What to Add to Your Water: Every Category, Evidence-Graded

Electrolytes, greens powders, magnesium, collagen. Four categories, four very different evidence bases. And one piece of label arithmetic that tells you whether any of them contains a real dose.

Most "what to add to your water" articles are shopping lists in disguise. Here's a different approach: take the four categories people actually buy, and grade each on what the published research supports, then hand you the arithmetic to check any label yourself.

Because the more useful skill isn't knowing whether greens powders work. It's being able to look at a label and tell whether the ingredient with the good research behind it is present in the dose that research used.

Usually it isn't.

The Grades

Bottom line: collagen has the strongest evidence of the four, electrolytes are genuinely useful but only situationally, magnesium rests on evidence its own reviewers called substandard, and greens powders have improved from nothing to thin-and-company-funded.

Electrolytes

Genuinely useful when you're actually losing fluid and sodium. Not a daily requirement for a person sitting at a desk.

This is the one category where the mechanism is uncontroversial: sweat carries sodium out, and replacing water without replacing sodium leaves you diluted rather than restored. That's real physiology, and it matters during heavy sweating, heat exposure, prolonged exercise, and illness involving vomiting or diarrhoea.

What it isn't is a daily baseline. If your day involves an office and a normal amount of walking, you're replacing what you lose through food.

The site covers this in depth elsewhere rather than repeating it here: how to read an electrolyte label for spotting a vague blend, and the format comparison for cost and concentration trade-offs across powders, tablets, capsules and ready-to-drink.

One honest caveat that runs against the category's marketing: our own review of the hangover evidence found that replacing fluid didn't measurably change hangover severity, and the cold-plunge research found volume mattered more than what was dissolved in it. Electrolytes are real. The claims stacked on top of them often aren't.

Greens Powders

The evidence has genuinely improved since the "no real studies at all" era. And it's still thin, short, small, and funded by the company selling the product.

AG1 is the category's reference point, so it's the fair test case.

What the criticism was

McGill University's Office for Science and Society reviewed the company's own early study and did not mince words:

"There was no control group; the outcome was evaluated via subjective self-reports; and only 35 adults participated."

Their summary of what that demonstrates: "people reported feeling better after drinking a green liquid they were told would make them feel better."

McGill also flagged the label itself: 467% of the daily recommended value of vitamin C and 1,100% of vitamin B7, both water-soluble vitamins where, as they put it, you largely "pee it out." And of the roughly 75 ingredients, the adaptogens are "whole plants and fungi, mixtures of molecules whose concentrations vary based on seasons and soil conditions, and whose final amounts in AG1 are undisclosed."

What's changed

There are now actual randomized trials, which is a real improvement and worth saying plainly. A 2026 randomized, double-blind, placebo-controlled crossover trial in Frontiers in Nutrition found AG1 "significantly increased the total number of EARs met compared to placebo [2.8; p = 0.0011]" and selectively enriched some beneficial gut bacteria.

Read the design before the result, though:

• 20 participants, resistance-trained adults
• 14 days per condition
• Funded by AG1, with multiple authors employed by AG1

And consider what the headline finding actually says. "Increased the number of estimated average requirements met" means: a product containing vitamins increased measured vitamin intake. That is what a multivitamin does by definition. It is not evidence of an energy, immunity, or wellbeing benefit: the things the category is actually marketed on.

The honest grade: greens powders are an expensive multivitamin with a produce-flavoured story attached. If your diet has genuine gaps, a conventional multivitamin closes them for a fraction of the price. No product recommendation in this section. Recommending a competing greens powder would undercut the section's own reasoning.

Magnesium

A real but small effect on falling asleep, resting on evidence the reviewers themselves described as substandard.

A systematic review and meta-analysis of oral magnesium for insomnia in older adults pooled three randomized trials covering 151 people across three countries.

Seventeen minutes to sleep onset is a genuine, noticeable difference if you lie awake. But the reviewers' assessment of the evidence behind it is the part that rarely gets quoted: all trials were at moderate-to-high risk of bias, and outcomes were supported by low to very low quality evidence. Their conclusion was that the literature is substandard for physicians to make well-informed recommendations.

So the accurate statement is: magnesium may modestly shorten time to fall asleep, on three small trials that the field's own quality assessment rates poorly. That's a reason to try it cheaply if you're curious, not a reason to build a routine around it.

Worth noting separately: magnesium is one of the supplements where the form matters for tolerability, and where kidney function is a genuine contraindication. That's a pharmacist conversation, not a label-reading exercise.

Collagen

The best-evidenced category here. And the one people are most sceptical of.

The reflexive objection is that collagen gets digested into amino acids like any protein, so swallowing it shouldn't do anything specific. It's a reasonable objection. The trial data doesn't support it.

A systematic review and meta-analysis of 26 randomized controlled trials in 1,721 participants found hydrolyzed collagen supplementation significantly improved both skin hydration and elasticity compared with placebo.

On the joint and training side, a separate meta-analysis of 19 studies in 768 participants found significant effects for long-term collagen peptide intake combined with training on fat-free mass, tendon morphology, muscle architecture, maximal strength, and recovery from exercise-induced muscle damage.

