Station 5 · Prediction Engine

The right call is not
a matter of effort.
It's a matter of timing.

For each customer, boostpilot.ai tells you not just whether they are worth it, but what to do now: defend, develop or win. And when.

Timing and course — the right moment in the cockpit.
Three answers, depending on the situation

Not every customer
needs the same thing.

Where a customer sits in the potential matrix, and what their signals say, decides the action.

Defend

A good customer is about to drift away. The patterns show it before revenue drops. Act before it is too late.

Develop

The signals show current buying intent. This customer is worth working harder now.

Win

A window opens at a company that is not a customer yet. You get in now, or not at all.

The special case

In machinery, who gets to supply is decided early.

When a manufacturer specifies its next machine generation, it fixes which suppliers and parts go in. That is your window. Once your part is in the drawing, it stays for the whole series.

Trigger / window
First contact
Evaluation
Design-inthis is where the OEM decides
Pre-series
Series

Sold at the design-in, not at the purchase.

Where the timing comes from

Not guessed.
Calculated.

The Prediction Engine reads the last design-in from the order history, knows the rhythm in which a customer develops new generations, and moves the next contact window forward from it, usually six months before to three after. Public triggers confirm or reject the forecast.

Example · running pilot, anonymised

388
existing customers
9,682
order lines
56
customers with a dated window, open today — 48 of them no one sees today
Why the forecast holds

Three levels,
in this order.

Tier A

Evidence at the customer

Concrete evidence at the customer itself beats everything.

Tier B

The customer's own rhythm

Without A, the customer's own rhythm counts, measured against their history, not guessed.

Tier C

Industry expectation

Only when neither yields anything does the industry expectation come in, explicitly as an assumption.

And where no window can be derived, it says "no window derivable", not a convenient number.

Two speeds

Some windows come every two years.
Some once a decade.

Fast segments · 1–6 years

Continuous presence

Semiconductors, battery, dispensing: the window comes round within a few years. Mode: short response times, trigger-driven.

Slow segments · 7–15 years

Early, rare, aimed high

Pharma, lab automation, materials testing: the window comes once a decade. Miss it and you wait a generation. At head-of-engineering level.

The triggers

25 signals,
three response levels.

From order patterns to tenders, patents, plant construction and trade-fair appearances: 25 signals report movement, with lead times of 9 to 48 months before start of production. Red means call this week. Orange means into the four-week plan. No signal means cadence by class.

The full trigger and evidence system in the demo →
Demo

Which of your customers are in the window right now?

With your order history we show where a window is open and what comes next.

Book a demo