Prophesee Horizon

Know how the year ends.

Performance Intelligence is knowing the probability that a plan will hit its target, what is moving that number, and which action moves it back. Every target has odds, every move has a driver, every claim has evidence.

FY plan: revenue
10,000 simulated year-ends
P 74%
In 74 of 100 simulated year-ends this plan hits target.
ActualsTodayYear-end
New accounts+40
Base P 48%P 74%
See it working

The planning workspace

Watch the lever move: P 48% recounts to 71%, the fan lifts toward target, and the negotiation signs off.

See it on your data →
Track

Odds, not status

Reviews tell you where you are. They dont tell you where youll finish. A status field stays green for ten months and turns red six weeks before the close, too late to change the year. The model saw the same year differently: odds fell from 74% to 61% in March, with the cause named as discount overrides in DACH, at twice policy.

  • NOV, the status turns red. Ten months of green, then red six weeks before the close.
  • MAR, the model flags it. Odds fall from 74% to 61%, cause named.
  • JUN, the action is funded. Re-pricing tested before any money moves: P 48% to 71%.
  • DEC, delivered. P 85%. The close confirms what the model had been saying since March.

Nine months is the difference between managing the year and explaining it.

Heuristic risk vs probabilistic risk
WHAT THE STATUS SAIDSuddenly redWHAT THE MODEL SAID74%48%85%JANJUNDEC
Plan

Plan by simulation

Most tools split the target equally, then discover in December who could never carry it. Horizon sets the target on thousands of simulated futures, not one guess, so the delivery risk is visible before anyone commits to the number, and distributes it according to the probability of success.

  • Target P 74%. In 74 of 100 simulated year-ends, this plan finishes on or above target.
  • Team A P 82% carries more. Team B P 61% carries its share.
  • Team C P 44% gets support, not a fantasy.
  • Uncertainty is carried from each driver rather than added at the top.

The best plan is the one most likely to succeed.

10,000 simulated year-ends
TARGETTODAYYEAR-END
P 74%of simulated year-ends finish on or above target
Explain

Follow the number down

A gap that cannot be traced becomes a room full of opinions. Every probability has a reason here: group to region to unit to objective to root cause, four clicks, and every step is a fact. Each level carries its contribution to the gap, so the room argues about size, not existence.

  • Sized at every level, so the trail is evidence rather than assertion.
  • It ends at a lever, never at ask around in the review.
  • Every input traces back to the system it came from.

This plan is at risk because discounting in DACH is out of policy. A fact, not a hunch.

A drill from the group revenue gap down through region, unit and objective to the root cause, with the share of the gap at every level

Group target P 48%, -4.2pts vs plan. EMEA -2.9pts, DACH unit -2.1pts, price realisation the failing objective, discount overrides at 2x policy.

Simulate

Test the move before you fund it

Knowing the cause is only useful if you can test the response. Before you change the plan, see what happens if you do. Accounts, growth per account, headcount: whatever moves your number. One pull ripples through every derived metric and objective it touches, and the plan moves from P 48% to P 71%.

  • Derived metrics included. EBITDA moves when revenue or cost moves, the maths is the graph, not a hope.
  • Objectives update. Every affected objective recomputes its odds the moment you move a lever.
  • Committed with governance. The plan of record changes once, approved, and everyone sees the same new number.

Test the move before you commit the quarter.

Pull the lever, watch it propagate
New accounts+40
Revenue+2.1%
Cost+0.4%
EBITDA+0.8%
P 48%P 71%
Govern

Negotiate inside the plan

The negotiation happens inside the plan, not in email: the counter-offer, its cost, and the agreement, on the record. Team B counters -24% to -21%. Cost disclosed, resets 3 sign-offs. Accepted, recorded forever. No forked spreadsheets, and the original ask survives every negotiation.

  • What was agreed, on record. The value, the confidence, the person, the moment, for every approval.
  • One number, per viewer. The number is computed once for each viewers scope.
  • A close you can sign. Attestation with evidence, blocking items block, history append-only.
  • Private plans stay private. They simply stop being a different version of the truth.

Structure for the organisation, flexibility for the individual: one version of the truth.

A negotiation ledger recording the offer and the signed-off counter, each with its time, its owner and its probability

Group distributes -24% to Team B, set on simulation at P 61%. Team B counters -24% to -21%, cost disclosed, resets 3 sign-offs. Accepted, recorded forever. On the record. History append-only.

One decision layer

Runs with the other engines

Horizon and Pulse run on the same event backbone, scoped to each viewers permissions at every step. The exception that interrupts someone this morning is the same event that moves the years probability.

Questions executives ask

See your plan with its odds attached

Performance Intelligence turns targets into outcomes. Plan, track, deliver, on your own numbers.