PPC is not a reporting goal—it is a measure of reliability

In LPS, the Weekly Work Plan is used to plan, commit, and then measure how much of it was completed as promised (PPC). The system thrives on commitments – not dashboards.

Typical mistake 2026:

Digital tools quickly deliver more status, more fields, more automation – but decisions and ownership remain diffuse. The result: more meetings, more discussions, the same instability.

Symptom vs. cause

• Symptom: PPC fluctuates greatly, lookahead seems “full,” Weekly still burns.
• Cause: Make-ready is not stable: constraints are identified too late, resolved too late, or managed without a clear owner.

Digital LPS can help here – but only if it strengthens make-ready and collaboration, not just creates transparency. An IGLC paper describes precisely this benefit of “situational awareness” through centrally managed information.

What “digital LPS” really means (and doesn’t mean) in 2026

Digital LPS = better situational awareness + faster response

In well-designed approaches, digital LPS delivers:

• Uniform database (instead of Excel islands),
• Visible constraints,
• Faster coordination,
• Shorter response times in sync.

But: Without data discipline, it becomes data noise

Many digital LPS implementations fail not because of the technology, but because of:

• inconsistent status definitions,
• unclear ownership,
• lack of routines,
• overly complex data models.

In short: Garbage in → Garbage out (only more expensive and faster).

AI in LPS: 3 use cases that can really support PPC

1) AI as a “constraint copilot” in make-ready

A useful application of AI is to suggest typical constraints for planned tasks (material, access, approvals, preliminary work) – based on similar tasks from the past.

Research into an “AI copilot” for make-ready planning in LPS is moving in precisely this direction.

Guiding principle: AI may provide hints – but it cannot replace commitments. Ownership and deadlines remain the responsibility of the team.

2) AI for “reason codes”: Accelerate learning, don’t assign blame

PPC without a learning loop is just statistics. AI can identify reason code clusters (e.g., “material,” “approvals,” “interfaces”) and suggest priorities for countermeasures.

Important: Reason codes must be few and stable categories – otherwise you are just training data garbage.

3) Early warning instead of rearview mirror (use with caution)

Generative AI and digital twin approaches are increasingly being combined with BIM/IoT/digital twins in construction to identify risks earlier. Reviews describe this direction of integration – including opportunities and hurdles.

Guideline: Forecasts are only as good as data quality + process stability. Otherwise, false security arises.

The 30-day plan: First stabilize, then digitize, then AI

At HSC, we first stabilize execution (cycle time, commitments, handovers) – only then is it worthwhile to scale digitally. HSC is not a standard consulting firm: We come in, bring order to complexity, and make implementation reliable again – with leadership and lean DNA. That’s why we recommend the following approach:

Week 1: Definition of Ready + WIP limit (immediately noticeable)

1. Definition of Ready for weekly tasks (e.g., “all constraints resolved”).
2. WIP limit in the weekly: fewer commitments, but reliable.
3. A status standard (e.g., “ready / not ready / blocked”) – no special cases.

Week 2: Constraint management with ownership

1. Constraint board (digital or analog) with: constraint, owner, date, status.
2. Daily “mini stand-up” (10 min) only for blocks/constraints.

Week 3: Launch “minimal” digital LPS

1. Minimal data model instead of field graveyard: task, owner, date, status, reason code.
2. One click from block → action (ticket/task), otherwise it remains observation.

Week 4: AI as assistant – only with action output

1. AI may:
   • Suggest constraints,
   • Consolidate daily reports on 3 risks + 3 actions,
   • Mark patterns in reason codes.
2. AI may not:
   • Create commitments “automatically,”
   • Dilute responsibility (“AI says…”),
   • Drive security/approval decisions without human review.

KPIs & guidelines: How to measure impact without a KPI graveyard

1. PPC (trend over 4–6 weeks) – not the daily value.
2. Constraint removal rate (resolved/planned in Lookahead).
3. Percentage of “Ready” in Weekly (tasks that meet the Definition of Ready).
4. Lead time “Block → resolved” (How quickly do you clear obstacles?).

Guidelines

• Max. 8–12 reason codes (otherwise noise).
• Max. 5 mandatory fields per task in the digital system (otherwise maintenance instead of production).
• Every AI recommendation must end with: Owner + Date + Next Step.

Risks & trade-offs

1) Data noise instead of decisions

More data points often increase the burden of discussion. The risk is real if data discipline is lacking or the data model is overloaded.

2) “False authority”: AI as an excuse

“AI says…” can shift responsibility. That’s exactly why AI needs clear roles, approvals, and checks in the project.

3) Adoption gap: Tool introduced, system not used

Many hurdles in AI/digital adoption are not technical, but organizational: skills, processes, training. This is also described as a key challenge in reviews of generative AI in construction.

Best practices: How to avoid data noise in digital LPS

• Start small: One project section, one cycle area, one team.
• One source of truth: No parallel Excel spreadsheets.
• Change meeting design: Weekly remains a decision-making meeting – not a status show.
• Schedule training: Foremen/supervisors as “last planners” must be able to lead the system, not just operate it.

FAQ

What is PPC in the Last Planner System?

PPC (“Percent Plan Complete”) measures how many of the tasks committed to in the Weekly Work Plan have been completed as promised.

How can AI increase PPC?

AI can indirectly increase PPC by supporting make-ready (identifying constraints earlier), evaluating reason codes (accelerating learning), and providing early warnings – if ownership and data discipline are in place.

What is the biggest mistake in digital LPS?

Too many fields, too many dashboards, too little ownership. Digital LPS must accelerate decisions – not status maintenance.

Conclusion: More PPC is possible – if AI strengthens the commitment system

Digital LPS + AI can be a real lever in 2026 – but only if it makes make-ready, collaboration, and learning loops more stable.

The principles remain the same. What is new are the tools. Those who stabilize first and then digitize will not be flooded with data – but will make better decisions at a steady pace.