An exposure figure is worth what its documents are worth.
A number assembled from memory gets through the first meeting. A number where each line names the document underneath it gets through the third, and reads the same way to anybody who picks it up later.
Treatable volume, depth, exchange with adjacent water, organic load and the length of your exposed window each move the cost of delivery by a large factor. The first work is on your side, before anybody quotes.
A figure has two audiences and they read it differently. Your board wants to know whether the loss is real. Whoever prices the transferred part of it a year from now wants to know which document each line came off, who signed that document, and over how many observed years it was collected. Build for the second reader and the first is covered on the way past.
Run this before a site read, and the conversation turns from technology to numbers. A competent analyst with maintenance records and revenue data needs about a fortnight. It does not require us.
Six steps
- Define the window your losses sit in Mark the weeks where the money goes. For most coastal assets that is a run of weeks inside one season; for some it is a single fortnight. Annualizing a seasonal risk is the commonest way to make a real exposure look like background noise. The window marks where the loss lands; step six covers the months either side.
- Count the revenue inside the window The revenue in those specific weeks, which for a resort, a marina or a fishery is usually a wildly disproportionate share of the year. Annual revenue divided by fifty-two hides it.
- Write each frequency as a count over the years you observed Throttled intake, mortality event, closed frontage, early membrane replacement, regulator notification. Give each one a rate, and carry the observation count with it wherever the rate travels. Written in the form four events across twelve years of continuous logging, a rate says what it is made of. Where the record only runs three years, say three, and let the thinness be visible. A rate with its denominator attached can be argued with. A rate without one gets discounted by whoever reads it next, usually silently, and usually by more than you would have conceded.
- Apply a multiplier to the direct loss, and justify it Published event studies consistently put total regional income loss at several times the direct figure. Pick your multiplier from a documented comparable, record which one, and let the committee take its argument to the source.
- Subtract what you already spend Additional dosing, extra cleaning cycles, contingency crew, monitoring added after the last event, the insurance loading you now carry. That money already leaves the business because of water.
- Compare against a year of the standing program Set the annual cost of a subscription holding that water against the expected cost of the window failing. Annual, because fouling load, nutrient and heat accumulate through the months either side of your exposed weeks, and continuous work keeps the peak below the level that costs production. An emergency quotation flatters the wrong option twice: a reactive force mobilized after the water has gone over costs more and arrives with the loss already booked.
What each line is worth
A line is worth the document under it. A maintenance order with a date on it, a mortality record with the oxygen trace beside it, an occupancy report, a notification letter: each of those can be handed to somebody who was not in the room. An estimate from the plant manager's memory cannot, however good the plant manager is.
So build the table in two passes. The number first. Then the document, named, with its date and whoever or whatever produced it. Where no document exists the line stays blank and gets shown blank, because a blank line with a reason attached reads better in a review than a filled one with a shrug behind it.
| Line | Where the number lives | The mistake that weakens it |
|---|---|---|
| Lost production | Production against nameplate, hour by hour, for the affected weeks in each of the last ten years | Using average annual availability, which buries a two-week collapse inside a good year |
| Energy penalty | Specific energy consumption during affected periods against clean-water baseline | Treating it as a tariff issue because it appeared on an energy invoice |
| Consumables and asset life | Membrane, element and media replacement dates against design life; cleaning-in-place frequency | Booking early replacement to routine maintenance, which erases the cause permanently |
| Stock loss | Mortality records with the dissolved-oxygen and temperature trace alongside them | Attributing the whole event to disease when the oxygen trace shows what preceded it |
| Closure and cancellation | Occupancy, cancellations, discounting and forward bookings for each season after an event, until back on trend | Stopping at the closure week and ignoring the soft demand behind it |
| Compliance and disclosure | Notifications made, consent variations sought, legal and consultancy hours, insurance loading at renewal | Excluding it as non-operational, though it compounds fastest |
| Mitigation already running | Additional dosing, contingency crew, added monitoring, standby arrangements introduced after prior events | Leaving it out, which understates both the exposure and the value of removing it |
The same folder does two jobs. Assembled once, it answers the internal question of whether an exposure is real, and it is the material anybody assessing the transferred part of that exposure asks for a year later. Duration per event, count of events over observed years, and what was done between them, each with the document behind it. Mitigation carried in that folder counts when an independent party measured it; carried as a description of intent, it does not.
