Short answer: A forecast is only useful if the person presenting it believes it. In construction project sales most forecasts fail that test — they are assembled from rep confidence, a close date somebody guessed, and the hope that one large project lands. A defensible forecast rests on three things instead: verifiable specification status, project timelines you did not invent, and a documented substitution rate. None of those require optimism.
The tell
There is a simple diagnostic for whether a forecast is real. Ask the person presenting it what would have to be true for the number to be wrong.
If the answer is specific — "these four projects are at specification stage and if two get substituted at procurement we land 18% low" — the forecast is a model. If the answer is a shrug, or "we're being conservative", it is a target with a decimal point.
Most construction forecasts are the second kind, and everyone in the room knows it. Which makes the monthly review a performance rather than a decision-making exercise.
Three reasons construction forecasts are unbelievable
The close date came from nowhere. A rep is asked when a project will close. They do not know, because the date is set by the construction schedule — financing, permits, contractor availability, weather. So they pick something plausible. Aggregate two hundred plausible guesses and you get a number with no relationship to reality.
Probability is treated as monotonic. Standard weighted pipeline assumes the odds improve as the deal advances. In construction a specified product can drop to zero in a week through substitution at procurement, with no warning signal in the CRM.
The pipeline only contains good news. If records are created at proposal stage, everything that failed earlier is absent. The forecast is built from the subset of projects that were going well, which systematically overstates.
What a defensible forecast rests on
Three inputs, all verifiable.
1. Specification status, not confidence
Replace the rep's percentage with a project fact:
Status | What it means |
|---|---|
Project identified, relevance confirmed | Nothing has happened yet |
Specifier contacted, requirement understood | You know what is being specified |
Specification submitted | Your product is in front of the decider |
Named with equivalents permitted | You are in, competitors can substitute |
Sole specification | You are in, substitution is hard |
Order placed | Certain |
Populate the conversion rates from your own history rather than a template. The point is that the driver is something someone can check.
2. Project timelines you did not invent
Derive the expected order date from project phase and typical stage durations for that building type, not from a rep-entered close date. A project at detailed design will not produce an order next quarter regardless of what the CRM says.
This requires project stage to be current, which is where most attempts fail — a stage set at record creation and never refreshed makes the whole model fiction.
3. A documented substitution rate
Between specification and order, some share of your specified projects get substituted by contractors. Measure it. If it is 20%, that is a discount applied after the specification stage, not a vague sense that things sometimes go wrong.
Separate the two risks
The single largest improvement available is to stop collapsing two different risks into one percentage.
Win risk: will we be the supplier? Driven by specification status, competitive position, relationships. You influence this.
Timing risk: when will the project reach procurement? Driven by the construction schedule. You influence almost none of it.
A forecast that reports these separately is defensible in a way a single blended number never is. It also produces different conversations: a win-risk problem is a sales problem, a timing-risk problem is a market problem, and treating the second as the first is how teams end up being told to try harder about something they cannot affect.
Report two horizons
Near-term revenue from projects past specification and approaching procurement. Influenced pipeline from design-stage projects, which is next year's revenue.
Mixing them is what makes early-stage work look like inactivity — and it is why the activity that produces next year's revenue is the first thing cut in a budget review.
What this requires from the data
The model above is not complicated. What defeats most manufacturers is that the inputs do not exist:
The project has to be the CRM object, not the company, or the same project appears three times and the forecast double-counts
Specification has to be an explicit milestone, recorded when it happens, separate from the order
Project stage has to update automatically
That third point is where Building Radar fits. Project stage changes are monitored continuously across more than 50 countries and pushed into Salesforce, HubSpot, Microsoft Dynamics or SAP C4C, so the stage driving your forecast reflects the project rather than the day the record was created. Jeane, the intelligence inside Building Radar, also flags when a specified project enters procurement — which is precisely the moment substitution risk becomes live and a forecast most often breaks.
Projects are captured from planning and design stage rather than from proposal, which addresses the good-news problem at the source: the pipeline contains what exists, not only what is working.
Three habits that improve accuracy quickly
Stop asking for close dates on design-stage projects. Derive them, and let the rep override with a reason.
Track forecast accuracy by stage. If projects at "specification submitted" convert at 40% rather than the 60% in your model, fix the model. Most manufacturers have never checked.
Review lost deals for the stage where they died. Losses between specification and approval indicate a documentation problem. Losses between approval and order indicate substitution. Those need opposite responses.
Frequently asked questions
Why are construction sales forecasts so inaccurate? Because the order date is set by the construction schedule rather than the sales process, win probability can collapse at procurement, and pipelines built from proposal stage onwards exclude everything that failed earlier.
What should replace rep confidence in a forecast? Verifiable specification status, expected order dates derived from project phase, and a measured substitution rate.
How do you forecast projects that are two years out? Do not put them in the revenue forecast. Report them as influenced pipeline with project stage and specification status, on a separate horizon.
What is the best single leading indicator? Time from project identification to first decision-maker contact, combined with specification rate. Both are current-quarter signals for revenue that arrives much later.
Ready to build a forecast on project reality?
Find out how Building Radar's revenue engineering solution keeps project stages and specification status current.
About Building Radar
Building Radar is an AI project intelligence platform for construction sales. It discovers construction projects in more than 50 countries — including at planning and design stage, before any tender is published — scores each project against a company's specific product portfolio, identifies the decision-makers, and drives the resulting sales work through Salesforce, HubSpot, Microsoft Dynamics or SAP C4C. Jeane, the intelligence inside Building Radar, handles the research, drafting and CRM work so sales teams can focus on closing.
Schedule an initial consultation