The CFO Asked What the Project Returns and Your Heat Map Had No Answer: Build the Before and After Expected Loss Memo
The views expressed are my own and do not represent those of my employer.
Imagine a CISO in the annual capital review. Three requests are on the agenda. A packaging line upgrade comes first, with a payback period and a net present value. A warehouse automation project follows, with the same two numbers. Then the CISO presents a risk register full of red and amber cells and asks for money to turn them green.
The CFO asks one question: "What does it return?" The CISO talks about threats, maturity, and best practice. The CFO nods and moves to the next item. Nobody rejects the request. It simply cannot be compared with anything else on the page, and in a capital review that amounts to the same thing.
I have seen versions of this scene play out at multiple organizations. The security case was sound. The format lost it.
The gap between spending and measuring is wide. PwC's 2026 Global Digital Trust Insights report, as summarized by CPO Magazine, found that 78% of organizations expect their cyber budgets to increase, yet only 16% significantly measure the financial impact of cyber risks. Money is coming. The leaders who can state their requests in dollars will get a bigger share of it.
This post is long, and it splits into three parts. Part 1 is strategic: why expected loss is the language of the capital review. Part 2 is tactical: how to build the decision memo, with three worked examples. Part 3 covers presenting it to executives. Read all three, or jump to the part you need.
Part 1: Stop Reporting Activity and Report Expected Loss
The core move is simple. Stop reporting what your team did. Report what the business is expected to lose each year before the project, what it is expected to lose after, and what the project costs. Put those three numbers in the same currency as every other capital request.
PwC's 2026 CISO material makes the case directly. It says cyber risk quantification gives you the tools to clarify what matters most and the credibility to make your case in the boardroom. It also says that embedding risk quantification into board dashboards helps shift the conversation from risk to resilience, and from cost to value.
Cyberstrategix, writing in July 2026, argues the same from the budget side. Its claim is that the CISOs winning the budget argument are the ones who have moved from red, amber, green to a defensible number. It also offers a test I like because a CFO can run it in their head. A control that costs two million dollars a year and cuts expected annual loss by seven million has a clear, board legible business case. A control with the same cost that cuts expected loss by only four hundred thousand is a much harder sell, and the model lets the board see that difference.
A request stated in expected loss can be compared. A request stated in colors can only be admired.
Expected loss is not exotic. The core formula behind the FAIR style of analysis is that probable loss equals loss event frequency multiplied by loss magnitude. How often does the bad thing happen in a year? What does it cost when it does? Everything else in this post is a disciplined way to answer those two questions and show your work.
I recently wrote about identity metrics and ROI from an IAM specific perspective, in the audit chair post. This post approaches it differently. It gives you one method that works for any project, whether it is identity, network, cloud, or a tool you are about to say no to.
What NIST CSF 2.0 Already Asks You to Do
You do not need to invent a mandate for this. NIST CSF 2.0 already points in this direction. Under GV.RM-06, the standardized method for calculating, documenting, and prioritizing cyber risk, one implementation example is to establish criteria for using a quantitative approach to cybersecurity risk analysis, and specify probability and exposure formulas.
Read that carefully. NIST does not say quantify everything. It says write down the criteria for when you will. That is a policy decision you can make this quarter. My recommendation: any request above a set dollar threshold, and any request going to the board, gets a quantified memo.
GV.RM-03 covers the other half. One of its examples is to aggregate and manage cybersecurity risks alongside other enterprise risks, such as compliance, financial, and regulatory. You cannot aggregate a color with a currency. Expected loss is what lets your cyber risk sit in the enterprise register next to credit risk and supply risk.
GV.RM-02 closes the loop. Risk appetite statements have to be translated into specific, measurable, and broadly understandable risk tolerance statements. A tolerance stated in dollars is what your expected loss number gets measured against. I walked through the whole Govern function in my earlier NIST CSF 2.0 series. Here I am using three of its outcomes as the foundation for a budget conversation.
If you built a risk register during your first 60 days, as I described in my Days 31 to 60 roadmap post, you already have your scenario candidates. I mentioned quantification there only in passing. This is the follow through.
Part 2: The Five Parts of the Decision Memo
The memo fits on one page. If it takes more, you are explaining the model instead of the decision. Here are the five parts, in order.
