The CFO Treasure Hunt: “where’s the ROI? 7 AI Finds That Pay Back

If AI doesn’t show money in the bank in 90 days, it’s not worth your CFO’s time.

Here are seven plays you can measure this quarter.

Generative AI has staggering long-term potential. McKinsey estimates it could add trillions in annual economic value as it works its way into sales, software, operations, and support. Deloitte’s recent research shows that many organizations are seeing encouraging returns from AI, particularly when they focus on clear use cases and treat AI as a business tool — not a science project. Yet other studies from MIT Sloan and BCG point out that most companies still struggle to translate pilots into measurable profit-and-loss impact.

Your job as a CFO or owner isn’t to chase the hype. It’s to turn AI from an experiment into a disciplined, repeatable source of AI ROI for business. That starts with a simple rule of thumb:

A win is anything that improves cash, margin, or risk in a way you can measure within 90 days.

In practical terms, you’re hunting for impacts in four places: labor efficiency (time back), revenue lift (more or better deals), working capital (faster cash), and risk reduction (fewer costly mistakes). Everything else is theater.

Below is your “treasure map”: seven concrete, finance-friendly AI plays that can pay back in a quarter or less — plus a simple numbers model and a 30/60/90 “CFO sprint” to make this real inside your business.

Redefining AI ROI in Plain English

Most AI initiatives fail the CFO test because they start from the wrong side of the problem. Leaders pick a tool, run a pilot, then ask, “So… did this help?” Unsurprisingly, the answer is often fuzzy.

High-performing organizations reverse that sequence. They start from a clear business outcome (for example, “reduce days sales outstanding by five days” or “double the number of proposals sent without adding headcount”), then work backward to the smallest AI intervention that could make that outcome happen. Research from firms like BCG and Deloitte shows that the companies actually getting strong AI returns tend to concentrate on a few high-value use cases, redesign the work around them, and scale those wins methodically instead of scattering pilots everywhere.

That’s the mindset behind this “treasure hunt”: fewer, better use cases — each one mapped to a visible line on your P&L.

1. Invoice Triage and AP Automation

Start with something boring and universal: accounts payable.

In most businesses, invoices arrive in every imaginable format: PDFs, emailed images, portal exports. Humans then retype details into your ERP, route approvals over email, chase corrections, and try not to miss payment terms. It’s repetitive, rules-driven work — exactly what AI is good at.

An AI “clerk” can read incoming invoices, extract key fields (vendor, amount, terms, GL hints), and gently push them into your workflow with suggested coding and pre-built approval routes. Humans still review and approve, but they’re checking, not typing.

Where does the money show up?

• Your AP team spends fewer hours per week on manual entry and chasing status.
• You reduce late fees and increase the share of invoices paid on favorable terms.
• You unlock the option to centralize AP for multiple entities without adding people.

In a typical mid-sized business, trimming even a dozen hours a week of manual AP work adds up to more than 150 hours over 90 days. Multiply that by a realistic fully loaded hourly rate and you have a clean AI cost savings example you can defend in front of the board.

2. Proposal and RFP Draft Assistant

If your revenue depends on proposals, quotes, or RFP responses, this is often the fastest AI win.

Today, sales, account management, and subject-matter experts spend days copy-pasting standard language, hunting for the last version of a case study, and manually tailoring boilerplate for each opportunity. The strategic thinking is high value. The repetitive drafting is not.

An AI assistant that’s trained on your existing proposals, product sheets, and case studies can assemble a first draft in minutes: a structured response that includes relevant proof points, pricing notes, and sections aligned with the RFP. Humans still shape the message and check the details, but they’re editing instead of starting from a blank page.

The value shows up in three places at once:

1. You send more proposals in the same amount of time.
2. Senior experts spend fewer hours on repetitive writing and more on strategy.
3. Your messaging becomes more consistent, which tends to help win rates.

If your team can move from six proposals per month to nine, with roughly the same win rate, you’ve just expanded your revenue opportunity pipeline by 50% without adding headcount. Price that on a per-deal margin basis and it’s very easy to quantify over a 90-day window.

3. Call-to-Lead and Inquiry Routing

CFOs know the math: response time kills or makes deals.

Many businesses still let leads sit in generic inboxes or voicemail boxes. A prospect calls with buying intent, leaves a message, and waits. Meanwhile, your sales team complains they don’t have enough pipeline.

