The Decision Table: Where Did the Saving Go?

The Decision Table: Where Did the Saving Go? Operational ITAM Podcast | Episode 016 | Bill Van Nort

Hey everybody, and welcome back to the Operational ITAM Podcast. I'm Bill Van Nort, and today we're following a saving that made it into the presentation before it made it into the accounts.

The licenses have been cleaned up. The change ticket is closed. The project report says the work is complete. Then finance asks why the supplier is still charging the old amount.

Nobody thinks that is their part of the project anymore. Fortunately, the invoice has brought everyone back together.

Today, The Decision Table: Where Did the Saving Go?

Good morning, good afternoon, or good evening, wherever you're listening from. This is the show where we take the unglamorous machinery of enterprise technology and make it make sense. Grab your coffee.

In episode fifteen, we established who can make the renewal decision. We gave that person evidence, options, and conditions they could actually approve. Today we start after the approval. We are going to follow one change until we can explain the result.

The FinOps Foundation's Reporting and Analytics guidance calls for comparing actual spend with the estimate behind a decision. That's a useful starting point. The practical challenge is explaining the distance between them without changing the original estimate every time something goes wrong.

My opinion, clearly labeled: the original business case should stay in the file. Update the forecast, absolutely. Keep the earlier version so the organization can learn from what changed. Otherwise, every project eventually achieves exactly the number somebody last typed into it.

Let's use a fictional company and an invented software agreement. These are illustrative U.S. dollar amounts, not vendor prices or a client result. We'll follow a calendar year from January through December.

The company pays for a thousand subscription seats at twenty dollars per seat per month. Twenty thousand dollars a month. Two hundred forty thousand dollars for a full year.

An assignment review finds that eight hundred seats can meet the continuing business requirement. The proposed reduction is two hundred seats, starting in January. At the same unit price, that would remove four thousand dollars a month, or forty-eight thousand dollars across the year.

That is our original gross forecast. Gross means before the cost of making the change. It is also conditional: the quantity must be commercially reducible, the change must take effect in January, and the remaining service must still meet the requirement.

Before we follow the money, establish what we're comparing. For this case, finance agrees that keeping the existing service at a thousand seats and the same twenty-dollar rate is a supported alternative for the year. We have the current invoices and an available unchanged renewal to support it.

The business requirement stays the same. The reduction removes unnecessary assignments, not a department that has closed. No price increase, tax change, currency movement, or service downgrade is hiding in the comparison. Those assumptions keep this example readable. In your own records, each needs checking.

That agreed comparison is the baseline. It tells us what the result is being measured against. Last year's payment, this year's budget, a supplier's opening quote, and a forecast of future demand are different baselines. The same invoice can look favorable against one and unfavorable against another.

If demand really changes, explain it separately. Perhaps the company serves more customers or acquires another division. Keep the approved comparison, then show an adjusted view with the new scope and its evidence. Don't quietly rewrite the starting point and call the difference performance.

You also need the period. A monthly reduction multiplied by twelve describes a full year at that rate. It doesn't establish that twelve months of benefit occurred. We'll see that distinction matter almost immediately.

An easy place to lose that discipline is the handoff between teams. The person finding unused assignments may estimate an opportunity. Procurement may record a negotiated position. Delivery may mark an action complete. Finance may report a recognized result. Those are useful milestones, but they need their own dates and evidence. If all four are labeled saved, the report stops telling you where the work actually stands.

Keep the original estimate alongside the latest forecast and the verified result to date. If the original estimate was wrong, leave a short explanation. If it was reasonable but circumstances changed, document the change. You are trying to improve the next decision, not arrange the columns so nobody has to discuss this one.

The team finishes the cleanup later than planned. January remains at a thousand paid seats. The signed change takes effect on February first.

There is another difference. The supplier agrees to reduce the commitment only to eight hundred fifty seats. That is the minimum in this fictional amendment. It is not a statement about a particular publisher's rules.

The authorized owner accepts that option, and the team records eight hundred assigned seats against eight hundred fifty purchased seats. There are fifty unassigned seats still being paid for. They are available capacity, not another saving already achieved.

Now we can explain the revised forecast. We lost the planned four-thousand-dollar reduction in January. For the remaining eleven months, those extra fifty paid seats cost a thousand dollars a month above the original target. Another eleven thousand dollars of the original forecast will not happen this year.

Forty-eight thousand, less four thousand for timing, less eleven thousand for the retained commitment. That leaves thirty-three thousand dollars of expected subscription reduction for the calendar year.

Here is a simpler way to check it. The monthly bill should fall from twenty thousand to seventeen thousand in February. Three thousand dollars less, for eleven months. Thirty-three thousand dollars.

The operational cleanup can be complete while the commercial result is smaller than first proposed. Both statements belong in the report. Calling the work a failure would ignore the reduction. Keeping forty-eight thousand in the forecast would ignore the agreement.

