5 forecasting and spend control practices for finance teams

The short version

The budget you locked in January is already wrong by February, and the CEO is editing the master spreadsheet. Meanwhile, recurring SaaS spend keeps growing without sign-off, and renewal quotes land on your desk with no benchmark to push back against. The reactive approach that worked at 10 people falls apart somewhere between 25 and 150, and the longer you wait, the harder it is to catch up.

Matthew May is president of Acuity Accounting and a "recovering public accountant" who advises roughly 600 companies. He shares what's working for finance teams in that 25-to-150-person range: three-tier scenario planning, department-level card controls, and a pricing intelligence approach that flagged a Slack quote running 48% above what other Ramp customers pay.

Most useful for: Finance leads, controllers, and fractional CFOs at 25-to-150-person companies who still run forecasts in spreadsheets and rely on expense reports to catch out-of-policy spend.

This webinar is CPE eligible.

Why does the annual budget go stale so fast?

The annual budget goes stale because most teams treat the budget, the reforecast, and the live model as one document. Separating them into three versioned documents is the single most important step to keeping your forecast useful.

Most finance teams at 25-to-150-person companies build a budget in December, lock it in January, and watch it diverge from reality by mid-February. Customer acquisition pace shifts, a hire slips, or a renewal lands higher than you modeled, and suddenly the budget is reporting against a snapshot that no longer matches reality.

The fix is version control. Keep a locked plan of record, a separate quarterly reforecast, and a live model that updates as actuals come in. Each has its own job, and none of them overwrites the others. That single discipline is what makes everything else possible.

5 forecasting and spend control practices

1. Run a budget, a quarterly reforecast, and a live reforecast in parallel

You should run three concurrent versions of your forecast:

• A locked budget for plan-of-record reporting
• A quarterly reforecast that resets every 3 months
• A live reforecast that updates continuously as actuals come in

Jirav lets you star each version and mark it reportable, which eliminates the "model_v6_final_final" spreadsheet problem. The budget gets measured against actuals for the full year. The quarterly reforecast captures what's materially changed (sales pace, churn, hiring), and the live reforecast is the working model you use to make decisions in any given week.

This is about making sure everyone's looking at the same numbers. When the CEO asks how the business is tracking, you have one answer, and when the board asks the same question, you have the same answer. When a department head asks what they have left to spend, you point them at the live reforecast rather than a version someone forwarded 3 weeks ago.

In Matthew's experience, it's not the math that breaks teams — it's having too many versions of the same file.

"The biggest nicest thing about spreadsheets, it can be a million things. The worst thing about spreadsheets is the CEO gets in it and screws it up all the time. And then the balance sheet doesn't fit, and then you're mad about what's going on... It's the version control thing is amazing once you get into the stage that most of you guys are talking about in that twenty five to one hundred and fifty person company."

2. Forecast off the GL, the CRM, and the HR system

If your forecast pulls only from QuickBooks, Xero, or NetSuite, you're missing the inputs that move the numbers. You need data from three places at once:

• The GL (QuickBooks, Xero, NetSuite) for P&L and balance sheet actuals
• The CRM (HubSpot or Salesforce) for pipeline and customer acquisition pace
• The HR system (Gusto, ADP, or similar) for headcount at the individual level, including role, department, start date, and wages

QuickBooks tells you total staffing expense, but it doesn't tell you which person, which role, or when they started. That detail is what makes a headcount-driven forecast work. Your model can project the next hire by triggering off customer count or pipeline volume rather than a flat assumption.

Matthew found that the biggest unexpected outcome when clients moved from spreadsheets to a tool like Jirav wasn't the math. Business users started thinking about the business the way the forecast was set up — which was completely different from their monthly reports. That shift ended up changing how the firm handled month-end reporting.

"The biggest unexpected outcome was that the business users thought about the business the way the forecast and the model set up. And that was a total disconnect from the data they were getting reported monthly out of QuickBooks, Xero, NetSuite, any of the sources of truth. If you go through this exercise of helping forecast and you go through the pain of linking it to your chart of accounts, you're gonna see that it's actually gonna change your month end reporting probably to something more sensible to the business user."

3. Build dashboards around the operator's top 10 metrics

Your dashboard should answer the 10 questions your CEO or department head is already asking, not reformat the P&L. Matthew's favorite question to ask is what those 10 things are and why they matter. Once you have those answers, build the dashboard around them:

• Visual tiles for each metric
• Live GL feeds, with the option to show only closed months so unclosed periods don't show up as misleading drops
• Department-level views so each owner sees a dashboard tailored to their own priorities

This matters because most operators ignore the monthly P&L. It feels like an audit, arrives in an email, and lands in a format nobody asked for.

The top-10 conversation flips that. Now the dashboard answers the questions operators are already asking, in the cadence and format they want, and that's how reporting becomes a tool people actually use.

The right build is incremental. Ship a few tiles, get feedback, and add tiles over time, because trying to ship all 10 on day one buries the signal.

