is a hypothesis with good graphic design. What I look for is whether anyone demonstrated that modulating the target changes the outcome — knockdown, knockout, antagonist rescue, dose-dependent reversal. If every citation is an association study or an in silico docking model, the mechanism has not been established; it has been asserted.
Test 02Was the concentration tested one a finished product can actually deliver?
This is where most licensing money is lost, and it is almost never in the summary. An active that performs at 50 µM on isolated dermal papilla cells may be undeliverable at the surface of intact stratum corneum at any tolerable formulation percentage. Ask for the exposure arithmetic: applied concentration, vehicle, penetration data in a relevant model, and the resulting tissue concentration. If nobody can produce that chain, you are licensing a cell-culture result, not a product.
Test 03Is the clinical work designed to detect a real effect, or to survive scrutiny?
Randomized, vehicle-controlled, adequately powered, with a pre-specified primary endpoint and blinded assessment. Anything else is supportive at best. In hair specifically: macrophotographic target-area hair counts under standardized conditions, not global photography scored by the sponsor, and not self-reported satisfaction. A 12-week study in a biology with a multi-month cycle is a safety study wearing an efficacy costume.
Test 04Does the intellectual property cover the thing you are buying?
Issued is not the same as valuable. Read the claims, not the abstract — composition claims covering the exact ratio and form you intend to sell, or method-of-use claims that a competitor can design around by changing a diluent. Then check the family: jurisdictions, remaining term, continuation status, and whether the prior art the examiner did not find is sitting in the licensor’s own earlier publications.
Test 05Can the claim you need to make actually be substantiated?
Work backwards from the label. The claim you must make to sell the product determines the evidence you must own — and in the regulated channels that matter commercially, a mechanism claim you cannot substantiate is a liability that travels with the asset. If the evidence supports “supports the appearance of” and your commercial model requires “regrows,” the deal is mispriced.
Test 06Is the effect size worth a commercial life?
Statistical significance is not commercial significance. A change that requires instrumentation to detect will not be perceived by a consumer, will not generate repeat purchase, and will not survive contact with a category already containing molecules with decades of clinical evidence. Ask what the effect looks like to the person paying for it.
Test 07Who has already walked away from this asset, and why?
Novel technologies with genuine promise are rarely virgin. If the licensor has been shopping it for three years, someone competent has looked and declined. Ask directly. The answer, or the discomfort produced by the question, is frequently the most informative datum in the process.
The three failure patterns I see most often
The translated-species error
A result obtained in rodent skin, or in an immortalized cell line, is treated as a human result. Hair follicle biology is especially unforgiving here: cycling dynamics, androgen handling, and immune privilege differ enough that a clean murine result carries far less predictive weight than its position in the dossier implies.
The composite-endpoint illusion
When a study reports improvement in a blended score assembled after the fact, the individual components almost always tell a duller story. Ask for the components. If they are unavailable, assume the reason is not administrative.
The exclusivity that is not exclusive
A supply agreement described as exclusive that is, in fact, exclusive to your territory, your channel, and your one SKU — while the same material ships to a competitor next quarter. This is a contract question with scientific consequences, because it determines whether you are building a brand or renting a commodity.
What I would ask for, in writing, before signing anything
- Full study reports — not summaries, not slide decks — for every study cited in support of efficacy.
- The exposure chain: tested concentration, vehicle, penetration model, and estimated tissue concentration in human skin.
- The complete patent family with claim text, term, jurisdictions, and prosecution history.
- Certificates of analysis and specification ranges across at least three production lots, with the analytical method named.
- Stability data in the intended finished-goods matrix, not in a solvent.
- A written statement of who else has evaluated the asset in the past thirty-six months.
- The regulatory position in every market where you intend to sell — the classification determines the claim, and the claim determines the model.
If seven documents feel like an aggressive ask on a licence you are about to pay for, that reaction is itself a finding.
The part almost nobody writes
What is the decision?
Given the available evidence, here is how I would advise a client — and if it were my own money, here is what I would do.
