Stop Guessing the Peak: In an Uncertain Era, Companies Need to Write Their Turn Rules First

Forecasts are useful. They provide a baseline scenario, let finance and operations work from the same set of assumptions, and remind a company not to bet all its resources on the most optimistic path.

But a forecast does not automatically produce an action. A company can forecast 20% demand growth next year without ever defining what order volume triggers a delay in equipment purchases; it can forecast a stronger Taiwan dollar without ever deciding how often to update quotes, or what margin level triggers hedging; it can worry that rates will stay high without ever sorting out which loan comes due first.

When the forecast and reality drift apart, management often tells every department to "stay flexible." That sounds prudent, but it actually hands responsibility to whichever future turns out to carry the most pressure. A trigger rule works the other way: it requires the decision to be made today — what evidence, once it appears, gives whom the authority to do what.

This is also exactly the management shift Taiwanese companies need most. Taiwan is riding the fast lane of AI demand, but that also makes it more exposed to swings in customer capital spending, energy costs and financial valuations. High growth does not cancel out risk — it raises the cost of turning late.

Many companies' idea of scenario analysis is simply swapping in an optimistic, neutral and pessimistic revenue growth number. Once the spreadsheet runs, the report declares the stress test done.

A genuinely useful scenario has to describe how the world actually works: why demand rises or falls, how prices and payment terms change, whether suppliers start slipping, whether banks tighten credit, what competitors do, and in what order the changes reach the company.

An AI downturn scenario isn't just lower revenue. It might first show up as customers delaying acceptance of delivered goods, then accounts-receivable days lengthening, inventory rising, prices getting discounted — and only then finally showing up on the top line. An energy shock isn't just higher costs either; it can transmit simultaneously through exchange rates, freight, interest rates and customer demand.

So the point of a scenario is not to describe the future accurately — it's to find the earliest observable signal on each path. The earlier a company sees the signal, the sooner it can take a small, reversible action, instead of waiting until the P&L turns red and cutting everything at once.

The "trigger card" proposed in this piece is an editorial method, not any official agency's standard. It needs at least seven fields.

Indicator: what are you watching? New orders, quote-to-order conversion rate, days of inventory, accounts receivable, exchange rates, interest rates, or supplier lead times.

Threshold: at what level has the situation actually changed? "Keeping a close eye on it" is not a threshold.

Duration: a single day, a single month, or two consecutive quarters? Without a time window, an organization is easily jerked around by noise.

Action: what exactly happens once triggered? Cut procurement, adjust pricing, activate a second supplier, increase hedging, or delay capital spending.

Owner and deadline: who declares the trigger? Who executes it? Within how many days? Without an owner and a deadline, the rule is still just a reminder.

Withdrawal condition: what evidence brings things back to normal? A rule with only an "on" switch and no "off" switch turns a temporary measure into a permanent cost.

The value of these seven fields isn't tidy formatting — it's turning "everyone sort of already knows" into a commitment someone can be held accountable for.

The first is the demand card. Don't just watch revenue, because revenue is a lagging indicator. Track new orders, quote-to-order conversion, customer forecasts, and the cancellation-and-delay rate; if two or more indicators weaken for two consecutive months, freeze any capacity expansion that can be delayed, and have sales confirm whether it's a single customer adjusting or the whole market shifting.

The second is the cost card. Break out how raw materials, energy, freight, exchange rates and wages each affect margin. The trigger doesn't have to be a specific price — it can be "unpassed-through cost pushes unit margin below a safe floor." The action might be repricing, changing purchase lot sizes, switching to an alternative material, or adjusting the product mix.

The third is the cash card. Track months of cash runway, accounts-receivable days, inventory days, available credit lines, and debt coming due over the next twelve months. If customers start paying late, inventory rises, and bank credit tightens all at the same time, the countdown on cash has already started even before the P&L turns red.

The three cards aren't meant to help a company predict the world — they're meant to translate the world's changes into actions the company can actually take. Every action should be tagged with whether it's reversible, what it costs, and what condition allows it to be withdrawn.