AI can quietly sit in the middle of that chaos. It can transcribe inbound calls, read contact form submissions, classify intent (“new prospect,” “renewal,” “support request,” “billing issue”), and route high-value inquiries directly to the right person with context attached. You can also set alerts for keywords or dollar ranges so that truly critical leads get human attention within minutes.

This kind of routing tends to improve:

• Average response time to high-value leads
• Conversion rate on those leads
• Sales team satisfaction with lead quality

Research into high-performing AI programs consistently shows that pairing AI with clear workflow redesign — in this case, “no important lead waits more than two hours” — is where value emerges. When you measure conversion rates on AI-routed leads versus your old baseline, you can see the revenue impact in black and white.

4. Smart Collections and Dunning

From a CFO’s perspective, there are few things more concrete than days sales outstanding.

Collections is another area where AI shines. Instead of sending the same robotic template to every overdue account, an AI assistant can segment customers by balance, age, and relationship, then generate personalized, polite outreach: emails that reference prior interactions, payment history, and relevant options. It can also suggest call scripts for your AR team and prioritize which accounts to contact each day.

The numbers to watch are simple:

• Change in DSO over 90 days
• Shift in distribution of 30–60–90-day past-due balances
• Cash collected from targeted cohorts

Leading consultancies have documented that AI-powered finance functions that focus on targeted, workflow-aligned use cases like collections can capture significantly higher ROI than those that simply “add AI tools” on top of old processes. Even shaving three to five days off DSO on a $5M revenue base represents tens of thousands of dollars in freed working capital — and that’s before you account for fewer write-offs.

5. Inventory and Demand Sanity Checks

Forecasting and inventory decision-making can feel like a mix of art and science. AI won’t replace your planners, but it can give them much better “sanity checks.”

By looking across historical sales, seasonality, open orders, and lead times, an AI model can flag SKUs that appear to be overstocked relative to likely demand and others that are at high risk of stock-out. It can’t see the future, but it can surface patterns that busy humans may miss.

Where does this help your P&L?

• You reduce excess inventory and carrying costs on slow-moving SKUs.
• You reduce lost revenue from stock-outs on high-velocity, high-margin items.
• You give the finance team clearer visibility into working capital at risk.

A 5–10% reduction in surplus stock on a subset of products can more than pay for an initial AI pilot. When you track inventory value and stock-out incidents before and after deployment, you get a quantitative view of the impact — and a story your board will understand.

6. Customer Support Deflection and Triage

Support is one of the first places many companies experiment with AI — sometimes in ways that frustrate customers. The trick is to use AI to augment, not replace, your team.

A well-designed AI assistant can greet customers, answer basic FAQs, pull up relevant knowledge-base articles, and draft suggested replies for agents inside your existing ticketing system. It can also tag and route complex issues to the right human with a summary of what the customer has already told you.

This is less about building a chatbot and more about building a digital coworker that:

• Reduces the volume of simple tickets agents need to touch
• Shortens handle time on tickets that do require human intervention
• Improves first-contact resolution and customer satisfaction

MIT Sloan and other research groups have found that companies actually seeing productivity gains from AI tend to redesign roles and workflows so people and AI work in concert, not in competition. That’s exactly how support deflection and triage can become a near-term ROI source instead of a reputational risk.

7. Board Pack and KPI Report Automation

Finally, a use case close to home: monthly reporting.

Closing the books is only half the battle. The other half is telling the story — building board decks, lender updates, and KPI reports that translate numbers into insight. In many organizations, the finance and FP&A teams spend multiple days a month exporting data into spreadsheets, cleaning it up, and building slides.

Here, AI can act as a junior analyst and report writer. It can pull updated data from your BI tools or spreadsheets, generate charts, and draft narrative commentary (“Revenue grew 7% this month, driven primarily by X; gross margin held steady at Y%.”). Humans still own the judgment and final messaging, but they’re editing instead of assembling from scratch.

The savings are obvious: fewer days per month spent on low-value formatting, more time for scenario planning, pricing analysis, or strategic conversations with the business. Over a quarter, the hours you reclaim from monthly reporting alone can rival the cost of the entire AI initiative.

A Simple 90-Day Numbers Model

To keep this grounded, imagine a $5M revenue business with a 10% EBITDA margin — $500K in annual earnings.

