Arcvue
Forecasting & planning—assumptions with authors
Somebody says grow eight percent. Here is what that sentence actually commits you to.
Arcvue Forecasting turns a growth assumption into a contract-level plan and scores it against your own trailing performance, on Arcvue’s ledger or your own. This page follows Arcvue carrying one assumption—grow eight percent—through nine steps, for an invented contractor.
Specimen Systems LLC · an example 120-person contractor · all figures synthetic
It begins as a sentence in a meeting, and at that moment it commits you to nothing at all.
Nothing is inherited yet. Somebody—a board member, a lender, the CEO—says the company should grow eight percent. It is a reasonable thing to say and it is not yet a plan, a forecast, or a number anybody can act on.
What happens next in most firms is a spreadsheet template emailed to division leads, who fill in their best guesses with no historical context, and send it back for somebody to reconcile by hand. The cycle takes weeks, produces a budget nobody fully trusts, and has to be repeated the moment an assumption changes.
The question this page follows. Not what is the forecast—that is an output, and outputs are easy to produce and hard to defend. The question is what does this assumption commit us to, and who will know it was ours?
Synthetic. An invented firm. Follow it down—the interesting number is not the 8%. It is the 16%, and nobody said it out loud.
The assumption is company-level, so it is set once rather than argued across two hundred contracts.
The sentence from stop 01 has to become something a system can hold. The instinct is to push it down—ask every division lead to find their share of the eight percent—and that is the step that turns one afternoon into six weeks.
Contract forecasting is one contract at a time. Scenarios are discrete named cases compared side by side. This is neither: it is the company-level assumption set the whole forecast is computed from. Revenue and gross profit by division, cost ratios, working-capital and capex levers. You are not editing a spreadsheet of outputs; you are moving the assumptions the outputs are derived from, and watching the derivation happen beside you.
The dial is scored against what this company has actually done, and eight percent is aggressive.
A single dial exists after stop 02. Scoring it against precedent answers the question a lender or a board will ask, and that is not the question that decides whether the year works.
The one that decides it is how much new business eight percent actually requires once the book you already hold is counted. Existing contracts carry their own forecast into next year—some growing on surge, some shrinking as scope comes off—and every recompete is revenue you defend rather than win. Net all of it against the target first. What is left is the only part that is genuinely new work, and it is rarely the number on the dial.
Which is why the base has to be current before any of it is worth computing. The requirement here is netted against what the business is carrying today—posted work and work still in flight, every contract at its present value and status. Netted against a book that closed a quarter ago, the same arithmetic is meaningless at best and wrong at worst, and it is wrong in the direction nobody checks: it looks finished.
Measured off trailing actuals, this business has been growing at four percent. So eight is not wrong. It is aggressive, which is a different thing: an assumption that departs from what the business has done, and therefore one that needs a story rather than a correction.
| Setting | Growth | FY2026 revenue | Score |
|---|---|---|---|
| Current run rate | 0.0% | 31,800,000.00 | The business as it stands |
| Trailing actual | 4.0% | 33,072,000.00 | On trend—what this company has done |
| The ask | 8.0% | 34,344,000.00 | Aggressive—double the trailing rate |
Screen Arcvue forecast cockpit, dial scoring against trailing actuals. Read this one row by row, not as a column. The three settings are alternatives, not additions, so no total rule is drawn under them. Basis the score is computed from this company's own history, not from a benchmark. “Aggressive” here means relative to you, which is the only version of the word that is useful. Synthetic sample.
Trailing growth answers whether the number departs from precedent. It does not answer whether it is reachable, because next year is not one number: it is funded work that continues, a recompete book that has to be defended, and go-get that has to be won. Eight percent on a portfolio whose recompetes all land in one quarter is a different proposition from eight percent where they do not—and the two score identically against trailing growth. Stop 06 splits the target three ways for that reason, and whether the go-get half is reachable is a coverage question rather than a growth one.
Half the year is already closed, so an eight percent year is a sixteen percent half.
This is the number nobody said in the meeting.
Stop 03 scored the dial against history. This stop asks something history cannot answer: where is the growth supposed to go? It is July. Six months are closed, audited and immovable—$15,900,000.00 of the year has already happened and no assumption can touch it.
So the entire $2,544,000.00 has to land in the half that is left, against a base of $15,900,000.00. That is sixteen percent, not eight. The ask did not change; the amount of year available to absorb it did, and it halves every month somebody spends deciding.
