Finance & Budgeting Free · self-study ~60 min

Post-SteerCo Cost Forecast Reconciliation

After a steering committee meeting, a senior leader requests a private discussion to verify the integrity of a revised cost forecast. The conversation focuses on validating governance controls, data versioning, and variance analysis before formal escalation.

 Level

What you’ll be able to do

Dialogue

Beginner version

Marcus Thorne
Hello Julian. Thank you for coming. You read the short report we sent yesterday, yes?
Julian Hayes
Yes, I read it. It was short and clear. Good. No big words. But now I want to know: did we really fix the problem? Or did we just put a small cover on it?
Marcus Thorne
I understand. You want to know if the fix is real. We can start with the numbers, the rules, or the systems. Where do you want to start?
Julian Hayes
Start with the numbers. Rules look good on paper. Show me the numbers are correct. Show me the data matches in all three systems. No hand work. If not, everything else is not important.
Marcus Thorne
OK. We did three checks last week. First, we set a time to lock the data. Brankside Holdings data locks two days after the period ends. Verolane Group locks at the same time. Caldmere Systems updates after that. Each update gets a number so we know which data we used. Second, we checked the cost codes. Third, we checked old numbers against new numbers and found the differences.
Julian Hayes
You say the data cannot change. But how do we stop someone from changing it after the lock?
Marcus Thorne
Each time we save the data, it gets a special number. The reports use that number. If we save new data, it gets a new number. The old data stays. So we can always see the old report again. Nobody can change the old data.
Julian Hayes
Good. That is important and many teams do not do it. Now tell me about the money difference. Give me numbers. No big words.
Marcus Thorne
The total cost went up by 4.8 million euros. 3.1 million is because we now share the indirect costs across the work. Before, those costs sat in one place and we did not see them in the forecast. 1.5 million is new costs added during the close. 0.2 million is small rounding. That is all.
Julian Hayes
OK. But the CFO will ask: the money was always there, so why did we not see it before? And did we make the forecast bigger to look safe?
Marcus Thorne
The money was always real. We just did not show it in the right place. We moved it, we did not add it. And we did not make it bigger on purpose. The rule is fixed. It is written down. We also tested it on old data from the last two quarters to check it works correctly.
Julian Hayes
You tested it on old data. Good. Did you find any problems?
Marcus Thorne
Yes, one. In one month, 600,000 euros moved between two areas because of a big change in staff. The plan did not show this change. So the plan was slow and the cost moved fast. We wrote this down as a problem to fix.
Julian Hayes
So we found another problem.
Marcus Thorne
Yes. But it is a different problem. The first problem was about where we put the costs. This new problem is about the plan being too slow. We can fix it. But someone needs to be responsible for it.
Julian Hayes
Let us talk about who is responsible. In big projects, everyone says the problem is in the middle. Then nobody fixes it. Last time you said you are responsible for the system. Is that still true?
Marcus Thorne
Yes. I made a clear list of who does what. My team owns the rules and the system checks. The cost team puts in the forecast numbers. Finance owns the Brankside Holdings data. Caldmere Systems builds the reports but follows our rules. This week we also made a new meeting every month to check the rules. Cost control runs it. My team and finance check and agree.
Julian Hayes
Who can say no to a change?
Marcus Thorne
My team and finance both must say yes. My team checks the rule is correct. Finance checks it matches the real money. The cost team can ask for changes but cannot change the rule alone.
Julian Hayes
Good. But what happens in real life? One day someone misses a deadline. Then people do their own thing. That is where rules break.
Marcus Thorne
Yes, so we made real controls. One: a calendar with clear deadlines and data freeze times. Two: a computer check that runs before every report. It checks for missing codes and errors. Three: the report cannot go out if the checks are not green. If someone needs to skip a check, they must ask for permission and write why.
Julian Hayes
Who gives that permission?
Marcus Thorne
The cost control leader or the finance controller. It depends on the problem. They must write what is missing, how big the effect is, and if the report is still safe to use.
