How do you measure the ROI of AI in accounting?

Set a baseline first. Dawn Hatch, CEO of AI-native accounting firm MATAX, recommended tracking time spent and accuracy for a month or two before implementation. ROI can also show up as broader review coverage, such as checking every transaction instead of a sample, and as staff shifting to analysis.

What is AI model drift, and how do you prevent it?

Model drift is a change in an AI workflow’s output quality over time, often after a model, prompt or automation update. The panel’s remedy is regular evaluations against a fixed set of known cases with known answers, rerun after every change.

How do you prevent shadow AI in a finance team?

Give people approved enterprise tools and a fast process for adding new ones. The panel said banning AI pushes employees to personal accounts, where data may be used to train public models.

Should finance teams use general AI models or vertical AI tools?

The panel suggested both, for different jobs. General enterprise models help staff build AI skills quickly. Vertical tools add what general models lack out of the box, such as legacy system integrations, data validation and workflow fit.

What finance work can be handed fully to AI?

Very little, today. Panelists said simple items, like a standard W-2, may not need review, but complex documents and anything AI agent does inside the accounting system should have a named person approving it.

AI in the Finance Function: What’s Working, What Stalls, and How to Govern It

Scrubbed

AI in the Finance Function: What’s Working, What Stalls, and How to Govern It

At A Glance
AI in the Finance Function was the second panel at The Future Is Fractional 2026, hosted by Scrubbed on September 24, 2026. Scrubbed Chief AI Officer Kendrick Kho moderated a conversation with Rusty Canada, partner and co-founder of Ternpoint Solutions; Dawn Hatch, founding partner and CEO of MATAX, an AI-native accounting firm; and Blaze O’Byrne, co-founder of Byron, a startup building AI agents for CPA tax workflows. The panel covered where AI delivers measurable ROI in accounting today, why pilots stall, how to run evaluations, how to govern AI without pushing staff toward personal tools, and how each would spend a $500,000 technology budget. The short version: measure before you automate, document the process first, and keep a named person accountable for every output.

Most finance leaders no longer ask whether to use AI. They ask where it actually pays off, what breaks when it meets a real workflow, and how to let people experiment without putting client data at risk.

That was the brief for Session 2 of The Future is Fractional, a virtual conference hosted by finance and accounting advisory firm Scrubbed on September 24, 2026. Kendrick Kho, Scrubbed’s Chief AI Officer, moderated a conversation with three practitioners who build, sell and run AI in accounting every day.

Key takeaways
AI pays off first in high-volume, reviewable work. Think transaction coding, document extraction, first-pass reconciliation review and narrative drafting.
Measure before you automate. Without a baseline for time and accuracy, you can’t prove ROI or know whether AI beats your current process.
Pilots stall without a documented process. Start with one narrow workflow, one owner, a written process and a way to check the output.
Test continuously. Run AI workflows against a fixed set of known answers, and retest after every model or prompt change, so quality doesn't slip unnoticed.
A person owns the output. Every panelist kept a named human reviewer in the loop.
Governance should create a “yes lane.” Banning tools pushes staff to personal accounts; classifying data and approving tools quickly keeps them inside the guardrails.

Meet The Panel

Kendrick Kho (moderator): Chief AI Officer, Scrubbed.

Rusty Canada: Partner and Co-founder, Ternpoint Solutions, with more than 15 years leading digital transformation and building technology practices.

Dawn Hatch: Founding Partner and CEO, MATAX, an AI-native accounting firm with more than 800 startup clients. She was named Xero’s 2026 U.S. Advisory Innovator of the Year and an Anthropic partner.

Blaze O’Byrne: Co-Founder, Byron, which builds AI agents that automate tax work for CPA firms. He was previously a partner at Citi Ventures, investing in early-stage AI companies.

Where AI delivers measurable ROI in accounting today

Hatch’s firm tracked time and accuracy before adopting AI, so it had a baseline to compare against. The clearest returns showed up in four places:

Transaction coding at volume. The bigger win wasn’t speed. It was that 100% of transactions got reviewed rather than a sample, so quality rose.
Document extraction. Bills, receipts and statements become structured data that a person reviews instead of keying in.
First-pass reconciliation review. AI drafts explanations for differences and flags anomalies; a person confirms and corrects.
Narrative drafting. Close summaries, client emails, reports and memos arrive about 90% complete for a person to finish.

How to measure AI ROI

“Measurable means that you have to measure,” Hatch said. Many firms can’t report ROI because they never set a baseline. Her advice: spend a month or two tracking both time spent and accuracy before implementation.

