Designing a Project Health Score in Salesforce: Signals, Weighting, and Ownership

Table of Contents
ToggleProject dashboards are very good at displaying confidence. Whether that confidence is justified is another matter. A project can look perfectly healthy on screen while the people working on it know otherwise. A milestone slipped last week. The team has already used more hours than planned. A key resource is about to disappear into another project. Yet the status indicator remains cheerfully green, apparently working from a more optimistic version of events.
The problem is not the colour. It is the evidence behind it. A useful project health score should combine current delivery evidence, show why it changed, and tell the right person what needs attention. Salesforce can keep the underlying project, task, time, resource, budget, and risk records close together. The design still matters.
Here is a practical way to build one.
Start by defining what “healthy” means
Before selecting fields or building a formula, agree on the decision the score is meant to support. For most delivery teams, a healthy project is not simply one with no overdue tasks. It is a project that is likely to meet its current commitments without an avoidable surprise in schedule, effort, cost, staffing, or risk.
Keep the definition short enough to use in a meeting. For example:
Project health measures our confidence that the project will meet its current delivery and commercial commitments, based on fresh operational evidence.
This gives you two tests for every signal you consider adding:
- Does it change confidence in the project outcome?
- Can someone act on it?
If the answer to both is no, it may be interesting data, but it does not belong in the health score.
Choose a small set of leading signals
Too many inputs create a polished average that hides why a project needs help. This is how a dashboard gradually develops a job description of its own.
Start with five dimensions.
| Dimension | Useful signals | What the signals tell you |
|---|---|---|
| Schedule | Milestone variance, overdue critical tasks, dependency delays | Whether the delivery plan is still credible |
| Effort | Actual versus planned hours, recent effort trend, missing time | Whether work is consuming effort as expected |
| Budget | Forecast at completion, budget consumed versus progress, unapproved billable work | Whether the commercial outcome is moving |
| Staffing | Capacity coverage, overloaded assignments, unfilled skill needs | Whether the plan has the people required to succeed |
| Risk | Open high-impact risks, overdue mitigations, unresolved issues | Whether known threats are being controlled |
Freshness is also a signal. If half the team has not entered time since last Thursday, an effort score calculated on Monday morning should not receive full confidence. Missing data is not good news. It is missing data.
Weight the score without creating false precision
A straightforward model is easier to explain and improve than a clever one.
You could begin with a 100-point score:
| Dimension | Weight |
|---|---|
| Schedule | 25 points |
| Effort | 20 points |
| Budget | 20 points |
| Staffing | 15 points |
| Risk | 20 points |
| Total | 100 points |
Each dimension receives a score based on agreed thresholds. For schedule, a project with no critical delay might receive the full 25 points. A small recoverable variance could receive 18. A missed milestone with no approved recovery plan might receive 5.
1. Use ranges, not theatrical decimals
A health score of 73.42 does not contain more wisdom than a score of 73. It simply looks as though the formula attended business school.
Use clear bands such as:
- 80 to 100: Healthy
- 60 to 79: Watch
- Below 60: Intervention required
2. Add hard-stop rules
A weighted average can disguise one serious problem. A major compliance risk should not be cancelled out by perfect timesheet completion.
Add a small number of override rules. For example:
- An unmitigated critical risk caps the score at 59.
- A missed contractual milestone without an approved recovery plan caps the score at 59.
- Stale or incomplete source data changes the status to “Unknown,” not green.
3. Show the components
Never display only the total. Put the five dimension scores beside it and retain a short explanation of the biggest change.
“Health fell from 82 to 68 because schedule variance increased and a staffing gap remains open” is useful. “Health: 68” is a conversation starter, usually followed by someone asking for the spreadsheet.
Give every exception an owner
A project health score is not finished when the indicator turns amber or red. That is when the operating process begins.
Every material exception needs:
- An owner
- A next action
- A due date
- A status
- A record of the decision or accepted risk
A schedule recovery action may belong to the project manager. A staffing gap may belong to a resource manager. A billing hold may require finance. In Salesforce, a threshold breach can create or update an exception, assign it by type, and place the action beside the project.
Track time to ownership and time to resolution. If red projects are visible but remain ownerless for five days, the dashboard is reporting a governance problem rather than solving one.
Test the score against completed projects
Do not launch the first formula across the whole portfolio and assume the mathematics will behave.
Select a mix of completed projects, including good outcomes, recoveries and quiet margin losses. Recalculate the score week by week using only the data available at the time.
Ask four questions:
- Did the score change before the outcome became obvious?
- Which signals produced useful warning?
- Which signals created noise?
- Would the stated owner have been able to act?
Pay particular attention to false greens. A score that occasionally asks a manager to investigate a healthy project is mildly inconvenient. A score that repeatedly labels troubled projects as healthy is decorative.
Adjust thresholds and weights based on what the back-test shows. Keep a record of those changes. Otherwise, six months later, nobody will remember why budget carries 20 points while staffing carries 15.
Put the score into the weekly operating rhythm
For many teams, daily calculation with a weekly portfolio review is sufficient. High-risk work may justify more frequent updates.
In Salesforce, keep the current score on the project record and store score history separately. That history lets you see direction, not just position. A project holding at 72 for three weeks is different from one that fell from 91 to 72 yesterday.
Your portfolio view should make four things easy to see:
- Current score and status
- Direction of travel
- Largest contributing exception
- Owner and next action
Then measure whether the process itself is working:
- Percentage of active projects with fresh source data
- Percentage showing “Unknown” because data is incomplete
- Time from exception to assigned owner
- Time from ownership to action
- Number of false-green projects discovered after the fact
- Accuracy of forecasted completion, effort, and margin over time
These metrics prevent the team from celebrating a beautiful score that nobody trusts or uses.
A sensible first version
Your first Salesforce project health score does not need predictive AI or a colour palette that requires its own legend. Start with five dimensions, a transparent calculation and two or three hard-stop rules. Route each exception to an owner, then compare the result with actual outcomes.
PK4 Project & Time brings project activity, time, resource, budget, and risk information into Salesforce, which gives teams the raw material for a health score tied to current delivery work. More importantly, the resulting exception can stay connected to the project, the person responsible, and the action required.
A project health score should not make management feel informed. It should help management intervene while intervention is still useful.
Explore PK4 Project & Time or schedule a short walkthrough to see how project and time data can support clearer delivery decisions inside Salesforce.
Co-Founder PK4 Tech