This is where you connect the evidence. Keep the approved proposal, the signed amendment, the effective date, and the assignment records together. Each answers a different question. What did we intend? What did the supplier agree to? When did that obligation change? What did the team actually implement?

Give the change a stable reference that can appear in the benefit record and the billing investigation. It doesn't need a new platform. A reference in the existing renewal record can be enough if people can find the supporting documents.

Microsoft's guidance on buying or removing business subscription licenses makes a useful distinction here. Unassigning a license from a user and removing a purchased license are separate steps. Removal timing depends on the billing arrangement and the applicable window. Check the actual subscription and agreement before predicting when the charge will fall.

That is a real product mechanism, separate from our invented eight-hundred-fifty-seat minimum. A screenshot showing fewer assignments proves something about assignments. It does not, by itself, prove a lower payable quantity.

And before removing access, verify the continuing service, data, and retention requirements with the responsible people. A cheaper bill is not a successful outcome if the change prevents authorized staff from doing required work.

Our fictional company also pays an outside specialist six thousand dollars to complete the cleanup. For this example, that is the only incremental implementation cost, it is incurred and paid during the year, and finance includes it in the benefit comparison.

Thirty-three thousand dollars of subscription reduction, less six thousand to implement it, gives twenty-seven thousand dollars of expected net benefit for the year.

Internal staff also spend time on the change. Record that effort. In this case it fits within existing capacity, with no extra payroll or displaced funded work identified, so we are not inventing an additional cash payment. If it displaced important work, say what was displaced and assess it. Paid invoices are not the only possible cost of a decision.

The completion check also needs more than a ticket status. Have the service owner confirm that the right assignments were removed, the required people retained access, and the purchased quantity matches the amendment. Preserve a dated record from the system that actually controls the assignments. If an automated rule can put those assignments back tomorrow, identify who owns that rule before closing the work.

Keep the check proportionate to the service. You don't need to retest an entire application because a dormant account was removed. You do need to know that the removal happened and that the reason for calling it unnecessary was sound. Where the evidence is only a request to make a change, the implementation remains unverified.

The counterargument is fair: this sounds like a lot of checking for a modest reduction. The effort should be proportionate. A small, straightforward change may need a few linked records and a short review. But somebody still needs to establish the effective date, the actual quantity, and the cost of getting there. The arithmetic gets shorter when the facts are simple.

At this point we have a revised forecast. We have not yet verified a full year's result. Now we need the invoices, and that is where our case becomes more interesting.

Let's take a quick break. If you want a practical procedure for this work, the Operational ITAM Store has one called Validate Benefits and Report Business Outcomes. It's procedure F02.

It includes an editable HTML procedure, an SVG flowchart, and a local adoption and evidence checklist. The starting inputs include the agreed baseline, approved action, invoices or quotes, implementation cost, currency, period, and business outcome evidence.

That gives you a structured place to start the conversation with finance. Adapt the responsibilities and measurement decisions to your organization. The procedure doesn't decide what your finance team will recognize, and it doesn't replace the records behind the number.

You can review the contents at operationalitam.com/store. This is my paid resource, and buying it supports the show. Today's homework uses information you already have and doesn't require a purchase.

You'll also find the podcast and practical resources at operationalitam.com. If someone keeps asking you where the saving went, this might be a useful episode to share with them.

Alright. Back to the invoices.

January is correctly billed at twenty thousand dollars. But February and March are also billed at twenty thousand, even though the signed amendment requires seventeen thousand from February onward. April and May arrive at the correct seventeen thousand.

At the end of May, those five invoices total ninety-four thousand dollars. Our unchanged baseline for five months is a hundred thousand. The invoices currently show a six-thousand-dollar reduction.

The agreement supports a different figure. January at twenty thousand, then four months at seventeen thousand, totals eighty-eight thousand. Compared with the baseline, that should be a twelve-thousand-dollar reduction through May.

The six-thousand-dollar gap is the extra three thousand charged in February and again in March. It is a billing dispute supported by the amendment. It is not another reduction in the contracted price.

Keep that distinction visible. At the May reporting cutoff, show six thousand supported by the invoices currently recorded, and a further six thousand under dispute. Finance decides whether the disputed amount warrants any accounting adjustment under the organization's policy. An expected credit is not evidence that the credit has arrived.

The investigation should be specific. Identify the subscription, the legal entity, both invoice numbers, the relevant service periods, the agreed quantity, and the amendment's effective date. Ask the supplier to correct the two identified differences. That is much easier to resolve than a message saying the savings report doesn't look right.

Microsoft's invoice guidance distinguishes the invoice date from the service period covered by a charge. That distinction matters beyond this example. A document arriving this month can concern an earlier period. Capture both dates so a late correction doesn't get mistaken for a new operating improvement.

In our fictional case, the supplier accepts the dispute and issues a six-thousand-dollar credit in June. It is applied against June's normal seventeen-thousand-dollar charge, leaving eleven thousand payable for that invoice.

June has not become an eleven-thousand-dollar service. The recurring charge is still seventeen thousand. Six thousand relates to correcting February and March.

