You can buy the tests. That's true.
Let me start with the honest part.
- The tests aren't secret. Blood chemistry, a GI-MAP stool test, DUTCH hormone testing, an organic acids test and a food-response panel can all be bought, some directly, some through any practitioner. The five-test approach I use is close to what FDN (the functional diagnostic nutrition training I did) teaches. I didn't invent it.
- Blood-testing services are good at what they do. In the US, Function Health offers "160+ lab tests annually", tested twice a year, for $365 (Function Health, Nov 2025). UK services offer smaller versions of the same idea. For tracking blood markers over time, cheaply and conveniently, that's hard to beat, and I'd say so to a client.
So if the question is "can I get the data without you?", the answer is often yes.
More results aren't more answers
Here's the part the "order everything" approach doesn't tell you.
A normal range on a lab report is usually set to include the middle 95% of healthy people. That means 5% of healthy people fall outside it on any single test, by definition. Order more tests and the chance that something comes back flagged climbs fast:
| Number of tests | Chance at least one is "abnormal" in a healthy person |
|---|---|
| 1 | 5% |
| 20 | about 64% |
| 100 | over 99% |
That's simple arithmetic (1 − 0.95n), and it assumes the results are independent. Real markers are linked to each other, so the true figure is a bit lower. But the direction doesn't change.
So a big panel almost guarantees a result that looks like a finding. Without a reason to test, and a way to decide what matters, more data mostly means more worry, more follow-up, and more supplements for things that were never wrong.
What the investigation adds
An investigation does three things a pile of results can't:
- It starts with a question. Your history, symptoms and timeline decide which tests are worth doing. Sometimes that's one test, not five.
- It looks for results that agree. One odd marker on its own is weak. Several findings from different samples pointing at the same problem, and matching your symptoms, is stronger. Findings that don't fit are noted, not forced.
- It knows when the answer is outside functional testing. Sometimes the most valuable thing a set of results does is send you back to your GP with a specific question.
The five tests I use look at different parts of the body, from different samples: blood, urine, saliva and stool. That's the real value of using more than one: they can't all be wrong in the same way.
And the tests themselves have limits, mine included. The only independent check of the GI-MAP stool test's pathogen panel I know of sent it stool samples spiked with known bugs, alongside samples with none. It picked up 80% of the real ones, but it also reported pathogens in many samples that had none: a specificity of just 26%, against 100% for a hospital-grade comparison test (Gingras & Maggiore, 2020). It was a small study, and it's only one. But it's why I treat a pathogen on a GI-MAP as a lead to confirm through your GP, not a diagnosis to treat.
A worked example
A woman in her forties comes in with tiredness and bloating. She's been told her bloods are "fine". Towards the end of the history she mentions, almost as an aside, that her bowels have been looser for about a year. She'd stopped mentioning it because nobody picked it up.
What the results show, read one at a time:
- Blood: haemoglobin normal, so not anaemic. But ferritin, the iron store, is low.
- Stool: calprotectin raised. Calprotectin is a marker of inflammation in the gut.
- Hormones: a cortisol pattern that's a little flat.
Read one at a time, each is easy to shrug off. Low-ish ferritin: take some iron. Flat cortisol: stress. Raised calprotectin: one number on a long stool report.
Read together, they tell a different story: iron that isn't being absorbed or is being lost, a gut that's inflamed, and a change in bowel habit. That combination isn't something to fix with supplements. It's a reason to see her GP, with a clear request: check for coeliac disease and investigate the gut inflammation properly.
Why calprotectin and not something more fashionable? Because it's one of the few stool markers with solid evidence behind it. In a meta-analysis of 18 studies, faecal calprotectin identified people with inflammatory or other "organic" gut disease, as opposed to IBS, with about 81% sensitivity and 81% specificity (An et al., 2019). NICE recommends it for exactly this job: helping tell inflammatory bowel disease from IBS in adults with recent-onset lower gut symptoms, when cancer isn't suspected (NICE DG11). And NICE lists unexplained iron deficiency as a reason to test for coeliac disease (NICE QS134).
And even then, it isn't a diagnosis. The same study showed that if only 1% of people tested actually have the disease, most positive results would be false alarms. That's why a raised calprotectin leads to a proper investigation, not a label.
The flat cortisol? Noted, and parked. It might matter once the gut is sorted. It might just reflect a year of feeling unwell. It doesn't change what happens next.
That's what I mean by the investigation: deciding which result leads, which ones support it, which one to set aside, and when to hand over.
"I asked ChatGPT about my results"
More people now paste their results into an AI chatbot before they ever speak to anyone. That's understandable, and it's often a decent start: a general chatbot can explain what a marker is and what a range means.
What it can't do is the investigation. It sees one set of numbers, not your history, your medications or your last three results. It can't tell whether a change is real or within the normal day-to-day wobble. And it tends to give a confident answer, with a threshold and no source, even when the honest answer is "that depends" or "see your GP".
Symmetry: my own assistant, AIdan, has failed some of the same tests in my evaluations. That's why it's built to say "I can't answer that" more often, and why I check it. If you've already got results and an AI's reading of them, the Health Audit is the human version.
What I used to say, and what I say now
I used to describe results that agreed across tests as "confirmation" of the root cause. That was too strong. Agreement across tests raises my confidence. It doesn't prove anything, and it certainly hasn't been validated as a method. The honest version is less exciting and more useful: several independent signals pointing the same way, matched to your story, is a better reason to act than any single result.
So what are you paying for?
Not the tests. You can buy those. You're paying for:
- the question that decides which tests are worth doing;
- the reading that separates a real pattern from the 5% of noise every panel produces;
- the restraint not to treat every flagged number;
- the referral when the answer belongs with your GP;
- and the follow-through: retesting what matters, when it matters.
If you only need blood markers tracked, a blood-testing service may be all you need, and I'll tell you that. If you've been tested plenty and still don't have an answer, it's usually not more tests you're missing.
Declared interest: I sell testing and the consultations that go with it. Everything above applies to my own tests too, and the grades are the same ones I'd give anyone else's.
If the tests you've had haven't added up, that's what the free 20-minute call is for. More on the approach: Treating the Test Result vs Reading the Pattern · The TDG programme.