O’Byrne added that people often judge AI against perfection. A tool that doesn’t hit 100% may still beat the accuracy a firm achieves today. Kho noted that some ROI isn’t time saved at all: when AI removes data entry from an accounts payable (AP) role, that person shifts toward vendor work and analysis.

What to Hand to AI, and What to Keep with a Person

O’Byrne’s product prepares tax returns end to end, but it’s designed so a CPA reviews key steps. It scores its own confidence in extracted data, tells reviewers where to look, and keeps an audit trail showing who approved what and when. A simple W-2 may not need human review; a complex K-1 probably does. “The onus is still on the user.”

How to Speed up AI Adoption without Creating Maintenance Debt

Canada recommended a “speedboat” approach: start small, build small AI agents (tools that carry out multi-step taks on their own) quickly to win user buy-in, and be ready to throw one out and rebuild it once you learn more. His own team lost a month or two trying to design governance for everything before shipping anything.

O’Byrne offered the counterweight. “Because you can build everything, you should build everything” is one of the biggest mistakes of the moment, he said, because everything you build has to be maintained. Hatch agreed that the time doesn’t disappear. Her team now spends more of it on evaluation, review and fine-tuning than anything else.

Vertical AI vs. general frontier models

Asked whether specialized AI companies can keep adding value as general models improve, O’Byrne said he once believed the models would absorb vertical tools. Building one changed his mind.

	General AI Models (e.g., ChatGPT, Claude)	Vertical AI Tools for Finance
What the panel said they’re good for	Broad staff use and upskilling; a first step before other tools	Specific workflows such as tax preparation
What they don’t do	Integrate with legacy systems, validate data or fit a firm’s workflow out of the box	Remove the reviewer: O’Byrne’s tool still routes key steps to a CPA

Most employees, O’Byrne said, want an application that fits their workflow, not a blank tool to build their own. His company uses the major models under the hood.

Why AI pilots stall, and how to run evaluations

Hatch’s rule: pilots succeed when there’s a documented process before AI arrives, and stall when there isn’t. Without it, “you’re just making confusion faster.”

AI pilot checklist

One narrow workflow
One named owner
A written process
A defined way to check the output
A record of every output

What is model drift? Model drift is when an AI workflow’s output quality changes over time, often after a model, prompt or automation update, without anyone noticing. Hatch’s defense is evaluations: a fixed set of known cases with known answers, run against every workflow and rerun after every change. “The worst thing is to have a model drift confidently.”

Where the panel disagreed: who owns evaluations?

O’Byrne questioned whether every firm should become an evaluation expert. He argued a good vendor should own that work inside its software, since many firms lack the bandwidth or skills. Hatch countered that a vendor knows the base rules but not a company’s nuances, so an internal person should own sign-off, much like governance. Early-stage firms, she added, often can’t afford specialized vendors for everything, and they’ll need that knowledge anyway to brief a vendor well.

The Practical Limits That Show Up First

Canada named two. The first is skipping a “phase zero”: organizing the documents and data AI will draw on before building anything. The second is assuming AI will fully automate a process. “It should scare companies to think about an agent doing something in your accounting system without some oversight.” Agents should help along the way, pulling and reading a document, then waiting for approval. Autonomy can grow over time, for example triaging a shared mailbox or preparing a morning brief.Kho tied it to data: agent sprawl (too many AI tools nobody tracks), hallucinations and runaway credit use often trace back to poor grounding, meaning the information AI draws is messy or out of date. Cleaning up the knowledge sources agents draw on, from shared drives to the ERP, is “the new version of garbage in, garbage out.”

Kho also flagged a naming risk. Products pitched as an “AI accountant” or “AI chief of staff” imply that no review is needed

An AI governance framework for finance teams

Hatch argued that governance that only says no creates shadow AI: staff using unapproved tools on personal accounts. The goal is a sandbox people will actually stay inside. Her framework has six parts:
1. Classify the data, not the tool Sort data into three tiers (public or anonymized, client-identifying, and regulated) and make sure everyone knows which is which. That gives people a “yes lane” to work in.
2. Keep an approved tools list with a fast path Check each tool’s training terms (whether your data train its models), data retention and sub-processors (other companies the vendor share data with). If approval takes six weeks, people will route around it.
3. Test in a limited environment Pilot tools in a protected setting before wider release, and retest after every update.
4. Strip sensitive data Remove personally identifiable information before it reaches a model outside protected team accounts.
5. Name sign-off points Every process needs documented review points. “The reviewer has to own the output, regardless of what is creating the draft.”
6. Train people “Governance that’s just on a document… that’s not governance.”