Over 90 days, you launch three of the use cases above: AP automation, smarter collections, and a proposal assistant. You track only conservative, first-order effects:

• AP automation frees about 12 hours of manual work per week, or roughly 150 hours over the quarter. At a fully loaded $40/hour, that’s about $6,000 in capacity.
• Collections improvements reduce DSO by four days. With average daily sales of about $13,700, that’s roughly $55,000 in freed working capital. Even valuing the liquidity at a modest 5–10% effective return over the period, you’re looking at $2,700–$5,400 of economic value.
• The proposal assistant helps the team send three more proposals per month. If that yields even one extra closed deal contributing $20,000 in annual gross margin, you might realize $5,000 of that margin in the first 90 days and more thereafter.

Add it up and your conservative 90-day impact is in the $15K–$16K range — from just three small, targeted changes. Extend that logic across more processes, and suddenly AI isn’t a line item under “innovation” anymore. It’s directly expanding your earnings and strengthening your balance sheet.

This pattern mirrors what large-scale studies are finding: a small minority of “future-built” companies are capturing outsized value from AI because they focus on specific, high-leverage use cases and then scale them, while the majority dabble in pilots that never touch the P&L.

Your goal is to behave like the first group, even if you’re a mid-market firm in the Carolinas, not a global giant.

How a CFO Picks the First AI Pilot

Not every idea deserves to be your first move. A strong pilot checks four boxes:

1. Clear financial impact. You can connect it to cash, margin, or risk quickly — with a simple before/after metric, not a 40-page model.
2. Decent data and a defined process. The work already happens in digital systems (email, ERP, CRM, ticketing) and follows a repeatable pattern, even if it’s manual.
3. A committed process owner. One person in the business cares deeply about this problem and is willing to test, give feedback, and help with adoption.
4. Manageable downside. If the AI makes a mistake, it’s annoying but fixable, not catastrophic. You can put approvals and guardrails around it.

If a use case like AP automation, proposal drafting, or collections checks these boxes, it’s a strong candidate for your first 90-day sprint.

At Queen City AI, we formalize this step in a Discovery Session: we sit with your finance and operations leaders, map candidate workflows, and score them on impact, feasibility, and risk before we ever touch a model.

Turning It into a 30/60/90 “CFO Sprint”

Think of the next quarter as an experiment in disciplined AI ROI for CFOs, not an open-ended transformation program.

Days 0–30: Discover and design.

You pick one or two of the plays above, map the current workflow, gather baseline metrics, and design a lightweight AI workflow. The question you’re answering is, “What would a digital coworker do here, step by step?” By the end of the month, you have a working prototype in a limited slice of the process.

Days 31–60: Pilot and prove.

You run the AI alongside the existing workflow. Humans are still in control, but they start using the assistant in real work. Each week, you compare outcomes to your baseline: hours spent, invoices processed, proposals sent, DSO, or tickets resolved. You refine prompts, rules, and guardrails as you go.

Days 61–90: Normalize and decide.

You decide which parts of the new workflow become “business as usual,” lock in documentation and ownership, and roll the pilot into a standard operating procedure. Then you build a simple ROI summary for the quarter and decide which treasure chest to open next.

External research is clear: the organizations that consistently get value from AI treat it as an ongoing operating-model shift, not a toy. They redesign work around AI, invest in training, and scale what works instead of chasing every new feature. This 30/60/90 sprint model lets you do that in a way that fits a mid-market finance function.

Download the CFO 90-Day ROI Model

If you’d like to run this treasure hunt without building your own tools from scratch, Queen City AI has packaged the math into a simple spreadsheet model.

The CFO 90-Day AI ROI Model lets you plug in your own:

• Current process costs and hours
• Expected time savings by role
• Target improvements in DSO or inventory
• Conservative revenue lift assumptions

From there, it calculates payback period, 90-day ROI, and annualized impact so you can judge AI projects with the same discipline you bring to any capital decision.

You can also take the next step and book a Discovery Session with Queen City AI. In a focused, on-site workshop, we’ll walk your finance and operations leaders through your current systems, identify the highest-value AI opportunities, and design a 90-day sprint tailored to your business.