"It's my favorite question to ask people. It's the richest discussion that you'll have with the business owner or CEO or the department heads is getting them to start talking about the ten most important things to them and why. And then that kinda drives your thinking from the finance chair to, how can we capture that data? How can we get that data on real time? What's the cadence they're gonna need it? How can we help? It reframes a lot of stuff for you."

4. Split corporate card spend onto 2 cards per department

Give each department head two cards: one for one-time charges and one dedicated to recurring software subscriptions, with its own budget and vendor parameters.

The reason for the split is visibility. When recurring SaaS lives on the same card as travel, lunches, and one-off purchases, it's hard to see what you're actually spending on subscriptions.

Subscription creep is the spend pattern most likely to slip past you. A department adds seats, upgrades a plan, and you don't find out until the bill spikes. Removing licenses when employees leave only works if you can see what's running.

The recurring card gives you a single source by department of every recurring license. Here's how to set it up in Ramp:

• Create a virtual card
• Assign the department owner
• Load the vendor and contract value
• The card only authorizes charges that match

"At the department head level, what we're doing is like a two card system. So they have like their one timers that they can monitor, but we're putting their recurring subscriptions like sales and marketing recurring subscriptions in a different one... You're gonna want insight and laser focus on that subscription one because those get out of control. Those ones you have to know, when people leave, how to deprovision those kind of things and ask those questions, and you can have a single source by department of the recurring licenses."

5. Benchmark every software contract before you renew

Before you renew any software contract, you can check whether you're overpaying. Ramp's pricing intelligence pulls from thousands of real contracts — for Slack alone, the tool drew on 21,254 data points. Tyler entered a hypothetical quote of $13 per license across 12 licenses ($156 per month), and the tool flagged it as 48% higher than what comparable Ramp customers pay nationwide. LinkedIn pricing in the same view had 2,500 data points.

You can use it two ways. You can manually enter the quote to benchmark before signing, or you can upload the contract PDF directly, let Ramp extract the data, and review the benchmark in-app. If you work with an accounting firm, they can run this check from Ramp's advisor console on your behalf.

When you walk into the renewal call with a benchmarked number, the conversation changes, because the vendor knows you have the data, and the price you accept stops being whatever they quoted first.

"Based on all these other data points and other customers on ramp paying for Slack, it will benchmark your quote or your price or your contract value against everyone else on ramp paying for Slack. So you can go back to that vendor and say, hey, that's not gonna work for me. You guys are kinda ripping me off. Forty eight percent higher than everybody else on ramp nationwide. I don't think that's a very good deal."

How do you apply Matthew's advice this quarter?

Three things you can do this week:

1. Lock your current forecast as the plan of record: Save a copy of your most recent forecast, rename it as the locked version for the current year, and share it as read-only. That one step gives you a clean reference the next time someone asks how you're tracking against budget.
2. Ask one department head for their top 10 metrics: Schedule a 20-minute conversation. Ask what 10 things they're tracking and why. Those answers become the dashboard spec for that department.
3. Move recurring SaaS spend onto a dedicated card per department: Pull a list of every recurring subscription from your credit card statements or accounting system. Issue one virtual card per department for recurring software, with the vendor and contract terms loaded.

Final thoughts

"The biggest unexpected outcome was that the business users thought about the business the way the forecast and the model set up. And that was a total disconnect from the data they were getting reported monthly out of QuickBooks, Xero, NetSuite, any of the sources of truth... It actually forced a conversation with the CEO and c level and business user to where then that ended up changing how we did our monthly reporting."

You get more leverage when your financial model reflects how your business actually runs. Lock your forecast versions, connect your live data sources, and set spend limits at the card and contract level so you catch problems before they hit your books.

See how Ramp fits in

Your forecast is only as good as the spend data behind it. Ramp helps you set spending rules at the card level, track recurring software on dedicated virtual cards, and feed real-time spend data into your live forecast.

Common questions about forecasting and spend control

How often should we actually reforecast?

Quarterly at minimum, with a live model that updates continuously as actuals close. The locked annual budget is for plan-of-record reporting, the quarterly reforecast is where you adjust what's changed, and the live model is the working number you use day to day.

If you only have one document doing all three jobs, you're guaranteed to overwrite something you needed to keep.

What data sources should we connect first?

Start with the GL, the CRM, and the HR system. Your GL tells you the total but not the breakdown, your CRM tells you pipeline pace, which drives revenue assumptions, and your HR system tells you headcount at the individual level, which is usually your largest expense line.

Connecting those three eliminates most of the manual copy-paste work that breaks spreadsheet forecasts.

What is the fastest way to find out if we are overpaying for software?

Upload your contract or latest invoice to Ramp, and you'll see how your price compares to what other companies are paying. The Slack demo pulled 21,254 data points and flagged a quote as 48% over market, and the renewal conversation changes when you have the number before the call.

How do vendor-specific virtual cards prevent surprise price hikes?

A vendor-specific virtual card only authorizes charges from the assigned vendor at the assigned contract value. If a vendor tries to charge a higher amount on renewal without notice, the transaction declines, which forces the renegotiation conversation before the charge clears rather than after.