1. One scenario and the asset it threatens. Pick a single, specific story. "Ransomware reaches the order processing platform through a stolen password" is a scenario. "Cyber risk" is not. Name the asset or business process, because the business owner will be the one who judges your loss figures.
2. Today's expected annual loss, as a range with stated assumptions. Give a low, a most likely, and a high figure. State the frequency and the loss per event behind each. A single number such as $2,317,000 signals false precision, and a CFO will distrust it on sight. A range with visible assumptions invites a useful argument about inputs, which is the argument you want.
3. Loss after the project, project cost, and net change. Run the same scenario with the project in place. Show the new range, the annualized cost, and the difference. The most useful single figure is risk reduction per dollar, which is the same idea as return on security investment, or ROSI. One common formula is monetary loss avoided minus the cost of the solution, divided by the cost of the solution.
4. Two or three options side by side. Do not defend one ask alone. Cyberstrategix recommends using the model for comparative prioritization, showing the board the expected risk reduction per dollar of competing options, not only the standalone case for the budget requested this quarter. Options also protect you. If the CFO asks "what else could we do," your answer is already on the page.
5. The cost of doing nothing. Show the baseline as a line item. It is not a scare tactic. It is today's expected loss, carried forward, set against the risk tolerance the business already approved.
Now the worked examples. Every number below is illustrative. I invented them to show the mechanics. Substitute your own frequency, loss, and cost data, and expect your results to look different.
Worked Example One: Identity Against Ransomware
Imagine a mid size manufacturer. The scenario is an attacker using a stolen password to reach the order processing platform and deploy ransomware. The asset is the platform that books orders and ships product.
Today. Suppose the team estimates one such event every three to seven years, a frequency of 0.15 to 0.30 per year. Suppose loss per event runs $6M to $14M, covering downtime, recovery, lost orders, and outside help. Multiplying the low ends and the high ends gives an expected annual loss of $0.9M to $4.2M. The most likely figure is about $2.25M, using 0.225 events at $10M.
The project. Phishing resistant MFA on remote and privileged access, plus cleanup of standing admin accounts. Annualized cost: $500K. Suppose it roughly halves the frequency, to 0.08 to 0.16 per year, and leaves loss per event unchanged. Expected annual loss after the project is $0.48M to $2.24M, with $1.2M most likely.
The net. The most likely reduction is $1.05M a year against a $500K cost. That is a net benefit of $550K, or $2.10 of risk reduction per dollar spent. Across the range, risk reduction per dollar runs from $0.84 to $3.92.
Look at that low end. At the pessimistic corner of the range, the project does not pay for itself. I put that in the memo. A CFO who sees you volunteer the weak case trusts the strong case more.
Doing nothing. The baseline is $2.25M a year most likely, or $6.75M of expected loss over three years if nothing changes. Suppose the business tolerance for this scenario is $1.5M a year. Today sits outside it. After the project, the most likely figure sits inside it, although the high end still pokes above the line. Say that too.
Worked Example Two: Edge and Segmentation, Three Options
Take a different scenario. An attacker exploits an internet facing remote access appliance and moves into a plant network. The asset is production at the plants. I covered the design thinking in my post on limiting the impact of compromise. Here I price it.
Today. Suppose frequency is 0.2 to 0.4 material events per year and loss per event is $3M to $8M. The range is $0.6M to $3.2M, and the most likely figure is $1.5M (0.30 events at $5M).
Now three options, each judged on most likely values:
- Option A, edge device replacement and hardening: $350K. It cuts frequency from 0.30 to 0.18. Reduction: $0.6M. Risk reduction per dollar: $1.71.
- Option B, full segmentation across every site: $1.2M. It cuts loss per event from $5M to $2M. Reduction: $0.9M. Risk reduction per dollar: $0.75.
- Option C, phased segmentation of the two highest value zones: $400K. It cuts loss per event from $5M to $3.5M. Reduction: $0.45M. Risk reduction per dollar: $1.13.
The recommendation writes itself. Fund A first. Fund C second. Defer B until the first two have produced data. The full segmentation project may be the right long term design, and the memo does not dispute that. It says the first dollars buy more risk reduction elsewhere.