Julian Hayes
Good. Last week you said you want to improve how we talk about problems. I saw your note. But talking about problems must be real, not just a document. What did you change?
Marcus Thorne
We made a list of important forecast ideas. Each idea has an owner and a limit. If something goes over the limit, we put it in the monthly report for leaders. For example: indirect costs growing fast, the plan being too slow, or contract prices changing.
Julian Hayes
Good idea. But will people really use it? Or will they only fill it in just before the meeting?
Marcus Thorne
We put it inside the forecast process. The forecast is not finished until someone checks the list and says it is OK or flags a problem. People must do it. It is not a choice.
Julian Hayes
Good. Now, what if someone outside asks us to explain every cost? Can someone explain it without calling you?
Marcus Thorne
Yes. We made three documents. One: a simple explanation of how we calculate the forecast. Two: a technical document about the rules in the systems. Three: a step-by-step guide to check the numbers from Brankside Holdings to Caldmere Systems. Both the cost team and Caldmere Systems can use the guide.
Julian Hayes
Did someone test the guide?
Marcus Thorne
Yes. A cost controller who did not work on the problem used the guide. He found the numbers in two hours. He did not call me. He had two small questions. Both were about missing words in the document. We fixed them.
Julian Hayes
Very good. But there is another risk. Sometimes after a problem, teams make too many rules. Then work is slow and people skip the rules. Did you think about this?
Marcus Thorne
Yes. We tried to keep the controls simple and automatic. The computer does the checks. The system saves the versions. The list is short because it only has the most important ideas. We also have a rule: we only add a control if it stops an error or helps us explain the data.
Julian Hayes
OK. Now a hard question. What is still not certain today? Say it without making it sound like a big problem.
Marcus Thorne
The numbers we have now are correct and we trust them. The questions we still have are about the future. For example, how fast will the indirect costs grow? How fast will the plan update? These questions are not hidden. We watch them and we will tell leaders if something changes.
Julian Hayes
Good answer. Now, what do you need from me? I do not want this to only live in your team.
Marcus Thorne
Two things. First, please make the monthly check meeting an official meeting with real power. Second, please tell cost control and finance that exception permissions must always be written down. If you say this, people will follow the rules.
Julian Hayes
I will do both. I will add the meeting to the calendar. I will tell cost control and finance it is a real rule. But I also want something from you. Tell me about problems early. Do not wait. No surprises in the big meetings.
Marcus Thorne
Agreed. We can also measure it. We can count how many problems we told you about early against how many we found late. It is simple but it shows if we are improving.
Julian Hayes
OK. But keep it simple. I do not want a big report about the report.
Marcus Thorne
Yes. We will keep it simple.
Julian Hayes
One more thing. You said the cost increase is inside the contingency. That is good. But it can make people think everything is fine and not be careful. How do we stop that?
Marcus Thorne
We treat contingency as real money, not free money. If we use any contingency, we must write why. Is it a one-time thing or will it happen again? What are we doing about it? The correction we made is not really using contingency. It is just moving costs. But we will explain it clearly so nobody thinks we have more money than we do.
Julian Hayes
Good. That is the right way. OK Marcus. One week ago I was thinking about making this a bigger issue. Today I think you are moving in the right direction.
Marcus Thorne
Thank you. I do not think the problem is closed. I think we are now working at a higher level. Big projects have problems. The important thing is to learn from them and make better rules.
Julian Hayes
Yes. Make it a rule for everyone. Please send me a one-page summary by Monday. Show the meeting plan, who does what, and the exception process. Keep it short and clear.
Marcus Thorne
I will send it. I will also add a short update on the guide and the list of forecast ideas. But it will be easy to read.
Julian Hayes
Good. We are done. Marcus, keep doing this. You can explain hard things in a simple way. That is a very useful skill.
Marcus Thorne
Thank you Julian. The work will show it.
Julian Hayes
Good. Let us move forward and be ready for the next problem.