Shadow AI: why banning tools backfires Canada added that refusing to provide tools pushes employees to personal accounts, where data may train public models. He compared AI resistance to the move from on-premises to cloud, only faster. O’Byrne recommended starting with an enterprise ChatGPT or Claude account with limited access. Kho shared examples of mid-market companies without a company ChatGPT subscription discovering that 80–90% of staff were already using it on their own. He also described a company-wide Microsoft Copilot rollout that was switched off three months later because usage costs ran too high. His advice: watch usage as you scale.

When AI touches the ERP, bank data or payroll

Once AI can reach core systems, Canada said CFOs should evaluate human review, the quality of the data agents draw on,, and who will maintain and improve the agents. He expects IT administrators to become internal consultants who study each department’s processes. Permissions matter too: an assistant with broad file access can surface salary data through the back door.

O’Byrne’s design for keeping human judgment real: surface exceptions for the reviewer’s attention, require a named approval on everything, and use a separate AI to look for mistakes. People check more carefully when their name is on it.

How to Allocate a $500,000 Finance technology Budget

Kho closed with a scenario: a mid-sized company’s CFO gets $500,000 for technology. How much goes to AI?

Panelist	First Priority	Then
Canada	The systems stack: move tracking out of spreadsheets into databases, the ERP or CRM so agents can use them	Expert guidance, and possibly a person to own AI governance and evaluations; ROI may not arrive right away
Hatch	Core systems and data hygiene, or “you’re just sort of buying the faster way to wrong”	Senior reviewers, then AI last. “I’d spend more on people reviewing the output than the actual tools that are producing it.”
O'Byrne	An enterprise ChatGPT or Claude in staff hands, since they’ll use AI anyway	Cybersecurity, since AI is making cyberattacks cheaper and more frequents; then Hatch’s path

Kho’s summary of the common thread:

Start small, build the foundation of clean data and defined processes, run evaluations consistently, and educate people from the ground up.

This session is part of TFIF 2026. To see how AI fits alongside talent and team design, read Own the Core, Access the Rest: What The Future is Fractional (TFIF) 2026 Revealed About How Finance Teams Are Being Rebuilt what finance leaders are rethinking about talent, AI and team structure.

Stop automating a process nobody has written down.

AI only pays off on top of clean data and documented workflows. This is the work our finance and AI teams take responsibility for. We map and document your processes, clean up the data underneath, and keep a qualified reviewer on every output, so your team gets the speed of AI without giving up control.

EXPLORE HOW WE BRING AI INTO YOUR FINANCE FUNCTION

Frequently Asked Questions

Still have questions?

Consult With Us

Related Content

Blogs

Building Your Finance Team: Four Decisions for a Hybrid In-House and Outsourced Model