One more detail. If you fund A and C together, do not add the reductions. Frequency drops to 0.18 and loss per event drops to $3.5M, so the new expected loss is $0.63M. The combined reduction is $0.87M for $750K, or $1.16 per dollar. That is less than $1.05M, the sum of the two parts. The controls overlap. Teams that add the numbers get caught in the first question.
Controls overlap. If you add up each project's benefit and call it the portfolio, the CFO will find the double counting before you do.
Worked Example Three: The Tool I Would Not Fund
This is the example that builds the most trust. A vendor pitches a $400K a year monitoring add on. The scenario is stolen customer credentials posted externally. Suppose today's expected annual loss is $0.1M to $0.3M, most likely $0.2M. Existing controls already catch most of it, and the tool would trim perhaps 10 percent. That is $20K of reduction for $400K of cost, or $0.05 per dollar.
The decision is no. Better still, notice that it was a no before any modeling. A tool cannot return more than the entire expected loss of the scenario it addresses. When the cost exceeds the whole number, you are done.
Put the rejection in the deck as a line item: considered, not recommended, with the reason in dollars. A CISO who shows what they would not fund is a CISO whose "yes" means something. Finance leaders notice that quickly.
Part 3: Ten Minutes, One Decision
A good memo can still die in the room. Gartner analysts at the Security and Risk Management Summit 2026 reminded attendees that a CISO, if lucky, will get only five to ten minutes to present to the board. Their advice, as reported by TechTarget, is to use monthly or quarterly financial reports as templates for cybersecurity board reporting.
The same coverage describes mapping reports to three financial structures: a balance sheet snapshot of current risk posture, an income statement view of threat driven financial impact, and a cash flow breakdown of resource allocation. Your decision memo fits that format naturally. Today's loss is the balance sheet. The project's effect is the income statement. The cost is the cash flow.
Kovrr, a CRQ vendor, publishes a board reporting guide with a compatible view. It says the board report should focus on aggregate financial exposure, top scenarios, movement over time, and investment ROI. It also says that operational metrics like patch rates and mean time to detect belong on the CISO's dashboard, not the board's.
Here is how I would run the room.
Prewire the CFO. Walk the finance chief through the memo a week early. Ask them to challenge the loss inputs. Their fingerprints on the numbers are worth more than any slide. Nobody likes learning a figure in public.
Lead with the decision. Your first sentence is the ask: "I need $500K a year to cut expected ransomware loss by about a million." Then show the evidence. The model is support, not the opening act.
Lead with the decision you need. The model is the evidence, not the opening act.
Show ranges and the confidence behind them. Say which inputs come from your own incident and loss data, which come from industry sources, and which are judgment calls. Directors accept uncertainty. They do not accept uncertainty you hid.
Tie every number to the appetite the business already set. This is GV.RM-02 in practice. Kovrr makes a useful point here: the test of an appetite statement is whether anyone could determine on any given day if the organization sits inside or outside it, which requires a current exposure figure and a threshold in the same unit. If no dollar tolerance exists, propose one in the same meeting and ask for it to be adopted.
Bring a follow up number. Tell the board when you will report back and what you will report. "In two quarters I will show you the observed change in the controls behind this estimate, and the revised expected loss." That promise turns one approval into a standing relationship. For the broader mechanics of earning and keeping executive attention, see Winning the Room.
I am planning a series on security metrics by audience under CSF 2.0. In that series, CRQ is the leadership level ROI layer. Operational metrics feed it. The patch rates and detection times your team tracks are the inputs that move the frequency and loss estimates in your memo.
Where the Model Lies to You
Now the caution. A model is only as good as its inputs, and I want you to hold every number in this post to that standard, mine included.
First, notice who writes much of the guidance. Kovrr sells a CRQ platform. Cyberstrategix is a consultancy. Several of the other ROI explainers I read while researching this post come from vendors. That does not make them wrong. It does mean the field's loudest voices benefit when you buy a platform. You do not need one to build the memo in this post. A spreadsheet and honest ranges will carry you a long way.
Second, PwC itself warns about the failure mode. CRQ can never be a plug and play solution, and too often it becomes a point in time exercise focused on generating numbers rather than delivering real insights.
Third, you cannot observe the breach that did not happen. Every reduction estimate is a counterfactual. Say so plainly, and show how you will check the estimate against evidence later.