No jargon, no tool-chasing — just a structured treasure hunt for the AI “finds” that actually move your P&L.The CFO Treasure Hunt: “Where’s the ROI? 7 AI Finds That Pay Back in 90 Days”

If AI doesn’t show money in the bank in 90 days, it’s not worth your CFO’s time. Here are seven plays you can measure this quarter.

Generative AI has staggering long-term potential. McKinsey estimates it could add trillions in annual economic value as it works its way into sales, software, operations, and support. Deloitte’s recent research shows that many organizations are seeing encouraging returns from AI, particularly when they focus on clear use cases and treat AI as a business tool — not a science project. Yet other studies from MIT Sloan and BCG point out that most companies still struggle to translate pilots into measurable profit-and-loss impact.

Your job as a CFO or owner isn’t to chase the hype. It’s to turn AI from an experiment into a disciplined, repeatable source of AI ROI for business. That starts with a simple rule of thumb:

A win is anything that improves cash, margin, or risk in a way you can measure within 90 days.

In practical terms, you’re hunting for impacts in four places: labor efficiency (time back), revenue lift (more or better deals), working capital (faster cash), and risk reduction (fewer costly mistakes). Everything else is theater.

Below is your “treasure map”: seven concrete, finance-friendly AI plays that can pay back in a quarter or less — plus a simple numbers model and a 30/60/90 “CFO sprint” to make this real inside your business.

Redefining AI ROI in Plain English

Most AI initiatives fail the CFO test because they start from the wrong side of the problem. Leaders pick a tool, run a pilot, then ask, “So… did this help?” Unsurprisingly, the answer is often fuzzy.

High-performing organizations reverse that sequence. They start from a clear business outcome (for example, “reduce days sales outstanding by five days” or “double the number of proposals sent without adding headcount”), then work backward to the smallest AI intervention that could make that outcome happen. Research from firms like BCG and Deloitte shows that the companies actually getting strong AI returns tend to concentrate on a few high-value use cases, redesign the work around them, and scale those wins methodically instead of scattering pilots everywhere.

That’s the mindset behind this “treasure hunt”: fewer, better use cases — each one mapped to a visible line on your P&L.

1. Invoice Triage and AP Automation

Start with something boring and universal: accounts payable.

In most businesses, invoices arrive in every imaginable format: PDFs, emailed images, portal exports. Humans then retype details into your ERP, route approvals over email, chase corrections, and try not to miss payment terms. It’s repetitive, rules-driven work — exactly what AI is good at.

An AI “clerk” can read incoming invoices, extract key fields (vendor, amount, terms, GL hints), and gently push them into your workflow with suggested coding and pre-built approval routes. Humans still review and approve, but they’re checking, not typing.

Where does the money show up?

• Your AP team spends fewer hours per week on manual entry and chasing status.
• You reduce late fees and increase the share of invoices paid on favorable terms.
• You unlock the option to centralize AP for multiple entities without adding people.

In a typical mid-sized business, trimming even a dozen hours a week of manual AP work adds up to more than 150 hours over 90 days. Multiply that by a realistic fully loaded hourly rate and you have a clean AI cost savings example you can defend in front of the board.

2. Proposal and RFP Draft Assistant

If your revenue depends on proposals, quotes, or RFP responses, this is often the fastest AI win.

Today, sales, account management, and subject-matter experts spend days copy-pasting standard language, hunting for the last version of a case study, and manually tailoring boilerplate for each opportunity. The strategic thinking is high value. The repetitive drafting is not.

An AI assistant that’s trained on your existing proposals, product sheets, and case studies can assemble a first draft in minutes: a structured response that includes relevant proof points, pricing notes, and sections aligned with the RFP. Humans still shape the message and check the details, but they’re editing instead of starting from a blank page.

The value shows up in three places at once:

1. You send more proposals in the same amount of time.
2. Senior experts spend fewer hours on repetitive writing and more on strategy.
3. Your messaging becomes more consistent, which tends to help win rates.

If your team can move from six proposals per month to nine, with roughly the same win rate, you’ve just expanded your revenue opportunity pipeline by 50% without adding headcount. Price that on a per-deal margin basis and it’s very easy to quantify over a 90-day window.

3. Call-to-Lead and Inquiry Routing

CFOs know the math: response time kills or makes deals.

Many businesses still let leads sit in generic inboxes or voicemail boxes. A prospect calls with buying intent, leaves a message, and waits. Meanwhile, your sales team complains they don’t have enough pipeline.