Building Your Finance Team was the closing conversation at The Future Is Fractional 2026, hosted by Scrubbed on September 24, 2026. Scrubbed’s Debra Andrews spoke with Ottavio Siani, a fractional CFO who also founded Triangle Coffee, which runs eight cafes across the Boston area and Washington, D.C. They walked through four decisions: whether to hire in-house or use a provider, how to vet an accounting partner, what to do in the first 90 days, and how to scale a hybrid team, plus where AI fits. The short version: divide work by context, vet partners on workload and industry fit, and expect in-house and outside capacity to grow together. Few finance leaders face a clean choice between building a team in-house and handing the work to a provider. Most end up with a mix. The harder questions are how to divide the work, how to choose a partner, and how to make the arrangement hold as the company grows. Session 5, the closing session of The Future Is Fractional 2026, a virtual conference hosted by finance and accounting advisory firm Scrubbed on September 24, 2026, took on those questions. Scrubbed Chief Marketing Officer Debra Andrews walked through four decision points with Ottavio Siani, a fractional CFO who also runs his own multi-location business, so he sees both the advice and the result. Key Takeaways: Divide work by context. In-house staff handle what’s new and unusual; a partner handles the routine. Vet partners on workload, industry fit and software knowledge, not just price. Four or five clients per accountant is a good answer, in Siani’s view . Be skeptical of full-automation promises. Several of his clients’ books went unclosed for months after trying them. Set cadences and communication rules early. Review cash weekly and the close monthly, and agree on response times. Bring the existing team along. Show them the work they’ll gain time for. Expect in-house and outside capacity to grow together. A provider scales in small steps; in-house hiring moves in big ones. Use AI, but keep a person accountable. AI can build the models; someone still has to give them a real read. What is a hybrid finance team? A hybrid finance team combines in-house finance staff with an outside provider, such as an outsourced or offshore accounting team or a fractional CFO. In Siani’s model, the in-house person supplies context and reviews the work, while the provider handles routine processing and produces reports. Meet the Speakers Debra Andrews (moderator): Chief Marketing Officer, Scrubbed. Ottavio Siani: Founder of Triangle Coffee, which runs eight cafes across the Boston area and Washington, D.C., and a fractional CFO for seven or eight years, serving more than 10 companies.  Decision 1: In-house or outsourced accounting? Usually Both Siani typically does both. Any finance support that isn’t full-time inside the company still needs someone inside to explain what’s happening day to day. He used his cafes as the example. A recurring charge from the same milk supplier needs no discussion; the offshore team knows the vendor from a shared vendor list and books it correctly. But a new coffee roaster is different. The name may not say what the company does, and the purchase might be an experiment that belongs in a research budget rather than cost of goods sold. Someone on-site has to explain it. How Siani divides the work In-House Finance Staff Outsourced or Offshore provider Transactions New vendors, new charges, anything new or unusual Recurring charges from known vendors on a shared list  Context  Explains what a purchase is for and where it belongs Books it correctly once the context is clear Reports Reviews them Produces them As the company grows An in-house controller may read the provider’s reports and design processes  Scales its time with the business That split gets more leverage out of the in-house person's time. “They’re not booking the debit and credit. Instead, they’re just dealing with the new vendors, the new charges, anything new that’s happening,” while the offshore team, in his case Scrubbed, handles the rest. Decision 2: How to Vet an accounting partner Across more than 10 fractional CFO clients, Siani often works with outside accounting firms, and has run the selection process many times. Accounting partner vetting checklist Question to Ask What Siani looks for How many clients does each accountant handle? Four or five is a good answer. He once heard 20; the team did good work but was grinding uncomfortably hard. Have they worked in your kind of business? His clients usually sell physical products, which brings inventory, payables, receivables and working capital challenges that software-only experience may not cover. Can they advise on software? Firms that work with many companies know which inventory or accounting tools fit which situations. How does the price compare with a full-time hire? He always runs that comparison, and it eliminates some options quickly. Are they promising full automation? A warning sign. On automation promises: AI tools are getting closer to automating larger portions of finance work, he acknowledged. But three or four of his clients signed up with companies promising fully automated accounting. It was “a disaster.” Books went unclosed for months while they tried to raise money, and he was brought in to clean up. “I’d be skeptical of any promise that… somebody will fully automate your accounting.” Decision 3: The first 90 days with a new finance partner It depends on the starting point, Siani said. Onboarding a partner at a company with little prior activity is straightforward. He also joined one company, working alongside Scrubbed, that had nine months of unclosed books and a botched attempt to consolidate two entities. There, the priority was indicative reports that gave executives a handle on the business, not a perfect close. Assuming no crisis, his foundation is cadence: Weekly, focused on cash: where it’s coming from and where it’s going. Monthly, reviewing the close. In both meetings, a roadmap and a prioritized issue list. Not everything gets fixed in the first close. If inventory balances are the biggest worry, start there and leave a mostly accurate receivables process for later. Agree on communication rules Companies and firms have different communication styles. Some companies only move when someone writes in all caps; a third-party firm sending polite emails and waiting may not realize that. Sometimes the fix is as simple as an agreed subject line: 'URGENT,' with a clear deadline and what happens if it's missed. So Siani sets explicit rules: Channel Expected Response Email   Within 48 hours Email marked urgent  Within 24 hours Messaging app  As soon as possible Bring the existing team along In-house staff often worry when a provider arrives. Occasionally, Siani acknowledged, someone is underperforming and being replaced. More often, the existing team is simply maxed out. He calms the nervousness by laying out the full list of work, showing it’s too much for the current team, and dividing it: “You will be reviewing this report. You will not be producing the report.” Convincing people that the provider frees them to focus on what they uniquely can do usually takes a couple of months. Decision 4: Scaling a Hybrid Finance Team Outside capacity scales more smoothly. A third party can scale “more linearly,” Siani said. Hiring in-house moves in steps, and part-time hires are hard to find: bring someone on, carry extra capacity until the company grows into it, then run short and hire again. A provider lets you add time as you grow. The two tend to grow together. He hasn’t yet seen a hybrid model outgrow its third-party resource at the small and mid-sized companies he works with. At some companies he has hired an in-house controller who then reads the provider’s reports and helps design the processes behind them. Large companies do this too, he noted, often using third parties for parts of the finance operation such as accounts receivable. AI for fractional CFOs: Faster models, Same accountability A person stays responsible. Siani agreed with the consensus from earlier sessions: a person must stay responsible for the output. Because AI tools can be so easy to work with, “it’s easy to kind of get a little lazy and not review something.” People signing off on information need to feel responsible enough to give it a real read. A year of progress.. He works mainly with Claude and says it has improved dramatically since a year ago, when the common complaint was that AI made things up and couldn't be trusted.. In one engagement, he spent nine months rebuilding a client’s finance function after a bad experience with previous providers. Much of the work was building financial models for a fundraiser. “I didn’t type a single formula into Excel.” He structured the models, commented on them and built in quality checks while the Claude did the building. A year earlier, he said, he’d have told you that building Excel models is what most people pay a fractional CFO for. “It’s a tool that’s too powerful to be ignored.” Across all four decisions, Siani's answer was rarely either-or. Keep the context and the review inside the company, let a partner carry the volume, and let both grow as the business does. This session is part of The Future is Fractional 2026.  For how team design connected every conversation at the event , read Own the Core, Access the Rest: What The Future is Fractional (TFIF) 2026 Revealed About How Finance Teams Are Being Rebuilt   what finance leaders are rethinking about talent, AI and team structure.

