I’ve sat through enough FP&A demos to know when a sales rep is hiding something. Most finance professionals watch a demo like a movie. They sit back, admire the dashboards, and get impressed. That mistake can cost your company six figures and your credibility. So I’m giving you twelve questions that flip the demo from a sales pitch into a due diligence session. These are the exact questions I wish someone had handed me early in my FP&A career.
WHY: Why Most FP&A Software Purchases Fail
Gartner research finds that 7 out of 10 finance transformation projects deliver slow, underwhelming results. Seven out of ten. And the most common reason isn’t bad strategy. It’s software that couldn’t do what you needed, and you found out after you’d bought it.
The FP&A software market is on track to more than double, from around $4 billion today to nearly $10 billion by 2032. Every tool fighting for your budget has a polished demo team. And a demo is designed to show you what the software can do, never what it can’t. The sales rep controls the data, the scenarios, the narrative. Everything works because they’ve rehearsed it a hundred times.
Your job in a demo is to find the cracks. Because once you sign, the cracks become your problem, and the vendor may walk away clean. And if you recommended the tool, your credibility takes the hit, not the sales rep’s.
Most teams still run on Excel, and that’s fine until you hit multiple entities, frequent reforecasting, and cross-functional collaboration. That’s when isolated workbooks turn into version chaos and reconciliation headaches. Modern FP&A platforms promise integrated planning, real-time collaboration, and AI-assisted forecasting. But those benefits only show up when the tool matches your bottlenecks, your data, and your culture. A misaligned platform doesn’t replace Excel. It just adds another layer of frustration on top of it.
So here are the twelve questions that find those cracks before you commit. By the end of this article, you’ll have a checklist that can save your company six figures and protect your credibility.
Not sure which tools deserve a demo in the first place?
Take our free Find Your Ideal FP&A Tool quiz. Answer six quick questions about your team, stack, and budget, and we score every tool in the directory against your profile and hand you a ranked shortlist in 60 seconds.
WHAT: Integration and Data (Questions 1–3)
I’ve organised these into four categories: integration and data, core capabilities, AI and automation, and total cost of ownership. The first three questions are about the data layer, because if the data doesn’t flow, nothing else matters.
Question 1: How does your integration with our specific ERP actually work?
This is the single most important question you can ask, because most vendors will say yes, we integrate with SAP, Oracle, NetSuite, whatever. But the word integrate is doing a lot of heavy lifting.
You need to dig deeper. Ask them whether it’s a native API connector, or whether it requires a middleware tool, and whether it needs manual CSV exports. Ask what the sync cadence is: real-time, hourly, or daily.
A native API connector that refreshes daily is fundamentally different from a solution that requires you to export a CSV file, manually map the fields, and upload it every night. And if that CSV sync fails silently, which happens more often than vendors admit, you’re making decisions on stale data.
Ask them to show you the integration setup live, not a screenshot and not a pre-recorded video. Watch them connect to your actual ERP system and see what happens. If the vendor can’t clearly explain what objects and fields are synced, or if they say that depends on your implementation partner, that’s a red flag.
Question 2: What happens when a schema change happens in our source system?
Here’s a scenario that plays out all the time. Your accounting team adds a new cost centre, or renames a GL account, and suddenly your FP&A tool stops pulling data correctly. Nobody notices until the monthly review.
Ask the vendor whether renaming a GL account in your ERP tomorrow would break the integration or cause it to adapt automatically. Some vendors handle this gracefully. Others require a support ticket, a manual remapping, and days of downtime. The difference between these two scenarios is the difference between a tool that works and a tool that creates more work.
If the vendor says schema changes shouldn’t happen often, or we’ll handle that during implementation, they’re telling you it’s going to be a problem.
Question 3: Can you show me how the platform unifies data from different sources into one metric?
Your ERP data alone isn’t enough for good FP&A. You need CRM data, HRIS data, billing data, and operational metrics. The real test is whether this platform can combine them.
Many tools can pull from multiple sources. Far fewer can actually combine that data into unified metrics, like revenue per FTE or customer acquisition cost to lifetime value ratio, without requiring a data engineer to build custom pipelines.
Ask them to build a custom KPI live. Pick a metric that matters to your business. Watch them create it using data from two different systems. If they need to export to a data warehouse or a BI tool to do the join, that’s not a unified FP&A platform. It’s a reporting tool with extra steps.
If the demo requires a separate data warehouse or SQL queries to combine data, you’re buying a visualisation tool, not an FP&A platform.
WHAT: Core FP&A Capabilities (Questions 4–6)
Once you know the data flows in correctly, the next question is whether the platform can actually do FP&A work. This is where I like to make sales reps nervous.