Fourth, use a sanity check on the output. One practitioner guide notes that a defensible ROSI for a mid market investment typically lands in the low hundreds of percent, and that results in the thousands usually signal inflated loss estimates or an overly optimistic mitigation ratio. If your memo shows a 4,000 percent return, you did not find a miracle. You found an input to fix.
Getting Organizational Support: Selling the Number From the Floor to the Boardroom
A quantified memo needs allies at every level. Each audience cares about a different piece of it, and none of them wants security jargon.
Engineers and IT owners. They supply the inputs on how controls actually work. Ask for their help early and credit them in the memo. People defend numbers they helped build, and they correct numbers they were never asked about.
Business unit leaders. They own the loss side. Nobody knows what an hour of line downtime or a week of delayed shipments costs better than the person running the business. Frame the conversation as "help me price what this would cost you," not "approve my security project." Their figures make the memo credible.
Finance. Ask FP&A to review your method, not just your totals. Use their discount rates and their definition of annualized cost. When the CFO sees their own conventions in your memo, the format question disappears and the substance stays.
The CIO and peer executives. Show them where the project removes friction or avoids rework, not only where it reduces loss. Peers support requests that make their own roadmaps easier. Run the memo past them before it reaches the CEO.
The CEO. Give the CEO one sentence: what the business is exposed to, what you propose, and what it changes. Connect it to revenue continuity, customer commitments, and the plant or service that cannot stop.
The audit committee and the board. Frame everything as risk the business is carrying against the appetite it chose. Directors oversee risk and allocate capital, so give them both views: the exposure and the return on reducing it. The Gartner coverage describes success criteria for this reporting model that include increasing approval rates for cybersecurity investment requests. Measure yourself the same way.
Key Points
- Report expected loss before and after the project, in the same currency as every other capital request.
- Write one scenario, one asset, one page.
- Give ranges with stated assumptions. A single precise number signals false precision.
- Compare two or three options by risk reduction per dollar.
- Show the cost of doing nothing as the baseline, measured against the tolerance the business already set.
- Show what you would not fund. It builds trust in what you would.
- Prewire the CFO, lead with the decision, and promise a follow up number.
- Treat vendor sourced guidance and your own inputs with the same skepticism.
Pro Tips
Start with the loss side, not the control side. Most CISOs begin by describing the project and then hunt for a number to justify it. Reverse that. Sit with a business owner first, agree on what the scenario would cost, and only then price the project. The estimate will be smaller and more believable.
Pair every estimate with an evidence trail. For each input, note whether it came from your own history, an industry source, or expert judgment. Three columns on a spreadsheet make an audit of your memo a five minute conversation.
Price the realistic step, not the perfect end state. A project that moves a control from partial to mostly complete this year should be priced on that move. Pricing it against full implementation overstates what you will deliver and leaves you owing a result you cannot produce.
Keep a decision log. Record the estimate, the date, the approver, and the follow up date. Next year, when you ask for money again, the log is your track record.
Rehearse the hostile question. The question will be "Why is your frequency that number?" Have a two sentence answer ready, with the source and the range, before you walk in.
Pitfalls to Avoid
Do not present a single point estimate as fact. A range with assumptions survives scrutiny. A precise number gets picked apart.
Do not sum project benefits across overlapping controls. Rerun the scenario with the combined set in place, as in the segmentation example, and report that figure.
Do not argue for one ask in isolation. A board with no alternatives will either say yes without understanding or say no out of caution. Options let them choose.
Do not build the model alone. Estimates the business owner never saw are the first ones challenged and the last ones defended.
Do not treat the quarterly update as optional. A number that was never revisited is the same as a number that was wrong. Report the change, even when it is unflattering.
Final Thought
Every capital request in your company is a bet on a return, and the leaders who get funded are the ones who frame their bet in terms the room already uses. Learn to say "this reduces expected loss by this much, for this much, and here is how I will prove it," and security stops being the item that gets nodded past and becomes the item that gets compared, and often chosen.
If you have built a before and after memo, or tried and hit a wall, I want to hear about it. Leave a comment with what worked and what did not. Subscribe if you want the upcoming series on security metrics by audience, and share this with a peer who is heading into a budget review. I read every one.
Comments ()