Data, Cash and Deal Readiness: Finance Lessons for Scaling Technology Companies

The Technology Spotlight was one of three industry spotlights at The Future Is Fractional 2026, hosted by Scrubbed on September 24, 2026. Anthony John Rogador, Accounting Advisory Services Manager at Scrubbed, spoke with Abdul Wahab Zafar, SVP of Finance at Studycast, who has led M&A integrations and financial systems consolidations at fast-scaling companies. They covered building a single source of truth, rationalizing the tech stack, forecasting usage-based revenue by cohort, managing cash with payment terms, and preparing a back office for M&A. The short version: tie customer, contract and billing data together, hold payment terms firm, and operate as if you might sell tomorrow. Fast-growing technology companies tend to add tools faster than they connect them. Finance ends up consolidating data from systems that don’t agree, forecasting revenue that won’t sit still, and hoping the back office holds up if an acquirer calls. At The Future Is Fractional 2026, a virtual conference hosted by finance and accounting advisory firm Scrubbed on September 24, 2026, Session 4 split into three industry breakouts running at the same time: real estate, nonprofit and technology. In the technology track, Scrubbed Accounting Advisory Services Manager Anthony John Rogador spoke with Abdul Wahab Zafar, SVP of Finance at Studycast, who has run M&A integrations and systems consolidations at fast-scaling companies. Key Takeaways: Accurate financials start upstream. Customer, contract and billing data need a common key tying them together. 90–95% data alignment is often enough. The last last few percent can cost more than it’s worth. Forecast usage-based revenue by cohort . In Zafar's experience, a cohort’s second year grows fastest, then growth settles to low single digits. Protect cash with payment terms. Hold customer terms firm and negotiate longer vendor terms to create a cushion. Operate as if you’ll sell tomorrow. Buyers judge revenue by how well customer, contract and billing data tie out. Don't rush to merge systems after a deal. Move acquired customers into the process that already works. Meet the Speakers Anthony John Rogador (moderator): Accounting Advisory Services Manager, Scrubbed. [LINK: LinkedIn] Abdul Wahab Zafar, MBA, CMA : Senior Vice President of Finance, Studycast, with experience leading M&A integration, financial systems consolidation and scaling finance operations at fast-growing companies .  Building a Single Source of Truth upstream Finance, Zafar said, takes data from everywhere else and consolidates it. So accurate financials start 'on the head of the pipe,' with customer data. He thinks of each data set as a 'cube,' and the order matters: customer first, then contract, then billing, then vendors. Data Set The Question To Answer Customer Data  Who is buying from you, and is that record correct? Contract Data  What exactly did they buy? Billing Does billing reflect the same customer name and contract terms? Vendor Data Who do you pay, under which contracts, against which invoices? When those line up, he said, the financial data naturally becomes accurate and trustworthy. Standardizing data across disconnected systems Growing businesses add the tool they need and forget to connect it. Zafar’s approach: Define the North Star for your data, then break it into the steps needed to get there. Get leadership aligned. If data alignment isn’t a priority for sales or customer teams, finance can only do so much. Create a common hook. Every record in the customer, contract and billing “cubes” should share a key so anyone can trace one customer across all three. Decide whether the last 5% is worth it. Reaching 90–95% alignment is often achievable. The final 5% may require retooling working processes or systems that already work, and in his experience is often less cost-effective than working around it. Rationalizing the tech stack before adding tools New finance tools launch constantly. Zafar’s discipline before buying one Identify the actual problem you’re solving, in depth. Check whether your current systems can solve it . Many companies use only part of what a platform like NetSuite can do. Only then evaluate third-party options, with a clear picture of what you need. Keep it simple. The fix is sometimes the smallest tool on the market and don't create processes 'just to make it look like you're doing a good job.' Complexity has a staffing cost. “If it’s more complicated processes, then a lot of times you need a bigger team to manage those processes,” and most small and mid-sized finance teams run lean. Forecasting Usage-based Revenue by cohort Usage-based revenue billed on how much customers use the product, not a flat fee) shifts with business days, holidays and seasonality. As Zafar put it, 'It's not as simple as, the customer signed a $12,000 contract, my revenue is $1,000 a month.' Studycast groups customers into cohorts by the year they went live, because cohorts tend to behave consistently over time. Cohort