Forecast Cockpit
The Forecast Cockpit is where you tune the company-wide financial-statements forecast by moving a small number of dials and watching the P&L respond. It is deliberately not the place you forecast a contract — that is Contract Forecasting, one contract at a time. The Cockpit sits a level above: it asks "what happens to the whole business if collections slip ten days, or if the services division grows three points faster than last year?" This guide explains what the dials mean, what the colors are telling you, and — the part that matters most — the difference between a preview and an Apply.
For: CEOs, CFOs, and anyone who owns the company forecast · Time: ~10 minutes · You'll need: the FP&A module and a forecast that already has a baseline (the Cockpit compares every dial against trailing actuals or your prior fiscal year).
How to access
The Cockpit is the first tab of Forecast. Open FP&A → Forecast and it is where you land.
There is no longer a separate Forecast Cockpit item in the sidebar — it became a
tab when the Forecast module was reorganized. An old /cockpit link still works
and redirects you to the tab, so bookmarks and older documents will not break.
Part 1 — The ideas you need first (read once)
A dial is an assumption, not a number you are typing in
Every dial is an assumption rather than a number you are typing into a statement — a growth rate, a gross-profit percentage, a collection period, a cost ratio. You are not editing the forecast; you are editing the inputs it is computed from.
Wired end to end. DSO (collection days), DPO (payment days) and CapEx as a percent of revenue are saved as balance-sheet assumptions the three-statement engine reads. The four indirect dials — Overhead Spend Scaling, G&A Spend Scaling, and both fringe rates — are wired too, as of 2026-07-31. Apply writes the pool elasticity and the fringe parameters the indirect forecast engine reads back, so moving them changes the applied forecast.
This paragraph used to say the opposite. Until 2026-07-31 those four dials wrote to a table no engine read, and this guide correctly warned you not to trust them. That gap is closed. If you remember being told these dials only record an intent, that is why — and it is no longer true.
The per-division dials are the exception, and the revenue one is worse than "partial". They write the closing year's targets while the current year is built from a monthly table the Cockpit does not write, so in the current fiscal year they can move the preview without moving the applied forecast.
The per-division revenue-growth dial was repaired on 2026-08-01 and this guide previously told you not to use it. That instruction is withdrawn — steer on it. It used to render an identical figure for every division, because the value was computed against a denominator derived from the very targets it exists to steer, so the division's own target canceled out. It now reads against that division's real prior-year revenue, the round trip closes (set a dial to 10% and it reads back 10%), and a shrinking division no longer badges itself "aggressive".
What the dial means, since the repair settled it: the number is new business as a share of that division's prior-year revenue — the Growth term of Existing + Recompete + Growth, not a total-revenue growth rate. A division reading 0.09 is targeting new business worth 9% of what it did last year.
A division's share of company revenue is your own revenue. The Cockpit splits company revenue across divisions by each division's share, and that share is read from Arcvue's own contract actuals by each contract's division, not from the project a document was filed under in your previous system. Revenue booked to a project that belongs to another firm, such as a pass-through venture you subcontract from, does not count toward any of your divisions, so a division's share can be smaller than an old report showed.
So: steer on all of them. The one caveat that survives is the timing one above — in the current fiscal year the per-division dials can move the preview without moving the applied forecast, because the applied number comes from a monthly table the Cockpit does not write.
There are more than twenty dials. They are grouped so you are never scrolling a flat list:
| Section | What lives there |
|---|---|
| Revenue / Gross Profit | Revenue growth and GP% — including per-division dials, with each division's revenue-growth row sitting directly above its GP% row |
| Cost | Cost ratios and per-division cost dials |
| Balance Sheet | The working-capital dials — collection and payment periods |
| Other | Anything that does not belong to the three above |
The color is a comparison, not a judgment
Each dial carries an aggression meter that classifies where you have set it relative to a baseline — either trailing actuals or your prior fiscal year. Four states:
| Chip | What it means |
|---|---|
| Conservative | You are assuming worse than the baseline |
| On trend | You are assuming roughly what the business has actually been doing |
| Aggressive | You are assuming better than the baseline |
| No baseline | There is no history to compare against, so no claim is made |
"Aggressive" is not an error and "Conservative" is not a virtue. The meter exists so that when you hand the forecast to a lender or a board, you can say exactly which assumptions depart from history and by how much. A plan built entirely of aggressive dials is not wrong — it is a plan that needs a story.
Click any dial's explanation to see what it feeds and how its baseline was derived.
Preview and Apply are not the same operation
This is the one thing to take away from this guide.
- Preview happens automatically as you drag. It uses analytic delta math — a fast approximation, sub-second, good enough to steer by. Nothing is saved and nothing else in Arcvue changes.