Intermediate version

Marcus Thorne
Thanks for making time again, Julian. I assume this conversation follows on from the executive briefing and the updated report pack we circulated yesterday.
Julian Hayes
That's right. And to be clear: I did read it carefully. It was focused and avoided unnecessary technical detail, so well done on that. The real question now is whether we've built something that will genuinely hold, rather than a short-term solution that looks good on paper.
Marcus Thorne
Understood. So you're asking whether the fix is properly embedded, not just cosmetic. We could start with the numbers, the governance structure, or the evidence that the data aligns correctly across systems. Where would you like to begin?
Julian Hayes
Let's start with the evidence. Governance can look solid in a document but still fail in practice. I need you to show me we can track cost movement, explain it clearly, and reconcile it across all three systems without heavy manual effort. If we can't do that, everything else is secondary.
Marcus Thorne
Understood. Since last Friday we've completed three full reconciliation cycles. First, we set clear cut-off rules: Brankside Holdings data locks two days after period close, Verolane Group captures a snapshot at the same point, and Caldmere Systems refreshes afterwards with a fixed dataset reference. This removes timing mismatches. Second, we confirmed that cost codes and work structure are properly aligned, including the indirect engineering pool. Third, we produced a variance analysis showing the movement from the previous forecast to the current one, broken down by category.
Julian Hayes
You're using the word 'fixed' for the dataset. Are we actually controlling it, or just assuming nobody will change it after the refresh?
Marcus Thorne
We've introduced version control in the Caldmere Systems layer. Each refresh produces a tagged extract with a unique ID, and the dashboards reference a specific version rather than whatever is most recent. New refreshes create additional versions without overwriting the old ones. So if anyone asks what figures were shown at a particular meeting, we can reproduce exactly that view.
Julian Hayes
Good, that kind of discipline is fundamental and often missing. Now, regarding the actual cost movement: can you explain it clearly in about a minute, using figures rather than general descriptions?
Marcus Thorne
Of course. The total forecast increase across the affected scope is €4.8 million. Of that, €3.1 million comes from properly distributing indirect engineering costs across the relevant work packages, rather than leaving them pooled and underrepresented in the estimate to complete. €1.5 million reflects actual costs posted during the closing period. The remaining €0.2 million relates to minor rounding and timing adjustments, which are within acceptable tolerance.
Julian Hayes
That breakdown makes sense. But the CFO will likely ask two things: why weren't these costs visible before, and are we now inflating the forecast to create a safety margin? How would you respond to that?
Marcus Thorne
The key distinction is between a cost existing and it being visible at the right level. The money was always in the pool, we just weren't distributing it in a way that showed its full impact on individual work packages. We haven't added cost; we've improved how it's attributed. On the second point, the allocation rule is fixed, documented, and not open to discretion. We also back-tested it against two prior quarters to confirm it produces consistent and predictable results.
Julian Hayes
The back-testing is exactly what I was hoping to hear. Did it reveal anything that concerned you?
Marcus Thorne
One thing, though it turned out to be useful. In one month, the new rule would have shifted around €600k between two work packages because of a staffing surge that wasn't reflected in the planning system at the time. This highlighted a timing gap between resource planning and cost forecasting. We've recorded it as a separate improvement action, we need to make sure plan updates happen more regularly so they keep pace with actual cost behaviour.
Julian Hayes
So we've identified another underlying weakness.
Marcus Thorne
Yes, but it's a different type of issue. The original problem was about cost attribution. This new item is about the timing between planning and forecasting. It's manageable, but it needs a clear owner and an action plan.
Julian Hayes
Let's talk about ownership. One thing that concerns senior leaders is when a problem sits at the boundary between teams and no one takes responsibility. You said last time that PMIS accountability sits with you. Does that still stand?
Marcus Thorne
It does, and we've now made the responsibilities more explicit. My team owns the rules, system configuration, and integrity checks. Cost control manages the forecast inputs within that framework. Finance owns the Brankside Holdings actuals and the close process. Caldmere Systems handles the reporting layer, but under my team's governance for versioning and traceability. We've also formalised a monthly 'Forecast Integrity Review' this week, chaired by cost control, with my team as design authority and finance as the validation check.
Julian Hayes
Who has the authority to block a rule change?
Marcus Thorne
My team and finance both need to agree. My team checks that the rule is technically sound and workable; finance checks that it's compliant and consistent with the ledger. Cost control can propose changes, but can't implement them independently. That prevents any one area from adjusting the model to suit local preferences.
Julian Hayes
That's the right approach. But explain how it works in practice, because the first time someone misses a deadline, people will start finding workarounds, and that's usually where governance breaks down.
Marcus Thorne
Exactly, which is why we've focused on practical controls rather than just principles. We've introduced three specific measures: a formal cut-off calendar with clearly defined data freeze periods; an automated pre-publication checklist that verifies mapping, cost codes, and rule execution; and a publication gate so the reporting pack can't be released unless those checks pass, or unless an exception has been formally approved and recorded.
Julian Hayes
Who can approve an exception?
Marcus Thorne
Either the head of cost control or the finance controller, depending on the nature of the issue. Any exception must include a brief explanation of what's missing, the likely impact, and a judgement on whether the report is still reliable enough for decision-making.
Julian Hayes
That's the right process. Now, last week you mentioned wanting to improve how the organisation communicates uncertainty, not just the mechanics. Your note was clear, but changing behaviour is harder than writing a note. What has actually changed?