Question 4: Can you rebuild one of our models in your platform right now?
This is my favourite question. Before the demo, take one of your actual financial models. It doesn’t have to be your most complex one. A simple revenue forecast or expense model works. Send it to the vendor beforehand and ask them to rebuild it in their platform during the demo.
A good platform should be able to build a simple model in 30 minutes or less with user-editable logic, not hard-coded formulas that require vendor support to change. Watch how they build it. Ask whether it uses familiar formula language, or whether they need a proprietary scripting language that only certified consultants understand.
If they refuse to do a live rebuild, or they say we’ll need to customise that during implementation, that’s your answer. The platform can’t handle your specific modelling needs out of the box.
Question 5: How does your three-statement model work? Specifically, show me what happens when I change a revenue assumption.
Three-statement modelling is the backbone of FP&A. Any modern platform should support it. The question is whether it’s automated or manual.
Here’s the test. Update a revenue assumption. Say, increase your Q3 forecast by 10%. Then ask to see the impact on the balance sheet and cash flow statement. If the changes flow through automatically, you’re looking at a properly integrated platform. If they don’t, if the statements update independently, you’re looking at three separate spreadsheets dressed up in a nice interface.
Manual reconciliation between statements defeats the entire purpose of moving away from Excel. If your team still has to manually check that the balance sheet balances after every forecast update, you haven’t saved any time. You’ve just added a new tool on top of the old problem.
If the vendor says those statements update overnight, or that’s handled during the monthly close, the integration isn’t real-time, and you’ll be making decisions on disconnected data.
Question 6: Can you run multiple scenarios side by side and compare them?
Scenario planning is where FP&A software should shine. But not all scenario capabilities are created equal, and some vendors claim they offer multi-scenario modelling when what they actually provide is different versions of the same model. That’s a workaround, not a feature.
A real scenario capability lets you create an unlimited number of parallel scenarios, change assumptions in one, and see the impact across all of them instantly. Ask them to create three scenarios live. Add a new driver to one scenario, say a change in headcount growth rate, and confirm that the change applies across all scenarios. Then request a comparison view with all three scenarios side by side.
If they can only compare two scenarios, or if creating a new scenario requires duplicating the entire model, the platform isn’t built for scenario analysis, and you’ll end up exporting to Excel to do the real work.
WHAT: AI and Automation (Questions 7–9)
Question 7: What AI capabilities are actually available today, not on your roadmap?
This is where the hype is thickest. Every vendor claims to be AI-powered. But there’s a big difference between AI features that are generally available, AI that’s in a beta program, and AI that’s only on the roadmap for next year.
Before the demo, go through the vendor’s website and marketing materials. Note every AI feature they claim. Then ask them to demo each one live, using your real data. Not their cleaned-up sample data. Your actual messy numbers with your seasonality, your outliers, and your real-world complexity.
Ask them specifically what forecasting methodology their AI uses, whether it’s statistical, machine learning, or a hybrid. Ask what the minimum data requirements are, and what happens if you don’t have enough historical data. A vendor that’s confident in its AI will be transparent about its limitations. A vendor that’s overpromising will deflect.
Question 8: How does your platform prevent AI hallucinations in financial outputs?
This is one of the biggest concerns finance professionals have about AI, and for good reason. AI hallucination, when the model generates information that looks correct but is completely fabricated, is a real risk in financial applications.
The consequences are serious: direct financial losses, misinformed strategic decisions, and regulatory compliance issues. Ask the vendor whether you can trace any AI-produced number back to its source records, and whether clicking a forecasted number reveals exactly which data points and assumptions generated it. If the AI produces a revenue forecast and you can’t explain how it got there, you can’t defend it to your CFO, your board, or your auditor.
This is called end-to-end explainability, and it’s non-negotiable for financial applications. If you can’t trace a number back to source data, you can’t trust it. And if you can’t trust it, you shouldn’t use it.
If the vendor won’t explain how their AI arrives at a specific forecast, or if they say the AI learns on its own, that’s not transparency. That’s a black box.
Question 9: What automated variance analysis and root-cause capabilities do you have?
Variance analysis is one of the most time-consuming tasks in FP&A. A good AI platform shouldn’t just flag that revenue was $2 million lower than planned. It should help you understand why.
Ask the vendor to demo their variance analysis. Give them a real example from your business. Say, last month our gross margin dropped by 3%. Ask whether the platform can explain why. A good platform will break down the variance by driver, whether it was price, volume, mix, or cost. It should show you the breakdown by dimension: which product line, which region, which customer segment. And it should do this automatically, without requiring an analyst to manually pivot through the data.