The evidence that exists for in-water treatment, and the parts of it that are still missing, are set out under what the record shows.
Sanity checks
What the published record says about order of magnitude.
Analysis of the global Harmful Algae Event Database attributes more than US$8 billion of economic loss from 2000 to 2020 to fish-killing blooms alone, with the largest documented impacts in China. One stressor, one loss type, two decades.
At the single-asset end there is no published benchmark worth quoting. The desalination downtime cost in circulation traces back to industry commentary, with no operator filing, regulator record or reviewed study beneath it. Your own throughput, tariff and replacement-water cost give a better number in an afternoon.
The stressor breakdown gives sector comparables: aquaculture, shellfish, desalination and tourism each fail differently and at different scales.
Two patterns recur in the situation reports Alarivean receives, and neither has a published statistic behind it, so no figure is quoted for either. Operators who stop annualizing land on an exposed window measured in weeks. A guest-facing closure leaves forward bookings soft for seasons after the water has cleared. Both belong in your model.
Two ways to buy
Per incident, or on subscription.
An incident-priced service earns more when your water behaves worse, and that incentive outlasts the people who signed for it.
A subscription inverts it. The commitment is an agreed condition inside a defined zone, with vessels, gases, crew, instrumentation and reporting on our side of the line. Proactive work outside the line, intercepting a bloom or a deficit where it is building, sits inside the same commitment with no separate invoice.
For finance, one is a variable cost correlated with your worst outcomes and the other a fixed cost that reduces them. They sit differently in a plan and with an insurer.
Two clauses belong in the written terms. The subscription ends if performance falls outside the agreed spec, on the reading itself. And once the minimum period has been served it ends whenever you are not satisfied, on no stated grounds.
The calibration phase decides everything: a bounded demonstration area on your own water, with viability parameters agreed before it starts, so both sides settled in advance how to read the result.
How Alarivean's own figures are bounded
A figure for your water arrives after the site read. Pond-scale and ballast-tank-scale validations are evidence about the approach; no open-water result has been published for it by anyone. Any percentage improvement in dissolved oxygen, bloom biomass or availability waits for a calibration phase to produce it.
Sources
- Reviews in Aquaculture — analysis of the Harmful Algae Event Database, 2000–2020, 2024.
- Alarivean, Inc. — resilience programs, controls and data services.
Asked by finance
Four commercial questions
What decides the price, and when do we get one?
Treatable volume, depth, exchange with adjacent water, organic load, window length, and distance from existing regional capacity. Each moves the answer materially, and none can be inferred from a web form.
The figure for your asset comes after the site read, in writing, with its assumptions printed underneath.
Our exposure is one week a year. Is this relevant to us?
Possibly not. A single short window with modest losses is usually better covered by contingency and redundancy than by a standing program. Reaching that conclusion early costs nothing.
Can we build the case on published loss figures from comparable operators?
Use them to test the order of magnitude of your own number. Do not use them as the number.
Published event losses are aggregated across regions, sectors and methodologies you cannot inspect. A case built on your own maintenance records and revenue data survives scrutiny.
What happens if calibration shows the treatment does not work here?
The agreed parameters say so and the program stops. Finding out inside a bounded demonstration area costs far less than finding out inside a service contract. Agree the exit before the entry.
When the number is built
What comes back from a site read.
Send the asset, the failure mode, the exposed weeks and the unit cost of an hour down. Alarivean returns a technical read on treatable volume and realistic effect, including the case where the arithmetic says to leave it alone.