Marcus Thorne
We've set up an 'assumptions under stress' register. It's not a general risk log, it's a focused list of forecast-relevant assumptions, each with an owner, a threshold, and a trigger for escalation. Examples include indirect pools growing faster than expected, planning updates lagging behind reality, and changes to contract rates. When a trigger is reached, it automatically feeds into the monthly executive narrative as an emerging cost driver.
Julian Hayes
That's a real step forward in terms of maturity. But I'll ask the difficult question: are you confident the team will actually use it consistently, or might it become something that only gets updated just before a meeting?
Marcus Thorne
We're not relying on good intentions. The register review is a required step before the forecast can be signed off. It's built into the process, so people have to engage with it, and that means having a real conversation about assumptions, not just confirming numbers.
Julian Hayes
Good. Let's consider external scrutiny. Suppose we receive an audit request, or a sponsor asks for a detailed justification of cost movements. Can the team handle that without you personally being involved in every discussion?
Marcus Thorne
That's an important point. The aim is to make the system self-explanatory. We've produced three specific documents: a plain-language overview of how the forecast is built, including EAC and ETC logic and allocation principles; a technical specification covering how the rules are implemented in Verolane Group and Caldmere Systems, with version control details; and a step-by-step reconciliation guide tracing figures from Brankside Holdings through to Caldmere Systems output, with checkpoints and screenshots. Both cost control and Caldmere Systems are responsible for running through that guide.
Julian Hayes
Has anyone tested it?
Marcus Thorne
Yes, we gave it to a cost controller who had no involvement in the incident, to simulate what an independent reviewer would experience. He was able to reproduce the full cost bridge in around two hours without needing to contact me. He did raise two questions, but both were about gaps in the wording, not errors in the model. We've updated those sections.
Julian Hayes
Excellent. Now, a different kind of risk: sometimes after an incident, teams add so many controls that work slows down and people start finding ways around them. Have you thought carefully about keeping the right balance?
Marcus Thorne
Yes, that was a deliberate consideration in the design. Where possible, controls are automated, the checklist runs as a script, and version tagging is system-driven. The assumptions register is kept brief because it only covers forecast-critical items. We also applied a clear principle: a control only stays if it either prevents a category of error, helps detect one early, or supports audit transparency.
Julian Hayes
All right. Let me ask you directly: if I asked what remains uncertain right now, what would you say, in a way that sounds measured rather than concerning?
Marcus Thorne
I'd say the current forecast is reliable within its defined parameters, and we've strengthened those parameters significantly. The remaining uncertainties are forward-looking, how indirect costs will behave as the project ramps up, and how quickly planning will respond to operational changes. These aren't hidden risks; they're being monitored, they have owners, and there are clear triggers for escalating them to leadership.
Julian Hayes
That's a well-framed answer. Now tell me what you need from me, because I don't want this improvement to stay confined to the project controls team.
Marcus Thorne
Two things. First, I need you to formally establish the Forecast Integrity Review as a standing governance meeting with real decision-making authority, not just an informal working session. Second, I need you to make it clear to cost control and finance that exception approvals must always be documented, even when people are under time pressure. If that expectation comes from you, it's much more likely to stick.
Julian Hayes
You'll have both. I'll get the meeting added to the governance calendar and brief cost control and finance that this is now a formal control requirement. In return, I want you to flag any emerging forecast issues to me before they become surprises. No last-minute escalations.
Marcus Thorne
Agreed. To make that measurable, we could track the ratio of issues flagged proactively versus ones that surface reactively. It's a simple indicator, but it reinforces the behaviour we're aiming for.
Julian Hayes
That's fine, as long as it stays meaningful. I'm interested in the outcome, not a metric for its own sake.
Marcus Thorne
Understood. We'll keep it straightforward and tied to real decisions.
Julian Hayes
One final point: you noted that the adjustment sits within contingency, which is reassuring. But I've seen teams use that as a reason not to worry. How do we make sure contingency is treated with proper discipline?
Marcus Thorne
By treating it as a managed resource rather than a buffer. We've added a rule that any draw on contingency must be accompanied by a brief explanation: what drove it, whether it's a one-off or a recurring pattern, and what mitigation is in place. The current adjustment is technically a reattribution rather than a draw, but it affects how much headroom appears to exist, so we'll report it transparently to avoid any false sense of comfort.
Julian Hayes
Good. That's the level of rigour I'm looking for. All right, Marcus. A week ago I was weighing up whether to escalate this. Today I'm assessing whether to trust the system you've put in place. You've made a strong case that we're heading in the right direction.
Marcus Thorne
Thank you. I genuinely see this not as closing a chapter but as raising the standard for the whole programme. Incidents like this are difficult, but they're also an opportunity, if we turn them into permanent improvements rather than one-off fixes.
Julian Hayes
Exactly, turn it into a standard. Send me a one-page summary by Monday covering the governance cadence, responsibilities, gates, and the exception process. Same tone as your last note, keep it executive-ready.
Marcus Thorne
You'll have it. I'll also include a brief status update on the reconciliation guide and the assumptions register, but I'll make sure it stays concise and readable.
Julian Hayes
Good. Then we're finished here. And Marcus, keep working at this level. Someone who can make complex systems clear to decision-makers is doing more than project controls work.
Marcus Thorne
That means a lot, Julian. I'll make sure the results back it up.
Julian Hayes
Good. Let's move forward, and make sure we stay ahead of whatever comes next.