AI can quietly sit in the middle of that chaos. It can transcribe inbound calls, read contact form submissions, classify intent (“new prospect,” “renewal,” “support request,” “billing issue”), and route high-value inquiries directly to the right person with context attached. You can also set alerts for keywords or dollar ranges so that truly critical leads get human attention within minutes.

This kind of routing tends to improve:

• Average response time to high-value leads
• Conversion rate on those leads
• Sales team satisfaction with lead quality

Research into high-performing AI programs consistently shows that pairing AI with clear workflow redesign — in this case, “no important lead waits more than two hours” — is where value emerges. When you measure conversion rates on AI-routed leads versus your old baseline, you can see the revenue impact in black and white.

4. Smart Collections and Dunning

From a CFO’s perspective, there are few things more concrete than days sales outstanding.

Collections is another area where AI shines. Instead of sending the same robotic template to every overdue account, an AI assistant can segment customers by balance, age, and relationship, then generate personalized, polite outreach: emails that reference prior interactions, payment history, and relevant options. It can also suggest call scripts for your AR team and prioritize which accounts to contact each day.

The numbers to watch are simple:

• Change in DSO over 90 days
• Shift in distribution of 30–60–90-day past-due balances
• Cash collected from targeted cohorts

Leading consultancies have documented that AI-powered finance functions that focus on targeted, workflow-aligned use cases like collections can capture significantly higher ROI than those that simply “add AI tools” on top of old processes. Even shaving three to five days off DSO on a $5M revenue base represents tens of thousands of dollars in freed working capital — and that’s before you account for fewer write-offs.

5. Inventory and Demand Sanity Checks

Forecasting and inventory decision-making can feel like a mix of art and science. AI won’t replace your planners, but it can give them much better “sanity checks.”

By looking across historical sales, seasonality, open orders, and lead times, an AI model can flag SKUs that appear to be overstocked relative to likely demand and others that are at high risk of stock-out. It can’t see the future, but it can surface patterns that busy humans may miss.

Where does this help your P&L?

• You reduce excess inventory and carrying costs on slow-moving SKUs.
• You reduce lost revenue from stock-outs on high-velocity, high-margin items.
• You give the finance team clearer visibility into working capital at risk.

A 5–10% reduction in surplus stock on a subset of products can more than pay for an initial AI pilot. When you track inventory value and stock-out incidents before and after deployment, you get a quantitative view of the impact — and a story your board will understand.

6. Customer Support Deflection and Triage

Support is one of the first places many companies experiment with AI — sometimes in ways that frustrate customers. The trick is to use AI to augment, not replace, your team.

A well-designed AI assistant can greet customers, answer basic FAQs, pull up relevant knowledge-base articles, and draft suggested replies for agents inside your existing ticketing system. It can also tag and route complex issues to the right human with a summary of what the customer has already told you.

This is less about building a chatbot and more about building a digital coworker that:

• Reduces the volume of simple tickets agents need to touch
• Shortens handle time on tickets that do require human intervention
• Improves first-contact resolution and customer satisfaction

MIT Sloan and other research groups have found that companies actually seeing productivity gains from AI tend to redesign roles and workflows so people and AI work in concert, not in competition. That’s exactly how support deflection and triage can become a near-term ROI source instead of a reputational risk.

7. Board Pack and KPI Report Automation

Finally, a use case close to home: monthly reporting.

Closing the books is only half the battle. The other half is telling the story — building board decks, lender updates, and KPI reports that translate numbers into insight. In many organizations, the finance and FP&A teams spend multiple days a month exporting data into spreadsheets, cleaning it up, and building slides.

Here, AI can act as a junior analyst and report writer. It can pull updated data from your BI tools or spreadsheets, generate charts, and draft narrative commentary (“Revenue grew 7% this month, driven primarily by X; gross margin held steady at Y%.”). Humans still own the judgment and final messaging, but they’re editing instead of assembling from scratch.

The savings are obvious: fewer days per month spent on low-value formatting, more time for scenario planning, pricing analysis, or strategic conversations with the business. Over a quarter, the hours you reclaim from monthly reporting alone can rival the cost of the entire AI initiative.

A Simple 90-Day Numbers Model

To keep this grounded, imagine a $5M revenue business with a 10% EBITDA margin — $500K in annual earnings.

