Five questions, asked in order, before anything else.
The Elevora Decision Framework is published in the front matter of our flagship reference and governs every chapter after it. It's reproduced here because it's the most portable thing we've written — it works on compounds we've never covered, and on claims made somewhere else entirely.
Why bother naming it
A library with no shared structure would be a stack of opinions in matching covers. What stops that is a fixed order of questions, so that opening a title you've never read, on a system you know nothing about, you still recognise the shape of the argument.
Five questions, and the order isn't decorative — each hands the next what it needs. Skipping one is the most common failure in health reasoning by a wide margin. Take almost any confident claim that turned out to be wrong, and you can usually name the question it jumped.
None of this is novel. It's what careful thinking already looks like. Writing it down made it repeatable, and made it obvious when a draft had cheated.
Two other sequences run alongside it, and it's worth knowing which does what. The five biological layers from the Female Optimization Framework describe where to look — nervous system, hormonal context, cellular health, inflammation, and only then modulation. The Decision Framework describes how to reason once you're looking. And every compound chapter in the flagship follows a third fixed order: plain English, mechanism, benefits, limitations, dosing, synergy — with limitations arriving before the dose rather than after it.
Five questions, in order.
What biological process am I trying to influence?
Say what you're actually after in physiological terms, not product terms. "Improve recovery" isn't an objective — it's a mood. "Shorten the inflammatory phase after mechanical loading" is something you could be wrong about, which is what makes it useful. Everything downstream inherits the precision of this sentence.
What level of evidence supports this approach?
Human, animal, or cell culture — and reproduced by how many independent groups? A mechanism that makes biological sense is a reason to investigate, not a reason to conclude. We label the model behind every substantive claim, because the same sentence carries very different weight depending on where it came from.
What outcome will tell me it is working?
Decided before you begin, not after. An outcome chosen in hindsight isn't a finding, it's a story — and one is remarkably easy to build out of ordinary variation. This question is also what makes the next one answerable.
What variables could limit the response?
Sleep, nutrition, stress load, medication, hormonal phase, training. Usually the actual explanation when something that should have worked didn't — and almost always cheaper to address than whatever is being considered instead.
Is there a simpler intervention that should come first?
The question with the least commercial interest behind it, and the one most likely to change what you actually do. If comparable progress is available from something better established, added complexity rarely returns proportional benefit.
Observe, and say what you don't know
Get a baseline and watch before adding anything; change three things at once and you've learned nothing about any of them. And where the literature is thin or contested, say so. Uncertainty is information — knowing a question is unsettled leaves you better placed than false confidence does, even though it feels worse.
The five questions are the analytical core. Observation and stated uncertainty are the practices that keep them honest.
Four kinds of knowing, labelled.
These appear throughout the library. They describe where a finding came from and how far it reaches — not whether it's true. A cellular result can be entirely correct and still tell you almost nothing about a person.
Structured observation
Understanding a system is worth something on its own. But at some point you're the only available data about you, and that's where this stops being reading and starts being work.
It means deciding what you'll record before you start, keeping conditions as steady as real life allows, waiting long enough for a real signal to separate from ordinary noise, and writing down the context you'd otherwise try to reconstruct from memory in four months. Most personal health records fail here — not through carelessness, but because nobody ever designed them to be read back.
None of this was invented for biology. It comes from behavioural observation, where our founder worked for two decades before any of this: establish a baseline, record what you see rather than what you expected, change one thing at a time, treat an unexpected response as data rather than as failure. It's a professional discipline with its own standards, and it transfers to physiology almost unchanged.
Several publications and tools exist for no other reason. They help you keep the record. They don't diagnose anything, don't tell you what your numbers mean clinically, and don't recommend action — a deliberate limit, and one we've turned down obvious ideas to hold.
A method is only worth something applied. Three places to see it working.
-
Curriculum
The nine stages
How the reasoning sequence maps onto an actual reading order.
-
Library
Publications by level
Every title marked with what it assumes — the method's most practical output.
-
Transparency
How we create
The editorial process this method governs, including what tools we use and where judgment stays human.