Advanced version

Marcus Thorne
Thank you for making the time again, Julian. I take it this follows the executive briefing and the revised reporting pack circulated yesterday.
Julian Hayes
Precisely. And to preempt your query: yes, I reviewed it thoroughly. It’s concise, decision-focused, and sidestepped the common pitfall of overwhelming readers with technical jargon. So, fair play. The real challenge now is convincing me that we’ve transformed this incident into a lasting control mechanism, rather than a superficial band-aid.
Marcus Thorne
I understand. If you’re assessing whether the solution is embedded in our processes rather than merely decorative, I’m on the same page. Where would you prefer to begin: governance architecture, operational uptake, or proof that the figures now align seamlessly across the board?
Julian Hayes
Proof first. Governance can look impeccable on paper. Show me we can track the EAC movement, articulate it clearly, and reconcile it from Brankside Holdings actuals through to Verolane Group and Caldmere Systems-without relying on strenuous manual workarounds. If we can’t, the rest is mere showmanship.
Marcus Thorne
Fair enough. Since last Friday, we’ve executed three reconciliation cycles. First, we established cut-off and versioning protocols: Brankside Holdings actuals locked at close-of-business on T+2 post-period close, Verolane Group snapshot at the identical timestamp, and Caldmere Systems refresh scheduled post-snapshot with an immutable dataset ID. This eradicates timing discrepancies. Second, we verified mapping integrity: cost codes and WBS alignment, encompassing the indirect engineering pool. Third, we conducted a variance bridge: prior EAC to current EAC, broken down into allocation logic adjustments, new actuals, and any scope-approved modifications.
Julian Hayes
You’re employing the term “immutable.” Are we genuinely controlling the dataset, or merely hoping no one alters it post-refresh?
Marcus Thorne
We’ve instituted controlled versioning within the Caldmere Systems layer. Each refresh generates a tagged extract bearing a unique version ID; the steerco dashboards reference a specific version, not the “latest.” Subsequent refreshes create new versions without overwriting previous ones. Thus, when leadership asks, “What did you present on Monday?” we can replicate the exact view.
Julian Hayes
Excellent. That’s fundamental governance hygiene, and it’s surprising how frequently it’s overlooked. Now, regarding the actual deviation: can you explain it in one minute, with figures, not adjectives?
Marcus Thorne
Certainly. The net EAC increase for the affected perimeter stands at €4.8 million. Of this, €3.1 million stems from formalising the allocation logic for indirect engineering pools-essentially distributing pooled costs across consuming work packages instead of leaving them aggregated and under-represented in ETC. €1.5 million reflects the natural impact of new actuals hitting the ledger during the closing cycle. The remaining €0.2 million comprises minor rounding and re-phasing effects driven by updated consumption curves, still within tolerance.
Julian Hayes
That decomposition is structurally sound. Now, the question I’ll face from the CFO’s side is: “If it was always real cost, why didn’t we see it earlier?” And “Are we now inflating the forecast to appear conservative?” How do you counter that?
Marcus Thorne
By distinguishing visibility from existence. The cost resided in the pool; it simply wasn’t allocated in a manner that made its downstream impact visible at the control account level. We’re not fabricating cost; we’re enhancing attribution. Regarding your second point: we’ve established guardrails. The allocation rule is non-discretionary. It’s parameterised, documented, and validated against historical periods. We also conducted a back-test across the last two quarters: applying the new rule to historical actuals to confirm the model behaves predictably and avoids artificial volatility.
Julian Hayes
Back-testing is precisely what I wanted to hear. Did it yield any unsettling findings?
Marcus Thorne
One, but it proved valuable. In one month, the allocation would have shifted roughly €600k across two control accounts due to a resourcing surge not captured in the planning system at the time. This exposed a planning-to-cost interface weakness: the cost pool reacted faster than the plan. We’ve logged it as a distinct improvement item: tightening the cadence between resource planning updates and cost forecasting assumptions.
Julian Hayes
So we’ve uncovered another latent fragility.
Marcus Thorne
Yes, but it’s a different category of issue. The original incident concerned model attribution. This follow-up item addresses cadence alignment-ensuring the plan isn’t stale relative to actual cost behaviour. It’s manageable but requires clear ownership.
Julian Hayes
Let’s discuss that. One factor that unnerves senior leaders is “intersection ownership.” Everyone agrees it’s at the intersection-then nobody claims it. You stated last time PMIS design accountability rests with you. Is that still your stance?
Marcus Thorne
It is, with a more defined RACI. PMIS owns the rules, configuration, and integrity checks across systems. Cost control owns the operational forecast inputs within that governed framework. Finance owns the Brankside Holdings actuals and close discipline. Caldmere Systems owns the reporting layer implementation, but under PMIS governance for versioning and traceability controls. The aspect we formalised this week is the governance forum that ratifies the rules: a monthly “Forecast Integrity Review” chaired by cost control, with PMIS as design authority and finance as validation authority.
Julian Hayes
Who holds veto power?
Marcus Thorne
For rule changes: PMIS and finance jointly. PMIS ensures the rule is coherent and implementable; finance ensures it’s compliant and reconciles with ledger reality. Cost control recommends but doesn’t unilaterally alter the model. That prevents local optimisation.
Julian Hayes
Good. Now, explain how this plays out in practice. Because the first time someone misses a cut-off, people will start improvising. And improvisation is where governance goes to die.
Marcus Thorne
Agreed. That’s why we engineered operational controls, not just conceptual ones. We’ve introduced three practical measures. One: a formal cut-off calendar with explicit “data freeze” windows. Two: an automated completeness checklist that runs before any forecast is published-mapping checks, missing cost code flags, and allocation rule execution logs. Three: publication gating. The steerco pack can’t refresh unless those checks are green, unless an exception is approved and logged.
Julian Hayes
Exception-approved by whom?
Marcus Thorne
By the Forecast Integrity Review delegate for that cycle-typically the head of cost control or finance controller-depending on the nature of the exception. And the exception must include a concise impact statement: what’s missing, what’s the expected effect, and whether the view is safe for decision-making.