Year Typical growth, in Zafar's experience Why Year 1  Partial year Customers go live partway through the year Year 2  Roughly 70% to 100% The first full year of revenue Years 3-4 onward  Around 4–5% a year Growth declines sharply, then settles Forecasting each cohort separately is more reliable than projecting the whole customer base at once. Separating enterprise and small-business customers to handle outliers The outliers at Studycast, single customers growing 100% or more a year, were mostly enterprise customers who went live at one site, then rolled out to others, compounding revenue month after month. The team now forecasts SMB and enterprise customers separately. Finance can’t forecast enterprise accounts alone, he added. Customer success and sales know whether an account is at capacity or has room to expand, which turns the forecast from “just a number on the paper” into a strategic decision. AI in Finance: Faster automation, Higher stakes for Data integrity Zafar sees AI making finance both easier and harder. It already helps automate bank and account reconciliations, reporting and financial decks. But deeper uses, such as combining data points to analyze a customer fully, depend on data that lines up across systems. Without that common hook, “it’s essentially garbage in, garbage out,” and AI may point you the wrong way. His team has also implemented third-party tools that automate accounts receivable (AR) and accounts payable (AP), significantly reducing processing time. They aren't cheap, he said, but they're 'not the Mercedes of the brands.' He spends two to four hours a week learning AI himself and encourages his team to practice and fail with it. “If you don’t do it, you’re gonna get left behind.” Managing Cash Flow with Payment Terms Predictable cash starts with predictable payment terms. “You live and die with your collections.” Net 20 means payment is due 20 days after the invoice date. Customer Terms (receivables) Vendor Terms (payables) Target   Net 20     Net 45, sometimes settling at net 30 How it’s held    Sales can’t change terms without finance approval; about 95% of customers are on net 20 Negotiate everything with vendors Exceptions Mostly larger customers or trades for something else Varies by vendor Support Dedicated people or systems chase late payers AP automation reduces processing time At those terms, collecting at net 20 and paying at net 30 to 45 works out to a cushion of 10 to 25 days. Not every customer pays on time, he acknowledged. The goal is for most (60–70%) to pay promptly and to have people or systems chasing the rest. M&A Due Diligence: Evaluating a Back office Zafar has been on both sides of deals. As the buyer, he starts with the P&L and balance sheet, but he spends most of his time on customer data. He pulls customer, contract and billing data and tries to merge them. If they tie out easily, revenue evaluation is straightforward. If not, the gaps tell him how much to trust the reported revenue. Most of the time the revenue is real, he said; the hard part is pinpointing the contract and customer behind it. As the seller, he makes sure the customer-to-billing chain is as accurate as possible, even if it lives in Excel, so the company can show the strength of its data. Just as important, finance, sales, customer success and the CRO should all tell the same story. Post-acquisition integration: don’t rush the systems The most common trap, Zafar said, is rushing to merge financial systems. The better move is to understand how data flows from customer to contract to billing, then move the acquired customers into your existing process. “If we are acquiring a business, we’re acquiring revenue.” Don’t break a process that works just to force systems together. Before closing, he said, the one thing finance leaders should understand is the target's billing process, from identifying a customer to signing a contract to sending an invoice. Finance is usually more involved before a deal closes, so it can then guide sales and revenue operations on how to bring that data across. Vendor data, by comparison, is 'a pretty easy lift.' M&A readiness checklist, drawn from the session Can every customer be traced across customer, contract and billing records with one common key? Would your customer-to-billing data tie out if a buyer tried to merge it? Do finance, sales, customer success and revenue leadership tell the same story about customers? Are payment terms consistent, and are exceptions approved and documented? Before closing a deal, do you understand the target’s billing process end to end? Rogador’s closing thought: Getting financial systems and data right doesn’t start when the deal shows up. It’s what makes you ready when it does. This session is part of The Future is Fractional 2026. For how deal readiness connects to the wider shift in finance team design, read Own the Core, Access the Rest: What The Future is Fractional (TFIF) 2026 Revealed About How Finance Teams Are Being Rebuilt   what finance leaders are rethinking about talent, AI and team structure.