Over 90 days, you launch three of the use cases above: AP automation, smarter collections, and a proposal assistant. You track only conservative, first-order effects:

• AP automation frees about 12 hours of manual work per week, or roughly 150 hours over the quarter. At a fully loaded $40/hour, that’s about $6,000 in capacity.
• Collections improvements reduce DSO by four days. With average daily sales of about $13,700, that’s roughly $55,000 in freed working capital. Even valuing the liquidity at a modest 5–10% effective return over the period, you’re looking at $2,700–$5,400 of economic value.
• The proposal assistant helps the team send three more proposals per month. If that yields even one extra closed deal contributing $20,000 in annual gross margin, you might realize $5,000 of that margin in the first 90 days and more thereafter.

Add it up and your conservative 90-day impact is in the $15K–$16K range — from just three small, targeted changes. Extend that logic across more processes, and suddenly AI isn’t a line item under “innovation” anymore. It’s directly expanding your earnings and strengthening your balance sheet.

This pattern mirrors what large-scale studies are finding: a small minority of “future-built” companies are capturing outsized value from AI because they focus on specific, high-leverage use cases and then scale them, while the majority dabble in pilots that never touch the P&L.

Your goal is to behave like the first group, even if you’re a mid-market firm in the Carolinas, not a global giant.

How a CFO Picks the First AI Pilot

Not every idea deserves to be your first move. A strong pilot checks four boxes:

1. Clear financial impact. You can connect it to cash, margin, or risk quickly — with a simple before/after metric, not a 40-page model.
2. Decent data and a defined process. The work already happens in digital systems (email, ERP, CRM, ticketing) and follows a repeatable pattern, even if it’s manual.
3. A committed process owner. One person in the business cares deeply about this problem and is willing to test, give feedback, and help with adoption.
4. Manageable downside. If the AI makes a mistake, it’s annoying but fixable, not catastrophic. You can put approvals and guardrails around it.

If a use case like AP automation, proposal drafting, or collections checks these boxes, it’s a strong candidate for your first 90-day sprint.

At Queen City AI, we formalize this step in a Discovery Session: we sit with your finance and operations leaders, map candidate workflows, and score them on impact, feasibility, and risk before we ever touch a model.

Turning It into a 30/60/90 “CFO Sprint”

Think of the next quarter as an experiment in disciplined AI ROI for CFOs, not an open-ended transformation program.

Days 0–30: Discover and design.

You pick one or two of the plays above, map the current workflow, gather baseline metrics, and design a lightweight AI workflow. The question you’re answering is, “What would a digital coworker do here, step by step?” By the end of the month, you have a working prototype in a limited slice of the process.

Days 31–60: Pilot and prove.

You run the AI alongside the existing workflow. Humans are still in control, but they start using the assistant in real work. Each week, you compare outcomes to your baseline: hours spent, invoices processed, proposals sent, DSO, or tickets resolved. You refine prompts, rules, and guardrails as you go.

Days 61–90: Normalize and decide.

You decide which parts of the new workflow become “business as usual,” lock in documentation and ownership, and roll the pilot into a standard operating procedure. Then you build a simple ROI summary for the quarter and decide which treasure chest to open next.

External research is clear: the organizations that consistently get value from AI treat it as an ongoing operating-model shift, not a toy. They redesign work around AI, invest in training, and scale what works instead of chasing every new feature. This 30/60/90 sprint model lets you do that in a way that fits a mid-market finance function.

Download the CFO 90-Day ROI Model

If you’d like to run this treasure hunt without building your own tools from scratch, Queen City AI has packaged the math into a simple spreadsheet model.

The CFO 90-Day AI ROI Model lets you plug in your own:

• Current process costs and hours
• Expected time savings by role
• Target improvements in DSO or inventory
• Conservative revenue lift assumptions

From there, it calculates payback period, 90-day ROI, and annualized impact so you can judge AI projects with the same discipline you bring to any capital decision.

You can also take the next step and book a Discovery Session with Queen City AI. In a focused, on-site workshop, we’ll walk your finance and operations leaders through your current systems, identify the highest-value AI opportunities, and design a 90-day sprint tailored to your business.

No jargon, no tool-chasing — just a structured treasure hunt for the AI “finds” that actually move your P&L.