7 Common Pitfalls When Contractors Use Excel Programmes on NEC Jobs

“Excel programme NEC” is a common setup on smaller jobs and fast tenders. It works right up until the moment someone needs the programme to do its real job: prove sequence, show cause-and-effect, and support decisions under change and delay pressure.

Excel is excellent for registers, dashboards and data capture. The pitfall is treating it as the programme engine on an NEC job.

When that happens, the project looks organised but becomes commercially fragile, because you can’t answer the questions a PM or commercial head will ask when risk is live:

What drove Completion? What changed this period? Which interface is blocking us? Are we double-counting time? Show me the impact pathway.

Here are seven pitfalls, with fixes that keep Excel where it’s strong and move the “truth” into a controllable planning system.

1. Dates without dependencies (you can’t demonstrate cause-and-effect)

What it looks like

A list of activities with start/finish dates and a simple Excel Gantt. Dependencies are implied, not engineered.

Why it fails

NEC conversations become causation conversations. Without a logic network, you can’t demonstrate the impact pathway to Completion or Key Dates—so discussions turn into opinion and negotiation.

Fix

Adopt a minimum viable logic network in a scheduling tool (P6/MS Project). Keep it lean:

• model the completion path and key handover gates

• include procurement, approvals, access and commissioning chains

• keep Excel for lists and registers, not sequence

  
  

2. Spreadsheet version chaos

What it looks like

Multiple tabs, copy-pastes, overwritten files, and unclear “issued vs working” versions.

Why it fails

When commercial pressure hits, you need a stable reference plan and a clear “what changed, when” record. Excel makes accidental rewrites easy.

Fix

Introduce an “issued pack” discipline:

• issued PDF + archived source file (read-only)

• defined data date each period

• archived revisions never overwritten

• one-page revision log (progress vs logic/duration vs assumptions)

  
  

3. Progress updates don’t re-forecast the sequence

What it looks like

Percentage complete is updated, dates are edited manually, and remaining work is adjusted by judgement. Site runs a separate lookahead because the spreadsheet isn’t trusted.

Why it fails

Even accurate progress data doesn’t help if the remaining sequence and critical path aren’t recalculated consistently. That’s why PMs lose trust: the programme stops matching reality.

Fix

Use Excel to collect progress inputs, but apply them in a logic-linked schedule with:

• fixed data date

• evidence-based actual starts/finishes

• measurable “definition of done”

• lookahead reconciled to the programme (one truth)

  
  

4. Interfaces live in meeting notes

What it looks like

Design release, access handover, permits/isolations, possessions, third-party approvals: tracked in minutes and emails, not modelled.

Why it fails

Interface risk is where NEC projects swing. If interfaces aren’t in the programme, you can’t show dependency, you can’t show when you needed inputs, and you can’t show knock-on effects credibly.

Fix

Turn interfaces into logic gates:

• programme gate activities/milestones for access/info/approvals/possessions

• ties to the work they enable

• Excel interface register that references programme activity IDs and evidence

  
  

5. Multiple changes become “manual date shifting”

What it looks like

Change after change arrives. Dates are pushed in Excel. The register grows. Nobody can confidently state whether impacts overlap, are already implemented, or are being claimed twice.

Why it fails

Once events overlap, you need controlled modelling and status discipline. Excel can track status, but it cannot robustly model combined impacts and preserve an audit trail.

Fix

Use a hybrid control model:

• Excel change register (status, dates, assumptions)

• schedule tool for modelling (insertion points / fragnets / scenario copies)

• implementation rule: accepted/implemented changes are reflected in the live schedule and marked implemented in the register

  
  

6. Excel creates “presentation programmes”

What it looks like

The Gantt looks professional. But there is no tested logic, no schedule health checks, and no defensible critical path.

Why it fails

Presentation creates false confidence. When the PM interrogates the plan (or when the job slips), the programme can’t explain itself.

Fix

Issue a decision-grade programme pack:

• schedule health snapshot (open ends, constraints, lags, out-of-sequence, negative float)

• PM review view: critical path + interfaces + key dates + next 6 weeks

• one-page narrative: what changed, why, what decisions are needed

Credibility is as much about governance and interrogation-ready packs as it is about bars on a page.

7. When the project turns noisy, Excel leaves you exposed

What it looks like

The team tries to prove delay/change with spreadsheets, screenshots and chronologies.

Why it fails

That evidence may prove disruption occurred, but without a logic model it rarely proves that the event drove Completion. Time becomes an argument rather than an analysis.

Fix

Recovery plan that works mid-project:

• stand up a logic-linked completion network (critical path + key gates)

• reconstruct an evidence-based as-built for the last 4–8 weeks on controlling workpacks

• start disciplined updates from now forward with archived issued packs

• keep Excel as the “front-end” for registers and dashboards, not the engine

  
  

Where Excel fits (and where it doesn’t)

Excel is excellent for:

• early warning register, interface register, change register

• cost and quantity trackers

• dashboards (Power BI-style outputs often start with Excel data)

Excel is risky as:

• the master programme that needs to demonstrate dependencies, critical path and cause-and-effect

If you keep Excel for what it’s good at and move the programme engine into a logic-linked schedule, you get the best of both: speed on site and defensibility when it matters.