Building a Resilient Nonprofit Finance Function When Funding Is Unpredictable

The Nonprofit Spotlight was one of three industry spotlights at The Future Is Fractional 2026, hosted by Scrubbed on September 24, 2026. Laurence Ruelo, Director of Business Development at Scrubbed, spoke with Costa John of CFO Assignments, who has decades of CFO experience across nonprofit and for-profit organizations. They covered the three pillars of financial resilience, tracking restricted funds, what to do when a major funder pulls out, board reporting, and how to build finance capacity without overbuilding. The short version: watch unrestricted cash, act early and in phases when funding is cut, and give your board the forward look rather than the history. For many nonprofits, the funding hasn’t disappeared, but its timing has become harder to predict. Grants arrive late. Proposals take longer to secure. Visibility six or twelve months out is thin. At The Future Is Fractional 2026, a virtual conference hosted by finance and accounting advisory firm Scrubbed on September 24, 2026, Session 4 split into three industry breakouts running at the same time: real estate, nonprofit and technology. In the nonprofit track, Scrubbed Director of Business Development Laurence Ruelo asked veteran CFO Costa John how nonprofits can build a finance function that is resilient, useful to leadership and scalable when resources are already stretched. Key Takeaways: Resilience rests on three legs: liquidity, flexibility and the cash cycle. Timesheets and integration drive restricted fund tracking. Programs are delivered by people, so staff time is what releases restricted funds. When funding is cut, act early and in phases. The first cost reductions are the cheapest ones. Don't hire strategy to fix bookkeeping. Get the books closing cleanly before bringing in a fractional CFO, or the CFO gets pulled back into reconciliations. Give boards the forward look, not the history. Show the outlook, its impact and a range of decisions. Watch operating cash flow, and show days of cash. “Reserves and profits are a matter of opinion. Cash is a matter of fact.” What is nonprofit financial resilience? It’s an organization’s ability to keep delivering its mission when funding is delayed, reduced or withdrawn. In John’s framing, that depends on having unrestricted funds(money not tied to a specific grant or purpose) available, enough cash on hand to absorb delays and a clear view of how long the organization must pay its costs before funding arrives. Meet the Speakers Laurence Ruelo ( moderator): Director of Business Development, Scrubbed. Costa John : CEO CFO Assignments, with decades of CFO experience across nonprofit and for-profit organizations, including as CFO of Sapiens North America.  The Three Pillars of Nonprofit Financial Resilience John frames resilience as a three-legged stool: Liquidity: The level of unrestricted funds the organization can draw on. Flexibility : The cushion to absorb a delayed grant or a slow proposal. The board usually sets it with the CFO as a target number of months of cash operating expenses. The cash cycle: How many days the organization must cover rent, salaries, travel and office costs before funding arrives. Together, they give the organization a stable footing. John acknowledged it’s hard for small nonprofits, but said building toward it should be a priority. How to Track Restricted Funds without the Spreadsheet Wrestling Match Restricted funds are grants or donations that can be spent only on a specific purpose or program. Start with timesheets. Nonprofit programs are delivered by people, so releasing restricted funds and matching expenses to grants comes down to how staff time is spent. “If you cannot have a good on-hand grip of where and how your program people are delivering against the mission, you’re already on a slippery slope." Integrate the systems. The more the organization’s systems connect end to end, the less restricted revenue tracking becomes “a wrestling match between confusing and sometimes broken spreadsheets.” Affordable, easier-to-integrate software has made that more realistic than it used to be. Contingency planning when a major funder pulls out Stay close to funders. Organizations with strong funder relationships can often see cuts coming, whether funders planned them or are reacting to their own budget pressures. Have the outline of a plan ready. John described a phased response, from least to most disruptive: Delay contractors and capital spending not required this year. Delay replacements when people resign. Delay salary increases. Delay benefits such as 401(k) matches. Only if those steps don’t buy enough time, consider reductions in force. Act early. Once you know funding is going away, move. “Your first cost reduction steps are ironically your cheapest ones.” Hesitating can make the eventual response more painful and more expensive. Board Reporting: Forward-looking Information, Not just data “Data is not the same as information,” John said. Boards can drown in numbers. He recommended spending little