Testing-Led or Guidance-Led: Choosing a Health Service Model

What Each Model Is Good At

Testing-led services exist because routine care frequently will not measure what you want. Apolipoprotein B, lipoprotein(a), fasting insulin and a urine albumin to creatinine ratio are the additions that change decisions and are commonly absent from a standard panel. A service that reliably provides those, through an accredited laboratory, with raw numbers and units, is solving a real access problem.

Their weakness is what happens next. A panel produces findings, findings require decisions, and a single set of results with an interpretive report attached does not constitute a decision process. Testing answers "what is my state", not "what should change and did it work".

Guidance-led services exist because the decision problem is genuinely hard: which candidates apply to your measured status, what interacts with your medications, what to change first, and whether it worked. Their weakness is that without measurement they are guessing, and their value depends entirely on the quality and consistency of the inputs.

The honest conclusion is that most people need both functions, and the question is which they are missing rather than which model is better.

Which Problem Do You Actually Have

The last row matters more than the rest. A person with irregular sleep and no resistance training who buys either kind of service is addressing the smallest available variable, and no service can supply adherence.

The scattered-results row is the most commonly unrecognised. Many people already have enough measurement and have never assembled it, and the value there is entirely in consolidation rather than in new tests.

The Structural Conflict Worth Checking

One question separates services more usefully than any feature comparison: does the interpreter sell the products it recommends?

Where a service interprets your panel and sells the supplements arising from it, the interpretation layer is not independent of the sales layer. That does not make the interpretation wrong, and it does mean the incentive runs toward addition rather than toward the correct answer, which is frequently no change or a removal.

What to look for as a counter-signal: whether a service ever recommends removing something, whether it ever concludes that no change is needed, and whether it has a published list of things it recommends against. A service with no exclusion list is not applying a filter.

The same applies to testing. A service whose panel is designed to generate findings, by measuring many analytes, will generate them, because reference ranges describe the central 95 per cent of a population and the more you measure the more falls outside.

And to devices. A platform selling a subscription has an incentive toward daily engagement, which conflicts with the correct review cadence of weekly for behavioural signals and quarterly for functional tests.

None of these is disqualifying. They are structural pressures worth knowing about before reading a recommendation.

The Non-Negotiables in Either Model

Raw numbers with units. A proprietary score or a traffic-light rating cannot be compared with a clinical result, tracked against a previous laboratory or read by a clinician. This is the single most important requirement.

A named, accredited laboratory, and the same one across time, since between-assay differences can exceed real biological change.

Assay-specific reference ranges, stated with the result.

A defined route for a serious abnormality. A service that flags a result is not a service that acts on it, and the gap between those has caused real harm through delay.

Data portability and a deletion path. If you cannot export the record, the asset belongs to the provider rather than to you, and the record is the thing that compounds in value.

Stated data handling, covering retention, research consent and third-party sharing. Health data, and methylation data in particular, carry information well beyond their stated purpose.

Medication checking, in any service making supplement recommendations. A service that does not ask what you take is not performing the highest-value safety function.

Where a Clinician Is the Right Answer Instead

Both models have a boundary, and recognising it prevents delay.

Any markedly abnormal result. ApoB well above target, blood pressure above threshold, a raised corrected calcium, a markedly raised ferritin with high transferrin saturation, pancytopenia, or a substantially raised glucose. These need clinical pathways, not interpretive reports.

Symptoms. New or progressive cognitive change, unexplained weight loss, persistent change in bowel habit, chest pain, breathlessness, any bleeding. Consumer services are not built for these and should route to care.

Where medication is indicated. Elevated apoB, hypertension and established atherosclerosis have outcome evidence for medication that no supplement approaches, and the years spent on an inadequate approach are not recoverable.

Anything requiring monitoring. Thyroid replacement, anticoagulation, glucose-lowering medication.

The useful framing: consumer services are good at the questions routine care does not address, principally measuring what is not routinely measured and deciding among broadly safe options. They are not a substitute for care where care is indicated, and a good service says so.

A Decision Sequence

First, establish whether the foundation is in place. Sleep regularity, resistance training, aerobic volume, protein, blood pressure measured properly. If not, that is the intervention and no purchase substitutes for it.

Second, consolidate what you already have. Request past records, gather scattered results, put the numbers in one place with dates and conditions. Frequently this reveals that measurement was never the problem.

Third, fill the specific gaps. ApoB, lipoprotein(a) once, fasting insulin, and a urine albumin to creatinine ratio are the usual omissions. This is where a testing service earns its cost.

Fourth, address the decision problem. Interaction checking, one change at a time, a three-month window, and a willingness to remove. This is where guidance earns its cost, and it is the part that compounds.

Fifth, keep the record. Whatever the tool, the sequence is the asset, and it is worth more in year ten than in year one.

Applied in that order, most people discover their gap is at step one or two, which is worth knowing before paying for step three.