Photo via Unsplash
The climatological peak of the Atlantic hurricane season falls around 10 September. That is two days away, which means any board reading this is reviewing its continuity plan during the exposure rather than ahead of it. So this is written to be actioned in a week, not a quarter.
NOAA's outlook for 2026 puts the probability of a below-normal season at 55 percent and an above-normal season at just 10 percent, with eight to 14 named storms and one to three major hurricanes expected. Acrisure Re projects around 13 named storms and roughly four landfalls. Caribbean landfall probability this year sits at 35 percent, below the 47 percent historical average. A warm Atlantic and an expected return of El Nino during the peak months are pulling in opposite directions.
I have watched what a forecast like that does in a boardroom, and it is the reason I am writing this now rather than in June. A below-normal outlook produces a very specific kind of relaxation, and it is built on a misreading. A seasonal count tells you how many storms will form across an entire basin. It tells you nothing about where any of them go or how strong they are when they arrive. For any one island, the year is decided by a single storm.
Melissa formed near the end of October 2025, in the tail of a season, and made landfall on Jamaica's southern coast near peak Category 5 strength. At least 93 people died. Physical damage across Jamaica has been estimated at 12.2 billion dollars. No firm on that coast was helped by the seasonal count.
Why the plans in the drawer do not work
Most BCPs I have reviewed for Caribbean firms fail in the same three ways, and none of the three is about effort.
The first is age. A plan last updated in 2019 or earlier was built around a Category 3 scenario, because that was the reasonable planning case then. Melissa and the seasons around it have moved the baseline. A Category 5 direct hit is now the case a plan has to survive, and treating it as an outlier is a decision, whether or not anyone made it deliberately.
The second is language. Too many of these documents read as though they were downloaded and lightly edited. A phrase like "ensure business continuity through robust protocols" tells nobody anything at 6am when the roads are flooded and the mobile network is down. The test for every sentence in a continuity plan is whether a person who has not read the rest of it could act on that sentence alone, in the dark, without internet. Most sentences fail.
The third gap is missing operational detail, and it is the expensive one. What are the first three calls the chief executive makes, by name. What are the payment terms if you need to pre-position fuel, agreed with whom, in advance. Who can authorize spending without a board resolution, and up to what limit. These questions are tedious to answer, which is why they get deferred to the moment when there is no time left to answer them.
What the Season Actually Says
- ~10 SeptemberClimatological peak of the Atlantic hurricane season
- 55% / 10%NOAA probability of a below-normal season, against an above-normal one
- 8 to 14Named storms forecast, with one to three major hurricanes
- 35%Caribbean landfall probability this season, against a 47% historical average
- $12.2BEstimated physical damage in Jamaica from Hurricane Melissa, with at least 93 lives lost
Three things worth doing before the next storm
Run the scenario, per business unit, as a Category 5 direct hit in the peak week. Every lesser case is contained inside it, so a plan that survives it survives the others. Model three things specifically: the disruption window, the point at which cash gets tight, and the decisions that have to be made in the first 48 hours. That last one is where most plans turn out to be silent.
Rank your critical dependencies by how fast each has to come back, not by how important each feels. For most Caribbean firms the top three are power, telecommunications and access to cash, and those three decide whether you can operate at all. Everything else matters and can wait, and saying so out loud is the useful part. A ranked list of three that people argued over beats an unranked list of thirty that nobody has read.
Then write the first three calls down. Names, mobile numbers, backup contacts, and the script for each: what you are asking that person for, and who you call instead when the number does not connect. One page. Laminate it. A plan gets used in a hurricane only if it was designed to be used in one, and a 40-page PDF on a server behind a login will not be.
Two things that have changed since 2019, in opposite directions
Insurance has moved against Caribbean firms. Hurricane cover is more expensive and narrower than it was, and catastrophe pricing is built on multi-year models rather than on this season's forecast, so a quiet outlook will not reduce your premium. That shifts the arithmetic on self-insurance and captive structures, and it makes physical hardening a cost saving rather than a precaution.
Telecoms have moved in your favour. Fibre and mobile network hardening across the region have made remote operation genuinely viable during the recovery window in a way it was not seven years ago. That changes the geography of the plan: where key staff need to be before the storm, and which functions can be run from outside the affected area. Several firms I work with have restructured their continuity staffing on this alone.
