Governance and monitoring

Why this matters

The best plan drifts once real work starts. The question executives ask is not “are you following the plan?” but “how far ahead or behind are you, and where will you end up if this continues?” Saying “we have spent half the money” means nothing without knowing how much work has been done. This module gives you the numbers to answer that, and the mechanisms that let a project stop in time when it should.

Earned value management

Earned value management (EVM) compares three values at the same point in time, as set out in the cost-estimating guide of the US Government Accountability Office (GAO):

• PV (planned value): the value of work the plan says should be done by now
• EV (earned value): the value of work actually done, at its budgeted cost
• AC (actual cost): the money actually spent
• BAC (budget at completion): the total project budget

bac = 1_200_000
pv = bac * 6 / 12
ev = bac * 0.45
ac = 620_000
cv, sv = ev - ac, ev - pv
cpi, spi = ev / ac, ev / pv
eac = bac / cpi
print(f"PV {pv:,.0f}  EV {ev:,.0f}  AC {ac:,.0f}")
print(f"CV {cv:+,.0f}  SV {sv:+,.0f}  CPI {cpi:.3f}  SPI {spi:.3f}")
print(f"EAC {eac:,.0f}   VAC {bac - eac:+,.0f}")

PV 600,000  EV 540,000  AC 620,000
CV -80,000  SV -60,000  CPI 0.871  SPI 0.900
EAC 1,377,778   VAC -177,778

KPIs you can decide with

A key performance indicator (KPI) needs a definition, a denominator, a data source, an owner and a review frequency. An example from the drone knowledge hub’s unit on project governance and monitoring: a survey project delivered 20 sets, 15 passed the first check, 18 were on time, and 2 are still waiting for coordinate confirmation.

delivered, first_pass, on_time, pending = 20, 15, 18, 2
print(f"first-pass acceptance {first_pass / delivered:.0%}")
print(f"on-time delivery      {on_time / delivered:.0%}")
print(f"pending confirmation  {pending} sets (not counted as passed)")

first-pass acceptance 75%
on-time delivery      90%
pending confirmation  2 sets (not counted as passed)

On-time delivery is 90% but first-pass acceptance only 75%: two different dimensions. Watching only timeliness hides that one job in four needs rework, and pending items must be counted separately, not as passed. More flights do not mean the users’ problem is being solved better either. KPIs must measure the outcome users need.

Risks and stakeholders

ISO 31000:2018 sets the risk-management process as identifying, analysing, evaluating and treating risks, with continuous communication and monitoring. A risk register records each risk, separating cause, event and impact, as in UAT 313 module 3.

risks = [
    ("R1", "permit delayed", 4, 4, "file early, assign an owner, weekly follow-up"),
    ("R2", "drone crash during pilot service", 2, 5, "OM, checklists, insurance, geofence"),
    ("R3", "mapping software price rises", 3, 2, "compare two vendors, budget reserve"),
]
for rid, event, likelihood, impact, response in sorted(risks, key=lambda r: r[2] * r[3], reverse=True):
    score = likelihood * impact
    level = "high" if score >= 15 else "medium" if score >= 8 else "low"
    print(f"{rid} score {score:>2} ({level:<6}) {event:<32} -> {response}")

R1 score 16 (high  ) permit delayed                   -> file early, assign an owner, weekly follow-up
R2 score 10 (medium) drone crash during pilot service -> OM, checklists, insurance, geofence
R3 score  6 (low   ) mapping software price rises     -> compare two vendors, budget reserve

Scores help rank risks but are not the whole story. R2 scores 10, lower than R1, but has the highest impact (people could be hurt), so it needs controls even though it is unlikely. The level bands here are practice thresholds; each organisation must set its own.

Stakeholders of the service unit include the mayor, communities being flown over, the Civil Aviation Authority of Thailand (CAAT), equipment vendors and the pilot team. A common tool is the power–interest grid: high-power, high-interest groups need close management, while low-power groups that are affected, such as communities, need regular information.

Change control and stop conditions

Every change to scope, time or budget must go through change control: log the request, assess its impact, have someone with authority approve it, then update the plan. Accepting small additions bit by bit without the process is scope creep, a leading cause of cost overruns.

A good project sets stop or review conditions from the start, such as “if CPI is below 0.85 for two reviews in a row, review the scope” or “if first-pass acceptance in the pilot is below 70%, stop expanding”. Stopping an ineffective project early is a good decision, not a failure.

Class activity

Common mistakes

Summary

• EVM compares PV, EV and AC to give CV, SV, CPI and SPI, and forecasts the final cost with EAC = BAC/CPI
• KPIs need a definition, denominator, data source and review frequency, and must measure what users need
• The risk register separates cause, event and impact, ranked by likelihood × impact plus judgement
• Control every change, and set stop or review conditions at the start of the project

Check your understanding

1. With EV = 300,000 baht and AC = 250,000 baht, what are CPI and CV?
2. With EV = 400,000 baht and PV = 500,000 baht, is the project ahead of or behind schedule?
3. With BAC = 2,000,000 baht and CPI = 0.8, what is EAC?
4. A risk has likelihood 3 and impact 5. What is its score?
5. Of 40 sets delivered, 30 pass the first check. What is the first-pass acceptance rate?