Link the credit to those original invoices and show when it was applied. If finance already recognized the correction in an earlier period, its later arrival settles that item. It must not create a second benefit in the savings report.

And if you want to say cash has been saved, check settlement. An invoice, an expense entry, a credit balance, and a payment are related records, but they aren't interchangeable. In our completed fictional year, all the relevant charges and the credit are settled. Before that point, use the label the evidence supports.

Let's finish the year. From July through December, the subscription stays at seventeen thousand dollars a month. No further billing errors, new seats, or additional project costs occur. The business owner confirms that the continuing service meets the agreed requirement.

The final subscription total, after the credit, is two hundred seven thousand dollars. You can check that as January's twenty thousand plus eleven months at seventeen thousand. Against our two-hundred-forty-thousand-dollar baseline, the reduction is thirty-three thousand.

Subtract the six-thousand-dollar implementation cost. The calendar-year net benefit is twenty-seven thousand dollars under our stated assumptions.

The credit is already included in that result. Adding it again would overstate the benefit. Leaving it out would understate the benefit. The credit corrects the billing record so it agrees with the amended obligation.

We can now explain what happened to the original forty-eight thousand. Four thousand was lost because the change started a month later. Eleven thousand was lost because the commitment could only fall to eight hundred fifty seats. Six thousand was spent implementing the change. The remaining twenty-seven thousand is supported by the completed case.

That's an explanation someone else can reproduce. It also gives the next project useful information. The effective date needed more attention. The minimum commitment should have been tested before the original forecast circulated. Billing required follow-through after the technical work was complete.

Now suppose someone asks for the annualized reduction. At three thousand dollars a month, the recurring subscription reduction would be thirty-six thousand over twelve months, assuming the same scope, rate, and commitment continue. That is a forward run rate. It does not replace this year's thirty-three-thousand gross result or twenty-seven-thousand net result.

And if management spends the released budget on another service, report that allocation separately. The original service can cost less even when the total technology budget stays level. Equally, a lower total budget doesn't prove that your particular action caused the reduction.

Amazon Web Services provides another useful example of why labels matter. Its Savings Plans utilization documentation defines total net savings against an estimated On-Demand cost for the same usage. That's a defined comparison. It does not mean the organization's bill fell by that amount compared with last month.

Amazon Web Services, or AWS as it is commonly referred to, also reports how much of the commitment was used. If a workload is reduced, check what happens to that commitment and whether other eligible usage absorbs it. A technical reduction and a financial reduction can occur at different times. The answer is in the usage, commitment, and billing evidence together.

In our seat example, the fifty unassigned paid seats might later accommodate new starters. If they do, record the actual reuse. If somebody wants to claim avoided purchasing, document the additional purchase that would otherwise have been needed and agree the comparison with finance. Don't count both the retained capacity and its later reuse as separate cash savings from this year's reduction.

What if the credit never arrives, or the records are incomplete? Leave the item open with the amount, evidence gap, responsible person, and next action. Report the supported result and the unresolved amount separately. You can have a useful result before every issue is closed, provided the report makes its limits clear.

What if the service got worse? Put that alongside the financial result. Track the agreed measures: required access, completion of the work, support demand, or whatever the business owner established. A subscription reduction doesn't erase rework or an operational problem elsewhere. The FinOps Foundation's Quantify Business Value guidance explicitly includes service and organizational performance, not just monetary cost.

My recommendation is to close a benefit claim only when another person can follow its comparison and evidence. Record the scope and period, retain the source documents, explain the adjustments, and have the agreed reviewer confirm the result. Credit the people who contributed without multiplying the money by the number of departments involved.

Today's principle is traceability. Someone should be able to start with the reported result and work back to the approved action, the implemented change, and the financial records that support it.

Class dismissed. Here's your homework. Set aside about an hour and choose one completed technology change, using records you're authorized to access.

Write down the original expected benefit, its baseline, and the period it covers. Find the effective date in the executed agreement or approved change. Compare what was implemented with what was purchased. Then inspect an invoice covering the affected period and any related credit.

Record the cost of making the change. Explain every material difference between the original estimate and the result you can support. If a document is missing, name it and assign the next action. Finish with one sentence stating what has been verified and what remains unresolved. Take that page to your finance partner.

If you want a procedure to help make that work repeatable, look for F02, Validate Benefits and Report Business Outcomes, in the Operational ITAM Store. The link is in the show notes.

There is a natural next question at The Decision Table: once you've verified a result, what would cause you to revisit it? Keep that question beside your completed record. We'll return to how decisions hold up as the business changes.

The case files are open. One situation, one page. The constraint, what you did, and what happened. Remove company names and sensitive details before sending it through the website. Send me one worth working, and I'll build an episode around it.

I'm Bill Van Nort, this is the Operational ITAM Podcast. Keep the comparison visible. Follow the change through the bill. Report the result you can support. I'll talk to you next week. Take care.