Twenty-six trials and 1,721 people is a substantially larger evidence base than anything else on this page. Caveats still apply: industry funding is common across this literature, and effect sizes for skin outcomes are measurable rather than dramatic. But on volume and consistency of evidence, collagen clears the bar the other categories don't.

The Label Math That Works on Anything

This is the transferable part. Three steps, and it works on any supplement label, in any category.

The problem it solves: an ingredient can have genuine research behind it and still be present in a product at a dose far below what that research used. Proprietary blends exist to make this hard to check.

1. Find the study dose. Look up what the research actually used: grams per day, not "contains ashwagandha."
2. Find the label dose. If the label gives a specific milligram amount for that ingredient, you can compare directly. If it's inside a proprietary blend, you can't. And that's the finding.
3. Check the blend's total weight. A blend listing 20 ingredients at a combined 2 grams cannot contain a clinical dose of any of them. Ingredients are listed in descending order by weight, so anything below the midpoint is present in trace amounts.

Applied to the greens category: a blend of 75 ingredients within a single scoop is arithmetically incapable of delivering research-grade doses of most of them. That's not a claim about any specific brand. It's what the total weight permits.

The one-sentence version: if a label won't tell you the dose, assume it isn't the study dose. Companies disclose numbers they're proud of.

Same skill, different aisle: reading an electrolyte label applies exactly this logic to sodium, potassium and magnesium amounts, and the certified-versus-tested distinction on water filters is the same instinct applied to certification claims.

FAQ

Are greens powders worth it? The evidence has improved but remains thin and company-generated. A 2026 randomized placebo-controlled crossover trial of AG1 found it increased the number of nutrient requirements participants met. But it ran 14 days in 20 people, was funded by AG1, and several authors were AG1 employees. Increasing measured vitamin intake is also close to tautological for a product containing vitamins; it isn't a health outcome.

Does magnesium actually help you sleep? Weakly, on weak evidence. A meta-analysis of three trials in 151 older adults found magnesium reduced time to fall asleep by about 17 minutes, but total sleep time did not improve significantly. And the authors judged all trials at moderate-to-high risk of bias with low to very low quality evidence, calling the literature substandard for making recommendations.

Does collagen supplementation actually do anything? It has the strongest evidence base of the common water add-ins. A meta-analysis of 26 randomized controlled trials in 1,721 participants found hydrolyzed collagen significantly improved skin hydration and elasticity versus placebo. Joint and training outcomes have a smaller but real evidence base.

What does a proprietary blend hide on a supplement label? The individual doses. A proprietary blend lists a combined weight for several ingredients without saying how much of each is present, so an ingredient with genuine research behind it may be included far below the dose that research used. AG1 lists exact amounts for a minority of its ingredients and covers the rest this way.

Do I need to add anything to my water at all? For most people on most days, no. Electrolytes earn their place when you're losing fluid through heavy sweating, heat, or illness. The other categories are supplements that happen to dissolve in water. Judge them as supplements on their own evidence, not as hydration products.

Keep Reading

• How to Read an Electrolyte Label: the same dose-checking skill applied to electrolyte products.
• Electrolyte Format Comparison: cost and concentration across formats, if electrolytes are the category you actually need.
• Do Electrolyte Drinks Actually Help With Hangovers?: a worked example of a popular claim that doesn't survive its own evidence.

A Note on Scope

This article is for informational purposes only and isn't medical advice. Supplements can interact with medication and are not appropriate for everyone. Magnesium in particular needs care if you have kidney disease, and high-dose single nutrients can interfere with prescriptions. Talk to a clinician or pharmacist before adding a supplement, especially if you're pregnant, breastfeeding, or managing a chronic condition. Supplements are not held to the pre-market approval standard that medicines are.

Sources

• McGill University Office for Science and Society: "You Probably Don't Need that Green AG1 Smoothie": the 467% vitamin C and 1,100% vitamin B7 figures, the undisclosed adaptogen amounts, and the assessment of the company's 35-participant uncontrolled study.
• Gonzalez DE, et al. "Effect of AG1® supplementation on nutritional adequacy and gut microbial composition in trained adults." Frontiers in Nutrition, 2026: the randomized, double-blind, placebo-controlled crossover design, the 20-participant sample, the EAR result, and the declared AG1 funding and author employment.
• Mah J, Pitre T. "Oral magnesium supplementation for insomnia in older adults: a Systematic Review & Meta-Analysis." BMC Complementary Medicine and Therapies: the three-trial, 151-participant pooled analysis, the sleep-onset and total-sleep-time figures, and the risk-of-bias and GRADE quality assessments.
• Exploring the Impact of Hydrolyzed Collagen Oral Supplementation on Skin Rejuvenation: A Systematic Review and Meta-Analysis: 26 randomized controlled trials, 1,721 participants, skin hydration and elasticity outcomes.
• Impact of Collagen Peptide Supplementation in Combination with Long-Term Physical Training: A Systematic Review with Meta-analysis: 19 studies, 768 participants, strength, tendon and recovery outcomes.

Not used: a widely-circulated figure describing the collagen evidence base as roughly 113 trials in around 8,000 people. No published review at that scale was found; the meta-analyses cited above are the largest located, and their actual participant totals are used instead. Also excluded: claims about AG1's litigation status, which could not be established from a primary court or regulatory record.