Screen Arcvue monthly view, closed periods marked. Basis the monthly view marks the periods already closed, so it is obvious at a glance which part of the year an assumption can still move. That marking is not a convenience—it is what turns 8% into 16% before somebody commits to it. Synthetic sample.
The statements redraw beside the dial, and none of it is saved.
Sixteen percent of the open half is now a specific, checkable claim. Before anyone commits to it, they should be able to see what it does to everything else—and be able to walk away.
Preview is analytic delta math, so the monthly and annual statements beside the rail answer immediately: revenue, direct costs, gross profit, indirect, EBITDA, operating cash, and ending cash, year by year to the end of the forecast window. It commits nothing. Nothing is written, no scenario is created, and the current plan is exactly where it was.
Why that distinction is load-bearing. A tool that saves as you explore makes exploring expensive, so people stop exploring and argue from intuition instead. Separating preview from apply is what makes it safe to ask a stupid question—which is most of what planning is.
Three and a half million of that target is work nobody has won yet.
The preview at stop 05 showed what the target does. This stop shows what it is made of, which is the question that decides whether it is a plan or a wish. Revenue is broken out as funded, recompete and go-get—and that split is the honest version of a growth plan, because it states how much of next year depends on work not yet won.
Screen Arcvue segment revenue composition, FY2026 forecast. Basis a segment with no forecast says so rather than reading as zero—an empty cell and a deliberate zero are different claims, and a plan that cannot tell them apart is not one. Synthetic sample.
Thirty percent of the target is not under contract. That is not an argument against the plan; it is the thing the plan is actually about, and it is better said in July than discovered in November.
Apply is the deliberate act, and it is the first thing on this page that commits you.
Everything through stop 06 was reversible. Apply is not: it runs the full pipeline—three-statement, indirect rates, out-years—and makes the dial set the forecast. Alternatively the positions are saved as a named scenario and today's plan is left exactly where it is, which is the right answer more often than people expect.
Scenarios run simultaneously—base, upside, and downside—against the actual debt structure and covenant thresholds, so a growth assumption and the covenant it might breach are visible in the same view rather than in two documents that meet at a board meeting.
The rates are not typed in. Apply recomputes indirect rates as part of the pipeline, off the filed provisional rates—the ones you actually bill at, which is what a forecast has to be priced on. Those sit beside the rates last night's pools actually produced, and the gap between the two is watched all year rather than discovered at settlement. A forecast built on a stale rate model is a forecast of a company that does not exist—and one built on a rate you are not billing at is a forecast of a different company entirely.
The people who actually know enter it themselves, and nobody consolidates anything.
A company-level dial is the fastest way to answer what would it take. It is not the plan. The plan comes from the people closest to each contract—program managers and division leads, who know more about what a contract will produce than anyone assembling numbers from outside it.
They log in, see their own contracts with full historical actuals as context, enter forecasts directly, and submit. Finance sees every submission as it arrives, reviews it in place, and approves. Approved forecasts feed the company forecast automatically—no manual consolidation, no re-entry, no chasing files. When a PM changes a number the company P&L moves with it.
| Method | Use it when | Built from |
|---|---|---|
| Manual | A PM has specific knowledge about upcoming months | Their entry, against visible history |
| Actuals-based | The contract runs at a stable rate | Prior performance, copied forward |
| Staffing calculator | T&M and labor-hour work | Headcount, position mix, and billing rates, on live indirect rates |
| Fixed price | FFP where revenue follows delivery | The milestone or delivery schedule |
| Locked | A lead has finalized a month and it must not move | Preserved through every recompute and rate update |
Screen Arcvue contract forecast methods. Basis the method is selectable per contract because contracts differ, and one forecasting method applied to all of them is a modeling convenience the business pays for later. Synthetic sample.
Six months later the useful question is not what the forecast says.
It is which assumption someone moved to get there.
The plan from stop 07 is now the number the business is being measured against, and it is December. Revenue came in under. The instinct is to interrogate the forecast, and it is the wrong instinct—the forecast was only ever arithmetic performed on an assumption.
Applied dial changes are recorded in the cockpit's own change log, so the assumption has an author and a date. A dial set is a decision. The eight percent was a sentence in a room at stop 01, with no author and no record; by stop 07 it was the company's plan. The change log is the only thing standing between those two facts.
Read the trail backward and it holds. Start at December's variance and the record Arcvue kept names the dial, who moved it, when, and what it scored against precedent at the time—aggressive, said out loud in July, six months before anyone needed it to have been said.