Example
Floor 4 is slipping. What should the team do?
This is the running example used across this site, told in full: a multifamily project where the same electrical rough-in package repeats on every floor, and Floor 4 has fallen behind. Walk through how the picture—and the recommendation—changes as evidence arrives.
This example uses simulated construction data to demonstrate the product concept. It is not a customer result or an industry benchmark.
Step 1 of 6
The current project state
Electrical rough-in has run at plan on Floors 1–3. On Floor 4, production has dropped to 62% of the planned rate. The milestone that matters is turnover of Floor 4 to drywall in nine working days.
The platform's current estimate, with evidence still incomplete:
- Progress 48% complete (plan: 60%)
- Material readiness 5 of 8 unit kits complete Uncertain
- Crew Crew B, 6 electricians
- Turnover probability 54% (likely range 41–66%)
Uncertainty reflects incomplete field information, not a system fault.
Evidence arrives from the field
Over the next day, three field updates arrive. Each one changes the estimated state of the work—and the value of the available actions.
- Mon 07:20 Daily report: two electricians pulled to punch-list work on Floor 2.
- Mon 09:45 Warehouse: 3 of 8 material kits missing device boxes and fittings.
- Tue 11:10 Superintendent: a second crew becomes available Thursday.
After these updates, turnover probability under the current plan drops to 38% (range 27–51%).
The decision point
A second crew is available Thursday. The instinctive response is to add it to Floor 4 immediately. The platform compares the practical options from the same view of current conditions:
| Action | Turnover probability | Main consideration |
|---|---|---|
| Continue current plan | 38% (27–51%) | No added cost; delay likely continues |
| Add crew Thursday | 47% (35–58%) | Incomplete kits limit useful work; congestion risk |
| Expedite kits only | 61% (50–71%) | Procurement cost; crew stays productive |
| Expedite kits, then add crew | 74% (63–83%) | Requires kits on site before Thursday |
Illustrative results based on a simulated project scenario.
The historical trap
Looking only at history, adding crews appears to make things worse: across past projects, work that received additional labor finished later than work that did not.
The history is misleading. Crews were added to areas that were already in trouble. The poor outcomes reflect the conditions that triggered the decision—not the effect of the added labor.
Cornerstone accounts for the conditions surrounding past decisions before estimating whether an action is likely to help now.
The context-adjusted conclusion
Under current conditions—incomplete kits, a partially staffed crew, congestion risk in occupied units—adding labor immediately is estimated to produce limited benefit. Added labor is likely to help only after readiness constraints are addressed.
The recommendation: expedite the missing kits now, then add the second crew Thursday, with the key assumption that kits arrive within two working days. If procurement cannot confirm that, the alternative is to resequence the second crew to Floor 5, where kits are complete.
The result — and what was learned
The team expedites the kits; they arrive Wednesday afternoon. The second crew starts Thursday on a ready workface. By the following Monday, production is back to 94% of plan and turnover probability has risen to 81% (range 72–88%).
The observed outcome—conditions, decision, result—feeds back into the platform, improving early estimates for the same repeated work on Floors 5 through 12.
- Production vs. plan 94%
- Turnover probability 81% (range 72–88%)
- Learning Kit readiness threshold updated for Floors 5–12
Illustrative product concept — simulated scenario.
What this example deliberately leaves out
This public walkthrough shows the decision logic: conditions, actions, expected consequences, confidence, and learning. It does not expose model parameters, adjustment sets, equations, or update mechanics. A more detailed demonstration is available privately under appropriate confidentiality.