Finance Teams Adopted AI, but 80% Still Plan in Spreadsheets

By one measure, finance has finished adopting AI. Two in three finance teams now use at least one AI tool, and the share has more than doubled in two years. By another measure, very little has changed. Four in five teams still build their plans and forecasts mainly in spreadsheets, and more than half still track cash that way. Both findings come from the same survey, and together they describe a function that has bolted AI onto its edges while its planning core runs on the same files it used a decade ago. That gap is where the next round of AI spending will either pay off or stall.

Adoption Went Vertical. The Workflow Did Not

The numbers come from CFO Connect's annual poll of 215 CFOs and senior finance staff in France, the UK, Germany and the US, carried out in June and July and reported by The Next Web on September 29. The survey, backed by Spendesk, Oracle NetSuite and Remote, found 67% of finance teams using one or more AI tools, up from 56% a year earlier and 31% in 2024.

Look at what those tools are doing, though, and the picture narrows. Financial analysis was the most common use at 20% of respondents. Reporting, modeling and forecasting came in at 12% each, reconciliations at 9%, and data queries and workflow automation at 8% each. Much of that is individual productivity: an analyst asking a chatbot to summarize a variance, draft commentary or restructure a table. It is useful work, but it does not change where the numbers live.

And the numbers still live in spreadsheets. The survey found 80% of teams carry out planning and forecasting primarily in spreadsheets, and 55% track cash in them. Meanwhile, 87% run dedicated payroll or HR systems. Finance has bought purpose-built software for the processes other functions own, while its own forward-looking work, the part leadership leans on hardest, remains the least systematized.

The Quick Wins Are Not Where the Spreadsheets Are

Gartner's latest research explains why this matters for the return on AI budgets. In a survey of 160 senior finance leaders fielded from January to April, covered by CPA Practice Advisor, the firm found that data extraction, accounts payable and receivable automation and report creation generally deliver returns within nine to 10 months. Data management, insight generation and forecasting need longer development before value shows.

Put the two surveys side by side and the problem is plain. The use cases that pay back fastest are transactional, and they sit in systems that are already structured. The use cases that pay back slowest are analytical, and they sit in the spreadsheets most teams have not replaced. AI cannot reliably forecast from a planning model that exists as a chain of linked workbooks with manual overrides, because there is no consistent data layer underneath it to learn from.

"AI adoption has reached a point where CFOs must adopt more deliberate portfolio management," said Marco Steecker, senior director analyst in the Gartner Finance practice. Gartner also found that low AI literacy has replaced talent acquisition as the primary barrier to AI success in finance. That reframes the spreadsheet question. Teams are not stuck because the tools are missing. They are stuck because the skills and data foundations needed to move planning onto something AI can work with are thin.

The Big Companies Feel It Too

This is not only a mid-market story. Deloitte's Q3 CFO Signals survey, which polled 200 North American CFOs at companies with at least $1 billion in revenue between August 24 and September 8, found 50% ranking technology deployment, including generative AI, as their top internal challenge. Cybersecurity, also at 50%, led the external concerns. Confidence is up, with Deloitte's score rising to 6.1 from 5.9, and CFOs expect capital spending to grow 4.3% over the next year. The appetite to invest is there. The difficulty is turning investment into changed workflows.

Buyers also appear to be tiring of point solutions. CFO Connect found that more than 90% of teams run six or fewer tools, two-thirds plan minimal additions next year, and nearly a quarter intend to move to a single platform. Spendesk CFO Pauline Babel told The Next Web, in substance, that finance teams want one joined-up place to work rather than more software. That points toward consolidation, and toward the planning layer as the next system worth replacing.

The Playbook: Get Planning Out of the Spreadsheet

• Audit where the numbers actually live: Map every forecast, budget and cash view to its source. If the answer is a workbook on someone's drive, that is the first constraint on any AI forecasting ambition.
• Bank the nine-month wins first: Fund data extraction, AP and AR automation and report creation now, and use the savings to pay for the slower planning work.
• Build the data layer before the model: Move plans and forecasts onto a structured platform with one version of drivers and actuals. Forecasting AI needs that foundation more than it needs a better prompt.
• Treat cash tracking as a priority system: With 55% of teams tracking cash in spreadsheets, a daily cash position fed directly from bank and AR data is one of the cheapest visibility gains available.
• Train for literacy, not just access: Gartner ranks low AI literacy as the top barrier. Budget for teaching analysts to test, question and validate AI output, not just for licenses.
• Consolidate with a destination in mind: If the tool stack is shrinking anyway, decide which platform planning will sit on before the next renewal cycle decides it for you.