- Apply runs the full forecast pipeline — the three-statement engine, the indirect forecast, and the out-year generator. It is slower, it is the real computation, and it commits the new assumptions as the live forecast.
So the numbers you steer by while dragging are indicative, and the numbers after Apply are the forecast for the dials that reach it — see the caution above.
Where a preview and an applied forecast disagree, the applied number is the truthful one. If a preview looks surprising, Apply and re-read it before acting on it.
This paragraph previously told you a preview/Apply disagreement on the indirect dials was EXPECTED, because Apply "hands the pipeline a rate it does not read back". That stopped being true on 2026-07-31. Those dials now write where the engine reads, so a disagreement there is a FAULT worth reporting rather than a known gap. The per-division revenue-growth dial was the remaining exception and no longer is — it was repaired on 2026-08-01; see above.
What Arcvue deliberately will not do
The Cockpit will not invent a baseline it does not have — dials with no history show No baseline rather than a guess. It will not let a dial leave a sensible range: day-based dials are held between 15 and 150 days and within about ±60 days of their anchor, and percentage dials are bounded relative to their own anchor. And it will not quietly overwrite your live forecast while you explore — only Apply does that.
Part 2 — How to run it
Step 1 — Read the baseline before you touch anything
Open the Cockpit and read the right-hand panel first. It shows the annual P&L as it stands: Revenue, Gross Profit, EBITDA, and EBITDA Margin. That is your starting point, and every number you see later is a movement away from it.
Step 2 — Move one dial and watch the right panel
Drag a slider. The right panel updates within a second, showing the previewed figure beside the baseline so you can see the size of the move rather than just the new value. Move one dial at a time until you have a feel for its leverage — several dials moved together are much harder to attribute.
Step 3 — Read the color, not just the number
After each move, check the chip. If a dial has moved from On trend to Aggressive, you have just made an assumption that departs from what the business has actually done. That may be exactly right — a signed contract, a new division — but it should be a decision, not a side effect of dragging.
Step 4 — Reset if you have lost the thread
Reset returns every dial to the applied baseline. It is the safe way out of an exploration that got away from you; it does not touch the saved forecast, because nothing you did during exploration was saved.
Step 5 — Decide: scenario or forecast
You have two ways to keep the work:
- Save as Scenario keeps this set of dial positions as a named scenario you can return to and compare, without changing the live forecast. This is the right choice for "what if" work, board cases, and anything you want to hold beside the current plan.
- Apply makes these assumptions the live forecast and runs the full pipeline. The button shows Applying… while it works. Use it when the plan has actually changed.
When in doubt, save as a scenario. A scenario is reversible in the sense that it changes nothing; an Apply is a decision about what the company's numbers now say.
Recent Changes is collapsed when you arrive, deliberately. The rail is for
driving the forecast, and a trail of past moves is reference rather than the
task. The count stays on its header, so collapsing hides the rows without
hiding that there ARE rows — an empty-looking section is telling you it is
closed, not that nothing has happened.
Annual Summary is the table under the monthly preview: one row per P&L
line, the year in one view, so you can see what a dial did to the whole year
rather than to a month.
Part 3 — When something looks wrong
"I moved a dial and nothing happened." Check whether that dial actually feeds the metric you were watching. A division-level cost dial can move EBITDA very little if that division is small. Open the dial's explanation to see what it feeds.
"The preview and the applied numbers disagree." Expected, within reason. Preview is analytic delta math; Apply runs the full pipeline including the indirect forecast and out-year generation. Small differences are the approximation. Large differences are worth raising — they usually mean a dial interacts with something the fast path does not model.
"A dial says No baseline." There is no trailing-actuals or prior-FY history for it. The dial still works; Arcvue is simply declining to tell you whether your value is conservative or aggressive, because it has nothing honest to compare against.
"The slider will not go where I want." Dials are range-bounded on purpose. Day-based dials clamp between 15 and 150 days and stay within roughly ±60 days of their anchor. If the value you need is outside that, the assumption probably belongs somewhere other than a dial — raise it rather than forcing it.
"I applied something I did not mean to." Move the dials back and Apply again. There is no undo stack; the forecast is whatever was last applied, so the fix is another Apply from the corrected positions.
One-line summary
The Cockpit is company-level assumption tuning: drag dials, watch an analytic preview, check the color chips to see which assumptions depart from history, then Save as Scenario to hold the idea or Apply to run the full pipeline and make it the forecast.
Related
- Contract Forecasting — the same forecast at contract level, one contract at a time. The Cockpit operates at portfolio level; that guide is where per-contract staffing, grids, and submissions live.
- Financial Statements — the three-statement projections the Cockpit's dials drive.
- Scenario Planning — what happens to a set of dial positions after you Save as Scenario.