Do not license on mechanism. Mechanism buys you a story; it does not buy you a product. Pay for an option, not an asset: a short, defined, jointly-designed exposure-and-delivery study in the finished-goods matrix you actually intend to sell, with the licence priced and gated on its outcome. If the licensor will not structure it that way, they are asking you to underwrite a risk they have already declined to underwrite themselves — which tells you what they believe.
If Tests 2, 3, or 5 fail, I would stop. Deliverability, trial design, and claim substantiation are not deficiencies you can market your way out of; they are the product. Tests 1, 4, 6, and 7 can be repaired with time, money, or renegotiation. The first three cannot, and every founder I have watched lose a large sum lost it by deciding the exception applied to them.
And to the question underneath the question: no, you are not crazy for asking. The founders who lose money are almost never the ones who asked too many questions.
References & primary record
The judgments above rest on a checkable record. These are peer-reviewed papers authored or co-authored by Dr. Marcovici on the mechanisms discussed in this note.
- A Randomized, Double-Blind, Placebo-Controlled Trial to Determine the Effectiveness of Botanically Derived Inhibitors of 5-Alpha-Reductase in the Treatment of Androgenetic Alopecia
Prager N, Bickett K, French N, Marcovici G. J Altern Complement Med. 2002;8(2):143–152.
- Inhibition of Inflammatory Gene Expression in Keratinocytes Using a Composition Containing Carnitine, Thioctic Acid and Saw Palmetto Extract
Chittur S, Parr B, Marcovici G. Evid Based Complement Alternat Med. 2011;2011:985345.
- Blockade of Androgen Markers Using a Novel Betasitosterol, Thioctic Acid and Carnitine-containing Compound in Prostate and Hair Follicle Cell-based Assays
Marcovici G. Phytother Res. 2016.
- An Uncontrolled Case Series Using a Botanically Derived β-Cyclodextrin Inclusion Complex in Two Androgenetic Alopecia-Affected Male Subjects
Marcovici G, Bauman A. Cosmetics. 2020;7(3):65.
Full list: publications and contributions · research and scholarship.
Where this connects
This note is one piece of a connected body of work. The evidence behind each judgment is published and checkable.
Hub
Strategic Scientific Evaluation
The practice this note belongs to — what gets evaluated, in which fields, and on whose judgment.
Mechanism
Intellectual property programmes
Senescence biology, intracrine dynamics, autoimmune privilege, and multi-nodal stabilization — the mechanism work behind Tests 1 and 4.
Translation
Commercial track record
Formulations taken from mechanism to pharmacy shelf across the GCC — the reason Tests 2, 5, and 6 are written the way they are.
Questions decision-makers ask
Short answers to the questions this note is most often searched for.
How do you evaluate a hair growth ingredient before licensing it?
Run seven tests in order: is the mechanism causal or merely correlated; was the tested concentration one a finished product can deliver; was the clinical work designed to detect effect or to survive scrutiny; does the intellectual property cover what you are buying; can the claim you need be substantiated; is the effect size worth a commercial life; and who has already walked away from the asset, and why.
Why do ingredients that work in vitro fail in finished products?
Because the concentration that produced the in-vitro effect is frequently one no stable, tolerable, regulatorily acceptable finished product can deliver to the target tissue. Deliverable exposure — not intrinsic potency — is the governing constraint.
Does an issued patent mean an ingredient is protected?
No. An issued patent means claims survived examination, not that they cover the commercial thing you intend to sell. Read the granted claims — not the title or abstract — and confirm they cover the composition, concentration, indication, and territory you are paying for.
What effect size justifies licensing a hair growth active?
One large enough to survive the gap between trial conditions and consumer conditions, and to support the price, claim, and marketing spend the programme requires. A statistically significant but clinically invisible change is a scientific result, not a commercial asset.
Enquiries
Have the evaluation before the wire transfer
If you are weighing a licence, a technology, or another tranche of spend, describe the decision in a paragraph. Enquiries are read personally by Dr. Marcovici.