Julian Hayes
That’s the correct pattern. Now, one week ago, you mentioned wanting to improve narrative control, not just mechanics. I saw your note, which was clear. But narrative control isn’t a note. It’s behaviour. What have you changed in how the organisation discusses uncertainty?
Marcus Thorne
We’ve implemented an “assumptions under stress” register. It’s not a generic risk log; it’s a list of forecast-relevant assumptions with leading indicators. For instance: indirect pools scaling non-linearly, resource plan lag, and contract rate renegotiations. Each assumption has an owner, a threshold, and a communication trigger. If the trigger is hit, it appears in the monthly executive narrative as an emerging forecast driver.
Julian Hayes
That’s precisely the maturity I seek. But I’ll pose the uncomfortable question: are you confident the team will actually use it, or will it become another document updated five minutes before a meeting?
Marcus Thorne
I’m not relying on goodwill. We’ve embedded it into the forecast cycle. The forecast can’t be “signed off” unless the register is reviewed and either confirmed stable or flagged. It’s a mandatory step, forcing a conversation about meaning, not just numbers.
Julian Hayes
Good. Let’s shift to external scrutiny. Assume we receive an audit request-or, more likely, a sponsor from another country asking us to justify cost movements line by line. Can we defend the model without bringing you personally into every conversation?
Marcus Thorne
That’s a critical point. The objective is to make the system defensible without single-point dependency. We’ve produced three artefacts specifically for that: first, a plain-English “forecast logic overview” explaining EAC/ETC composition, allocation principles, and cut-offs. Second, a technical specification describing rule implementation in Verolane Group and Caldmere Systems, including version control. Third, a reconciliation playbook: step-by-step trace from Brankside Holdings actuals to Verolane Group cost codes to Caldmere Systems aggregation, with screenshots and checkpoints. Cost control and Caldmere Systems both own running the playbook.
Julian Hayes
Who tested it?
Marcus Thorne
We ran a dry run with a cost controller uninvolved in the incident, specifically to simulate fresh eyes. He followed the playbook and reproduced the EAC bridge within two hours, without escalating to me. He raised two clarifications-both were documentation gaps, not model gaps. We corrected them.
Julian Hayes
That’s excellent. Now, another risk: over-correcting. Sometimes teams respond to an incident by building so many controls that delivery slows down and people start bypassing them. Have you considered the balance between control and agility?
Marcus Thorne
Yes. We explicitly designed controls to be lightweight and automated where possible. The completeness checks are scripted. The version tagging is system-driven. The register review is brief because it’s focused-only forecast-critical assumptions, not every operational concern. And we’ve set a principle: no control step should exist unless it either prevents a class of error, detects it early, or provides audit defensibility.
Julian Hayes
All right. Let me put you on the spot. If I ask, “What remains uncertain today?” what do you say-without sounding like we’re out of control?
Marcus Thorne
I’d say: the core forecast is reliable within defined parameters, and we’ve strengthened those parameters. The remaining uncertainty isn’t about the correctness of actuals or baseline integrity; it’s about forward-looking behaviour-how indirect pools will respond to ramp-up, and how quickly planning will reflect operational reality. Those uncertainties aren’t hidden; they’re monitored, owned, and have triggers for executive communication.
Julian Hayes
That’s a strong answer. Now, tell me what you need from me. Because I don’t want this to live and die at the PMIS layer.
Marcus Thorne
Two things. First, sponsorship to formalise the Forecast Integrity Review as a standing governance cadence with decision authority, not an optional working group. Second, clarity that exception approvals must be documented and not treated as “quick favours,” even under schedule pressure. If that tone comes from you, adoption becomes much easier.
Julian Hayes
You’ll get both. I’ll have my office add the cadence to the governance calendar, and I’ll brief cost control and finance that this is now a formal control gate. But I want something in return: I want you to proactively surface any assumption under stress before it becomes a steering committee surprise. No heroics, no last-minute fixes.
Marcus Thorne
Agreed. And to make it measurable, we can add a KPI: number of forecast drivers flagged proactively versus discovered reactively. It’s not perfect, but it will reinforce the behaviour we want.
Julian Hayes
Fine-just don’t turn it into theatre. I care about substance, not vanity metrics.
Marcus Thorne
Understood. We’ll keep it simple and meaningful.
Julian Hayes
One last point: in your note, you said the adjustment remains within contingency. That’s reassuring, but it can also breed complacency. How do we ensure contingency isn’t used as a comfort blanket?
Marcus Thorne
By treating contingency as a governed resource, not a cushion. We’ve added a rule that any draw on contingency must be accompanied by a driver narrative: what triggered it, whether it’s one-off or trend, and what mitigation exists. The allocation correction isn’t a contingency draw per se-it’s attribution-but it impacts perceived headroom. So we’ll report it transparently to avoid false comfort.
Julian Hayes
Good. That’s the level of discipline I expect. All right, Marcus. One week ago, I was deciding whether to escalate. Today, I’m deciding whether to trust the system. You’ve made a compelling case that we’re moving in the right direction.
Marcus Thorne
Thank you. And for what it’s worth, I don’t see this as “closing an issue.” I see it as elevating the program’s maturity level. If we’re building megaproject-grade controls, incidents like this are painful but informative-provided we convert them into standards.
Julian Hayes
Exactly. Make it a standard. And send me, by Monday, a one-page summary of the new governance cadence-RACI, gates, and the exception process. Same style as your last note. Keep it executive.
Marcus Thorne
You’ll have it. And I’ll include the reconciliation playbook status and the assumptions register triggers, but in a way that stays readable.
Julian Hayes
Good. Then we’re done. And Marcus-keep operating like this. If you can make complex systems legible to decision-makers, you’re not just a PMIS specialist. You’re a control leader.
Marcus Thorne
I appreciate that, Julian. I’ll ensure the work proves it, not the words.
Julian Hayes
Perfect. Let’s move on-and stay ahead of the next one.

