Foresight
Forecasts demand for every flight across a 363-day booking horizon.
Three engines work from the same demand forecast
Foresight forecasts demand, Optima prices against that forecast, and Radar checks the result against competitor fares. Each example below is an output of the pilot test pack, shown with its basis.
Foresight forecasts the booking curve of each flight
For every departure, Foresight projects how many seats will be booked on each day until the aircraft leaves, with a range. It compares bookings to date with that projection and marks the flight as ahead, on pace or behind.
The forecast covers a 363-day booking horizon.
Booked to dateForecastForecast range
Table view
| Days before departure | Booked | Forecast | Range |
|---|---|---|---|
| 363 | 0.4 | 0.4 | 0.3 to 0.4 |
| 270 | 0.5 | 0.4 | 0.3 to 0.5 |
| 180 | 2.7 | 1.6 | 1.3 to 1.8 |
| 120 | 14.1 | 8.3 | 6.9 to 9.6 |
| 90 | 28.1 | 18.1 | 15.1 to 21.1 |
| 60 | 44.7 | 33.2 | 27.7 to 38.7 |
| 55 | 47.3 | 35.9 | 29.9 to 41.8 |
| 30 | not yet | 48.3 | 40.3 to 56.3 |
| 0 | not yet | 58.1 | 48.5 to 67.7 |
Optima sets one price and files it inside the fare classes you already have
Optima computes a continuous price for each fare product and maps it to the nearest filed class, so reservation systems that sell by class keep working. Your direct channel can sell at the continuous price.
A guardrail stops the direct price from undercutting the filed fare by more than 8 percent.
$212Light$262Standard$348Flex
Y
B
M
K
Q
V
W
S
T
L
$470
$410
$355
$305
$262
$225
$188
$152
$120
$95
| Fare product | Continuous price | Filed class | Filed fare | Gap |
|---|---|---|---|---|
| Light | $212 | V | $225 | 5.8% |
| Standard | $262 | Q | $262 | none |
| Flex | $348 | M | $355 | 2.0% |
Radar compares your fare with every carrier on the route
Radar lists the carriers on a route with their weekly flights and current fare, and flags each fare below yours.
It needs a competitor fare feed. The pilot used fares collected from public shopping sites in July 2026.
| Carrier | Flights a week | Fare | Against your fare |
|---|---|---|---|
| You | 14 | $235 | |
| Competitor A | 21 | $248 | $13 above you |
| Competitor B | 7 | $262 | $27 above you |
| Competitor C | 7 | $205 | $30 below you |
Bid prices keep seats for the passengers who pay most for them
A bid price is the lowest fare worth accepting on one flight leg. An itinerary is sold only when its fare covers the bid prices of every leg it uses. At a hub this stops a cheap connecting fare from taking a seat that a local passenger would pay more for.
One decision, as the engine records it:
- Request
- A to C via the hub, fare $240
- Leg 1
- A to the hub, bid price $120
- Leg 2
- The hub to C, bid price $200
- Rule
- Accept when the fare is at least the sum of the bid prices
- Sum
- $320
- Decision
- Closed. The fare is $80 short.
Full legLeg with spare seats
| Itinerary | Fare | Bid prices | Margin | Decision | Seats |
|---|---|---|---|---|---|
| A to the hub | $120 | $120 | $0 | Open | 19 |
| B to the hub | $135 | $135 | $0 | Open | 36 |
| Hub to C | $200 | $200 | $0 | Open | 50 |
| Hub to D | $305 | $0 | $305 | Open | 22 |
| A to C via the hub | $240 | $320 | -$80 | Closed | 0 |
| B to C via the hub | $230 | $335 | -$105 | Closed | 0 |
| A to D via the hub | $360 | $120 | $240 | Open | 10 |
| B to D via the hub | $360 | $135 | $225 | Open | 14 |
The pilot test pack passed 87 of 87 checks on a regional carrier's network
The engines were configured with the fleet, network and fare ladders of a regional African carrier. Fares are anchored to July 2026 public shopping data. Bookings are simulated, because the carrier had no forward-booking history to share. No airline runs Yieldcraft in production yet.
Bid-price control earned $2,288 more per run than selling first come, first served
The simulation sells the same demand at a 4-flight hub under four rules. Each figure is the average revenue of 300 simulated selling runs.
Each result below comes from a check that can be rerun
| Check | Result | Basis |
|---|---|---|
| Engine checks | 87 of 87 pass, in 22 groups covering pricing, fare filing, revenue management and network bid prices. | The shipping engine code, configured with the carrier's fleet, network and fare ladders |
| Two-flight connection | Bid-price control earned $25,000 against $16,000 with no control. It refused a $200 connecting fare that would have displaced $500 of local revenue. | Simulated bookings; fares from July 2026 public shopping data |
| Overbooking limit | On one 50-seat flight, authorising 55 seats raised revenue from $12,092 to $13,288 with 0.1 denied boardings expected. Authorising 63 added no revenue and raised expected denied boardings to 2.8. | Simulated bookings; fares from July 2026 public shopping data |
| Fault found and fixed | On a domestic route one shared fare ladder let the direct price of $172 undercut the filed fare of $188 by 8.5%, past the 8% limit. A separate domestic ladder removed the conflict, checked in the same run. | Fares from July 2026 public shopping data |
The same engine paces every departure
| Flight | Aircraft | Days out | Booked of seats | Forecast load | Pace | Empty seats expected | Turned away |
|---|---|---|---|---|---|---|---|
| Hub to C | 50-seat jet | 2 | 57 of 50 | 99% | On pace | 0 | 1 |
| Hub to A | 29-seat turboprop | 3 | 32 of 29 | 97% | On pace | 1 | 7 |
| Hub to C | 50-seat jet | 5 | 39 of 50 | 84% | Behind | 8 | 0 |
| Hub to B | 29-seat turboprop | 7 | 32 of 29 | 97% | On pace | 1 | 8 |
| Hub to E | 29-seat turboprop | 9 | 28 of 29 | 84% | On pace | 5 | 0 |
| Hub to D | 50-seat jet | 14 | 43 of 50 | 80% | On pace | 10 | 0 |
| Hub to C | 50-seat jet | 21 | 55 of 50 | 100% | Ahead | 0 | 8 |
| Hub to F | 50-seat jet | 26 | 31 of 50 | 83% | Behind | 9 | 0 |
The 8 nearest departures in the pilot run. Booked can exceed seats because the engine authorises overbooking against expected no-shows. Airport names are withheld. Source: pilot test pack, simulated bookings.
Yieldcraft works from a ticketing extract and the fares you already file
| You provide | A ticketing extract from your reservation system, with issue dates. Your filed fare ladder and fare rules. |
|---|---|
| Yieldcraft returns | A forecast for each flight, a price for each fare product mapped to a filed class, a bid price for each flight leg, and a list of competitor fares for each route. |
| Stays as it is | Your fleet, your fare classes and your distribution channels. |
Booking curves are rebuilt from the issue dates in a ticketing extract, so a carrier without stored booking history can start. The pilot tested this on a synthetic extract.
Your commercial data is used to forecast and price your own network. It never trains another airline's model.
The demo console runs in your browser with sample data
It shows the four screens an analyst uses: overview, Foresight, Optima and Radar. The numbers in it are illustrative.