Where I would push back on my own advice
The Category 5 rule is the right default and it has a cost I should name. Planning every unit against the worst case produces plans that are expensive to maintain and that some managers stop believing in, and a plan nobody believes gets ignored under stress exactly like no plan at all. If your firm has limited capacity to do this properly, plan two business units honestly against the severe case rather than eight superficially against all of them.
The one-page call sheet has no such trade-off. It costs an afternoon, it works when everything else has failed, and I have never seen a firm regret having it. If you only do one thing this week, do that.
Before your next executive meeting
Pull out the current plan and check the date on it. If the last substantive update was before 2024, the document describes a threat environment that no longer exists, and everyone in the room already suspects that. Put the Category 5 scenario for your largest business unit on the agenda, name who owns the rebuild, and give them a deadline inside this month rather than this quarter.
Every board I advise has learned the same thing across the last two seasons. The plans that get used are the ones that were rebuilt for the storms we are actually getting. The ones inherited from a calmer decade get opened once, during the event, and closed again.
Rebuild before the next storm rather than during it.
"A below-normal season is not a below-normal risk. Melissa arrived at the end of October and did 12.2 billion dollars of damage to one island. The forecast was never the thing that was going to protect you." - Adrian Dunkley, AI Boss
Frequently Asked Questions
When does the Atlantic hurricane season peak?
The climatological peak of the Atlantic season falls around 10 September, with August through early October carrying the bulk of activity. That timing is why a board reviewing its business continuity plan in September is reviewing it during the exposure rather than ahead of it, and why the review has to be short enough to finish in a week.
What is the 2026 Atlantic hurricane season forecast?
NOAA's outlook puts the probability of a below-normal season at 55 percent and an above-normal season at 10 percent, calling for eight to 14 named storms, three to seven hurricanes and one to three major hurricanes. Acrisure Re has projected around 13 named storms and roughly four landfalls, with Caribbean landfall probability at 35 percent against a 47 percent historical average. A warm Atlantic and an expected return of El Nino during the peak months are the two factors shaping the outlook.
Does a below-normal forecast mean lower risk for a Caribbean business?
No. Seasonal counts describe how many storms form, not where they go or how strong they are at landfall, and a single storm determines the year for any one country. Hurricane Melissa made landfall on Jamaica's southern coast near peak Category 5 strength in late October 2025, killed at least 93 people and caused an estimated 12.2 billion dollars in physical damage. A quiet season with one direct hit is not a quiet year for the island that was hit, and a forecast of fewer storms is the condition under which boards most reliably relax.
Why do business continuity plans written before 2024 fail?
Three reasons recur. They were built around a Category 3 scenario, when recent seasons have made a Category 5 direct hit the case a plan has to survive. They are written in template language such as ensuring continuity through robust protocols, which tells nobody what to do at 6am with flooded roads and a mobile network down. And they omit operational detail: who the chief executive calls first, what the payment terms are for pre-positioning fuel, who has the authority to spend without a board resolution.
What scenario should a Caribbean board actually model?
A Category 5 direct hit in the peak week, run for every business unit separately. It is the case that tests operations hardest, and every lesser scenario is contained inside it. Model the disruption window, the point at which cash runs short, and the decisions that have to be made in the first 48 hours. If the plan works for that case it works for the others, and if it only works for a Category 3 you have not tested it.
How should a firm rank its critical dependencies?
By how quickly each one has to come back before the business stops functioning, rather than by how important it feels. For most Caribbean firms the top three are power, telecommunications and access to cash, and those three determine whether you can operate at all. Everything else matters and can wait. A ranked list of three that people have argued over is more useful than an unranked list of thirty that nobody has read.
What belongs on the one-page call sheet?
Actual names, actual mobile numbers, actual backup contacts, and the script for each call, on a single laminated page. Not who to call in an emergency as a category, but the person, the number, what you are asking them for, and who to call instead when that number does not connect. The test of the page is whether a manager who has never read the plan can act from it in the dark with no internet.
What has changed most for Caribbean businesses since 2019?
Insurance and telecoms, in opposite directions. Hurricane cover has become more expensive and narrower, and because insurance pricing is built on multi-year catastrophe models rather than the current season's forecast, a quiet outlook does not lower the premium. That shift strengthens the case for physical hardening as a cost saving rather than a precaution. Telecoms have improved, with fibre and mobile network hardening making remote operation more viable during the recovery window, which changes where key staff need to be positioned before a storm.