Check your understanding

1. What is the 'immutable' dataset control mechanism Marcus Thorne describes for Caldmere Systems?

Show answer
Each refresh generates a tagged extract with a unique version ID, and steerco dashboards reference a specific version rather than the 'latest.' Subsequent refreshes create new versions without overwriting previous ones, allowing leadership to replicate the exact view presented on a specific day.

2. How does Marcus Thorne distinguish 'visibility' from 'existence' to answer Julian Hayes' question about why the cost wasn't seen earlier?

Show answer
He explains that the cost resided in the pool but wasn't allocated in a way that made its downstream impact visible at the control account level. They are not fabricating cost but enhancing attribution.

Grammar practice (mixed)

Tenses

Marcus noted that the revised allocation logic __________ predictable results during the back-testing phase.

Show answer & why
yielded · 💡 The sentence refers to a completed action in the past (the back-testing phase), requiring the simple past tense.
Prepositionsself-check

The reconciliation process ensured traceability __________ Brankside Holdings actuals and Verolane Group data.

Show answer & why
between · 💡 The preposition 'between' is used to indicate a relationship involving two distinct items (Brankside Holdings actuals and Verolane Group data).
Verb forms

Without clear ownership, forecast integrity issues __________ easily reappear at scale.

Show answer & why
could · 💡 The modal 'could' expresses possibility or potential risk, which fits the context of a warning about what might happen if controls are weak.
Linking Words / Phrases / Conjunctive Expressions

The allocation logic was corrected; __________, additional governance controls were introduced to prevent recurrence.

