Evidence standard

How we count: every figure says where it came from.

Every performance figure on this site carries one of four provenance tiers — KST field data, published research, independent laboratory characterization, or our own model. They carry different weight, so we never let one stand in for another. This page explains each, with a real example, so you can check any figure you read here.

The four tiers

Four sources, never blurred

Listed from the tier that is ours and measured, to the one that is ours and computed. Each has its own wording in our copy, so you can tell them apart in a sentence.

KST field data

Measured on a customer's site

Measured in a Kairospace deployment or pilot, against that site's own baseline or an untreated control.

How we word it

"we measured", "in our deployments", "field data".

−35.4% salt compaction, measured at Yorba Linda Country Club — one treated green, two untreated greens as controls, a roughly 3-month pilot.

One site, one green: a demonstrated result there, not an expected value elsewhere.

Published research

Other people's trials

Independent, peer-reviewed literature — other people's crops, soils, basins and ores. Never our own result, and never written as if it were.

How we word it

"in published research", "in peer-reviewed trials".

~20–25% less fertilizer at matched yield, in published research.

"At matched yield" is part of the figure: those trials held the harvest constant and measured how far the input could fall.

Independent lab

A third party measured our water

An outside laboratory characterized our water or our hardware, outside peer review. It is not our field measurement and not the literature, and it carries different weight from both.

How we word it

We say who measured it — an independent laboratory — and the method it used.

< 200 nm bubble size, characterized by an independent laboratory using nanoparticle tracking analysis, sized under ISO 20480.

ISO 20480 is the measurement framework. The < 200 nm is the laboratory's measured size class, not a threshold the standard defines.

KST model

Computed, never measured

Our own calculation on our own inputs. It is the weakest of the four and is never stated as a measured outcome.

How we word it

"our model, run on our own field data", "computed, not measured".

< 1 yr payback, from our model run on our own field data.

It varies by site, and we never attach it to a named customer, crop or season. Run it on your own numbers instead.

The house rules

The tiers say where a figure came from. These rules say how it is allowed to appear.

  • Ranges over points. Where the evidence is a range, the page states the range — ~20–25%, not 25%. Picking the top of a range and printing it alone is how the same metric ends up quoted four different ways across a site.
  • Units always. A figure travels with its unit — < 200 nm, mg/L, mm/day — so it cannot be read as something it is not.
  • "Up to" is part of the figure. It marks the top of a published range, not an expected value. It stays in a headline, an impact tile and a sentence alike; we do not drop it to make a tile read cleaner.
  • No tier borrowing. A laboratory result is never written as field data, published research is never written as if we measured it, and a model is never presented as a measurement. Where a figure rests on two tiers, the sentence names both.
  • A figure without a source comes off the page. Every number here traces to an entry in our claims register that records its source, its tier and the conditions that travel with it. If a number cannot be traced, it is removed, or rewritten as mechanism without a number.
  • Constants are not claims. Textbook values — the solubility of oxygen in water, a stoichiometric ratio, a figure derived from Stokes' law — take no tier. They name their own basis instead: "textbook solubility", "computed, not measured".

What "customer-reported" means

Some field figures come from the customer rather than from our instruments: the grower or operator measured or observed the change and told us. They sit in the field tier because they come from a real deployment, but nobody on our side took the reading. So we write "<customer> reports", never "we measured", and the evidence ledger marks the row customer-reported. A customer-reported figure also carries the customer's own metric — cuttings are not flower yield, and a growth rate is not a harvest.

Calculators built to say no

The ROI calculators run on the same discipline. Every assumption carries its source and tier, or is flagged where no supporting figure exists. They open on a conservative preset. The cannabis and aquaculture calculators credit extra output at margin rather than revenue, and cap it at what the equipment can physically reach. Two go further and are built to return a negative or zero answer when that is the honest one:

  • Wastewater costs both sides of the aeration question — your blowers after derating, against our concentrators plus the pump. On an already efficient fine-bubble plant, our side can be the dearer one, and the page says so.
  • Heap leach returns zero when your circuit already runs at its cyanide-to-oxygen optimum, and states that oxygen is not your limit rather than manufacturing a benefit.

A calculator result is an estimate from inputs you choose, not a measured outcome — which is why it belongs to the model tier and never to the field tier.

Check any figure yourself

The evidence ledger on the homepage lists the headline figures with their source and tier side by side. The case studies give the conditions behind our field data — the site, the control, the duration. If you find a figure on this site that does not follow these rules, tell us and we will correct it.

Your numbers

The only figure that settles it is the one from your site.

Run the model on your own inputs, or measure the delta on your own water with a structured pilot.

Run my numbers Start with a pilot