If the tool flags the variance but requires manual investigation to find the root cause, you haven’t gained much efficiency. And if the vendor says our AI generates narrative commentary but can’t show you how that narrative is grounded in actual data, be sceptical. Generic commentary isn’t analysis.
WHAT: Total Cost of Ownership and Implementation (Questions 10–12)
Flashy AI doesn’t matter if the total cost breaks your budget or the implementation drags on for a year, and these last three questions are about protecting your wallet and your timeline.
Question 10: What is the full total cost of ownership over three to five years?
The licensing fee is just the beginning. Ask the vendor for a detailed breakdown. Not just the annual subscription, but everything else: implementation services, data migration, system integration, user training, customisation costs, and the cost of adding future modules or users.
Ask about the renewal terms. Ask whether there’s a cap on annual price increases, or whether they can raise your rate by 20% next year with no recourse. Ask specifically whether adding a new entity next year, or a new data source, or doubling your data volume, would change your pricing, and if so, how.
Some vendors have pricing models where scaling triggers a complete renegotiation. Others have transparent tier structures. You want the latter.
Question 11: What is your average time-to-value, and can you prove it?
Every vendor will give you an optimistic timeline. They’ll say three months to go live, or six months tops. But the reality is often different. Ask for data. Specifically, ask for their median time-to-value for implementations over the last twelve months. Ask what percentage of customers launched on time, and what percentage were delayed by more than four weeks.
A vendor with confidence in their delivery process will share this data. A vendor that’s hiding something will say it depends on the customer, or your situation is unique. Also ask about the implementation model. Ask whether it’s done in-house by the vendor’s team, or outsourced to system integrators. Partner-led implementations tend to cost more and take longer.
And here’s the critical question. Ask who is accountable if the implementation runs over time or over budget. If the vendor says the implementation partner handles that, you have no direct accountability. The vendor still gets paid. The partner still gets paid. You’re the only one who loses.
Question 12: Can you give me three customer references with similar complexity to ours?
This is the ultimate reality check. The vendor can show you anything in a demo. But customer references tell you what it’s actually like to work with them.
Ask for references matching your profile. Similar company size. Similar industry. Similar complexity, multiple entities, multiple currencies, multiple data sources. And don’t just ask the reference whether they’re happy. Ask specific questions.
Ask what the actual implementation timeline was versus what was quoted. Ask what surprised them, positive or negative, about the platform. Ask how responsive the support team is when something breaks. Ask whether they’d make the same decision again. If the vendor can’t provide three references that match your profile, that’s a risk signal. It means they don’t have experience with organizations like yours.
Shortlisted a few vendors? Put them head to head with our side-by-side comparison tool.
Run Your Demo Like a Due Diligence Session
Now you have the questions. But how you ask them matters just as much.
Send the vendor a list of your top questions before the demo. A good vendor will welcome this. They will prepare. A vendor that pushes back or says let us just show you the platform first, that tells you something.
During the demo, take control of the agenda. Don’t let them walk you through their standard presentation. Start with your questions. Say I’d like to start with Q&A, and then see the specific features that address our needs.
Ask them to use your data, not their sanitized sample data. Your actual charts of accounts, your actual cost centers, and your actual business scenarios, and if they can’t do that in a demo, they may not be able to do it in production.
And here’s the most important thing. Take notes during the demo. Not just on features, but on behaviours. Note whether they answer directly or deflect, and whether they volunteer limitations or make you pull every answer out of them. The way a vendor acts during the demo is how they’ll act during implementation. If they’re evasive now, they’ll be evasive later.
WHAT IF: What Happens If You Skip These Questions
If you skip this approach and just sit through the standard demo, here’s the picture six months later. The integration breaks every time the ERP updates. The AI features don’t work on your data volume. The reporting module can’t handle your multi-entity structure. And the vendor is asking for a change order to fix things that you thought were included.
Your CFO is asking why you chose this tool. Your team is asking when they can go back to Excel. And you’re stuck in a multi-year contract with a tool that doesn’t work, and you’re the one who has to explain the budget overrun in the next leadership meeting. Not the vendor. Not the implementation partner. You.
This isn’t hypothetical. This happens all the time. Gartner’s sixty percent failure rate isn’t about bad software. It’s about bad evaluations. The cost of buying the wrong FP&A software isn’t just the contract value. It’s the lost productivity, the frustrated team, the delayed insights, the wrong decisions based on unreliable data, and the opportunity cost of not having the right tool in place.
It’s much easier to say no before you sign than to get out of a bad contract, and these twelve questions give you the confidence to say no when you need to.
Closing
Save this list. Take it into your next demo. I promise you will make a better decision. I’ve also put together a free downloadable checklist of all twelve questions, linked at the end of this article.
Have you ever been burned by a software demo? Let me know in the comments. I’d love to hear your story.