Show answer & why
furthermore · 💡 The second clause adds information that supports or extends the first clause (correcting logic AND adding controls), requiring an additive linker.

Discussion (practise speaking)

How can an organization ensure that new governance rules are adopted by staff rather than being bypassed during high-pressure periods?

🤔 Think about a time when a new rule was ignored in your workplace and what could have prevented it.

Show sample answer
  • Implement mandatory steps in the workflow so rules cannot be skipped.
  • Communicate the 'why' behind the rules to build buy-in.
  • Use automated checks to reduce the burden on manual compliance.

Ask Phil: Practise explaining how to enforce compliance without slowing down delivery with the Pickle AI tutor.

What are the risks of relying on a single person to explain complex financial models to external auditors?

🤔 Consider how you would handle a situation where you are the only expert on a critical system.

Show sample answer
  • It creates a bottleneck if that person is unavailable.
  • The explanation may become inconsistent if not standardized.
  • It increases risk if the person leaves the company.

Ask Phil: Practise describing how to make a technical system defensible without single-point dependency with the Pickle AI tutor.

How should a leader communicate remaining uncertainties to stakeholders without causing alarm or loss of confidence?

🤔 Think about how you would present a difficult update to your manager or client.

Show sample answer
  • Acknowledge the uncertainty but show it is monitored and owned.
  • Provide clear triggers for when issues will escalate.
  • Focus on the controls in place rather than just the problems.

Ask Phil: Practise delivering a status update that balances transparency with confidence with the Pickle AI tutor.

Why is it important to distinguish between 'visibility' and 'existence' when discussing cost allocation?

🤔 Reflect on a time when a problem was hidden by poor reporting and how you would have exposed it.

Show sample answer
  • It clarifies that costs were always there, just not tracked correctly.
  • It prevents accusations of fabricating data or inflating forecasts.
  • It highlights the need for better attribution rather than new spending.

Ask Phil: Practise explaining the difference between visibility and existence in a business context with the Pickle AI tutor.

Vocabulary

decision-focused
reveal definition Concentrating on supporting choices rather than providing excessive detail. “It’s concise, decision-focused, and sidestepped the common pitfall of overwhelming readers with technical jargon.”
↗
control mechanism
reveal definition A systematic method used to manage and regulate processes. “The real challenge now is convincing me that we’ve transformed this incident into a lasting control mechanism, rather than a superficial band-aid.”
↗
governance architecture
reveal definition The structural framework that defines how rules and oversight are applied. “Where would you prefer to begin: governance architecture, operational uptake, or proof that the figures now align seamlessly across the board?”
↗
reconciliation cycles
reveal definition Repeating processes to verify that different sets of data match correctly. “Since last Friday, we’ve executed three reconciliation cycles.”
↗
variance bridge
reveal definition A breakdown explaining the difference between two financial figures. “Third, we conducted a variance bridge: prior EAC to current EAC, broken down into allocation logic adjustments, new actuals, and any scope-approved modifications.”
↗
immutable dataset
reveal definition A set of data that cannot be changed after it is created. “Caldmere Systems refresh scheduled post-snapshot with an immutable dataset ID.”
↗
RACI
reveal definition A matrix defining who is Responsible, Accountable, Consulted, and Informed. “It is, with a more defined RACI.”
↗
cut-off calendar
reveal definition A schedule that sets specific deadlines for data submission. “One: a formal cut-off calendar with explicit “data freeze” windows.”
↗
assumptions under stress
reveal definition A register tracking forecast assumptions that are at risk of failing. “We’ve implemented an “assumptions under stress” register.”
↗

Key phrases (useful expressions from the dialogue)

More Finance & Budgeting lessons

Annual Financial Review and Strategic Planning

Elena, the Financial Controlling Manager, conducts a virtual annual financial review with the CFO to analyze t…

Derivative Trading Risk and Compliance Review

An auditor meets with a trading department head to review derivative strategies, risk management, and regulato…

Explaining Budget Variances and Corrective Actions

A management controller presents a detailed budget variance report to a regional operations manager, discussin…

Negotiating EBITDA Adjustments in M&A Due Diligence

Alex Vance reviews Caldmere Systems's financial data and raises concerns about revenue sustainability, invento…

Post-SteerCo Cost Forecast Variance Review

After a steering committee flagged a cost forecast discrepancy, the Program Director requests a focused one-on…

Q3 Financial Results and Cost Variance Analysis

A finance manager presents Q3 results to the executive committee, highlighting revenue growth and profit margi…

Keep going

Practise this scenario live with Phil, and get a personalised lesson built from your own mistakes. Interactive in the browser, with a booklet to keep.

Book a lesson with Phil