Here is an illustrative decision, not pointing to any specific company: a mid-sized Taiwanese exporter fulfilling AI equipment orders sees a full-year demand forecast that still looks great, but at the same time notices two customers delaying acceptance, receivables stretching out by a month, and inventory days climbing. If it only looks at total revenue, it will keep buying equipment and hiring ahead of need; if it has the three cards, the demand card freezes unsigned expansion first, the cash card reschedules debts coming due, and the cost card tells sales to renegotiate delivery times and payment terms.

The point of this illustration isn't to guess whether AI demand will cool — it's to separate "demand is still there" from "cash can't take it." The company can keep serving confirmed orders while preserving room to pull back, which is closer to real risk management than waiting for a market-wide pessimistic consensus to form and then cutting capacity all at once.

A trigger rule is not autopilot. Set the threshold too sensitively, and a single month's order decline could trigger unnecessary layoffs; reprice across the board after one sharp one-day jump in energy prices, and you might lose the customer instead; an exchange rate crossing a round number doesn't necessarily mean the long-term trend has actually changed.

So the rules need at least three brakes built in.

First, confirm with multiple indicators. A demand decline is best confirmed by orders, quotes and inventory together, not revenue alone.

Second, set a duration requirement. Unless it involves a security, liquidity or safety incident, don't overhaul strategy because of a single day's swing.

Third, take the small reversible action first. Shorten procurement commitments, increase monitoring frequency, and confirm credit lines with the bank before deciding whether to cancel investment or cut capacity.

This is also why the trigger card can't be written by finance alone. Sales knows whether a customer's pullback is real or not; operations knows whether lead times can actually be substituted; HR knows whether an adjustment will damage a critical capability. The rule has to be executable across departments, or it just becomes one more table nobody looks at.

The exchange rate is the easiest thing for a company to outsource to prophecy. A company asks what the year-end rate will be, gets a number, and ties its entire year's pricing and budget to it; when the rate moves away from that number, it decides at the last minute whether to hedge.

A better approach is to calculate exposure first: for every 1% of appreciation, how much margin is lost? How much can foreign-currency revenue and raw-material costs offset each other? How large are the signed-but-unpaid orders? Will customers accept an exchange-rate adjustment clause?

Then set tiered triggers: entering the first band, extend the validity period on new quotes; entering the second band, add partial hedging; entering the third band, re-examine low-margin products and payment terms. If the rate returns to an acceptable range and order margins recover, withdraw the measures step by step.

This approach will not eliminate exchange-rate risk, and it is not investment advice. It simply turns "I think the Taiwan dollar is going to strengthen" into "at what margin loss does the company need to change its pricing and hedging."

A threshold that worked in 2025 will not necessarily fit 2026. Product margins, customer mix, the cost of capital and the supply chain all change; if a company adds a second market, its customer-concentration threshold should be adjusted; if it switches to a different material, the cost trigger needs to be recalculated; if a bank cuts its credit line, the cash safety floor should be raised.

Every card needs a review date and a data owner. After a major event, the company should look back: did the alarm go off too late? Were there indicators that shifted first but got ignored? Did the action actually reduce the loss? Was the withdrawal too slow?

The BIS's discussion of financial-system resilience notes that under stress, markets need not just assets but a mechanism that can absorb shocks, sustain function, and reconfigure. ^[4] Applying that principle to a company means letting the rules themselves be stress-tested too, instead of turning a single meeting's conclusion into something permanent.

In an uncertain era, the most expensive mistake isn't necessarily guessing the wrong direction — it's failing to have a mechanism that admits the direction has changed. The equipment has already been ordered, the budget has already been committed, the executive has already said it publicly — so the organization keeps waiting for the original forecast to come true.

What a trigger condition offers is a different kind of certainty: we don't know when AI investment will slow, but we know which order and cash indicators worsening means it's time to pause expansion; we don't know where the exchange rate will land by year end, but we know what margin level means it's time to adjust pricing; we don't know when interest rates will fall, but we know which debt needs to be refinanced ahead of time.

Forecasting still has to happen. It's just that behind every forecast, there should be one more sentence that matters more:

If it doesn't turn out this way, what do we see — and what do we do instead?