time on historical statements beyond a status update. “It’s hard asking a board to govern an organization looking in the rearview mirror.” Include in Board Updates What It Means The forward look The outlook for the coming period The impact If this happens, here's what it means financially Decisions,plural A range of choices and their impacts, not a single option to approve or reject Non-financial metrics, with the CFO at the table John pushed back on the idea that non-financial metrics are outside the CFO’s lane. Measures like schools reached, children served or outcomes achieved all have financial drivers underneath, and the CFO can help program leaders see the dollar value of their impact. He sees a missed opportunity with funders: payback . In one illustration, a program costing about $72,000 per school produces an estimated $112,000 per school in lifetime health and education benefits. Framed that way, funders hear two things: the program pays for itself, and more money would do more good. Aligning boards on cash reserves Reserves aren’t shareholders’ equity. John warned that board members without a finance background shouldn't read nonprofit reserves that way. Restricted reserves reported under GAAP are, in economic substance, obligations to deliver future programs, and the cash to deliver them needs to be available. Deficits change the math. With $10 million in grant revenue and $12 million in expenses, every dollar of programming costs $1.20. Delivering more of it without a plan to close the gap can dig the hole deeper, so healthy-looking reserves can offer “a comfort that turns out to be razor thin.” Boards don’t need the calculations, he said, but they should know to ask. Scaling Finance Capacity: Compliance vs. Value-added work Executive directors already juggle fundraising, programs, people and operations. John divides finance into two parts, with different ways to resource each: Non-negotiable: compliance and reporting Value-added: strategy and planning What it covers The Form 990, the audit, paying vendors, payroll Forecasting, budgeting, support at the fundraising and budget tables The risk It “can become a self-consuming monster” without integration and automation It gets crowded out if the CFO is pulled into bookkeeping How John would resource it Outsource it, or at minimum keep it separate from the CFO role A part-time or fractional CFO, kept out of the bookkeeping weeds Don’t hire strategy to fix bookkeeping. Bringing in a fractional CFO for strategic work while the books still don’t close cleanly means the CFO gets dragged back into reconciliations, “the most expensive way to solve the wrong problem.” Messy reconciliations also pull program leaders away from the mission. Start with the lightest system that works . John warned against buying the most sophisticated accounting system too early. A lightweight option such as QuickBooks with nonprofit modules usually serves growing organizations better, while complex systems can “grind the process to a halt.” Ruelo agreed, noting organizations that implement complex systems, don’t get the information they need and end up returning to simpler tools. Non-financial metrics, with the CFO at the table A dedicated CFO role, even a part-time or fractional one, onshore or offshore, kept out of the bookkeeping weeds . Relationship-driven roles such as the grant accountant, who knows program leaders, funders and nuances that may not translate if outsourced. The metric every Nonprofit leader should watch: Operating cash flow “Reserves and profits are a matter of opinion. Cash is a matter of fact.” A balance sheet can look healthy while cash is tied up in grants not yet funded, and “you cannot make payroll with grant certificates or invoices.” John recommended practicing short-horizon cash forecasting, starting at 30 and 60 days, until it becomes muscle memory, especially as the funding environment for nonprofits shifts. One Change to the Board packet: An Unrestricted Cash Schedule What to cut: Liability-side schedules, such as an accounts payable aging report showing how long vendors have waited to be paid. John called that too operational to help a board make a decision. What to add: a schedule showing how many days the organization can run on unrestricted cash. Start with unrestricted reserves. Subtract amounts already committed. Divide by daily cash expenses. The board then knows, at every meeting, how many days the organization can operate: 111 days, say, or 192. It's easy to focus on and easy to understand. Ruelo’s summary of the conversation:   Resilience starts with visibility, good reporting translates information into decisions, and finance capacity is about access to the right expertise at the right time, whether internal, fractional, outsourced or a mix. This session is part of The Future is Fractional 2026. For the themes that ran across every industry, read Own the Core, Access the Rest: What The Future is Fractional (TFIF) 2026 Revealed About How Finance Teams Are Being Rebuilt   what finance leaders are rethinking about talent, AI and team structure.