Is the mid-market ready for a wholly automated finance function? A new vendor survey indicates that most midsize businesses haven’t even automated their accounts payable processes entirely.
In an online survey of 225 mid-market finance and accounting leaders in September 2025, only 4% of respondents said they had fully automated AP “from invoice to payment with no manual touchpoints,” according to San Francisco-based AP automation firm Ottimate. The pool of respondents reported annual revenue between $20 million and $499 million.
The survey, which queried respondents across healthcare, retail and hospitality, goes on to say that 48% of respondents said they’ve seen “little to no cost savings from their AP automation tools.” Ottimate attributed that to the prevalence of “partial” automation, where some tasks are handled by technology tools and others are handled manually. Specifically, 89% of respondents said they use partial automation in the AP function, while 7% reported not using any automation at all.
“Partial automation may seem like a step in the right direction,” the company wrote in the report. “But it actually sets the stage for some of the key challenges finance teams face every day.”
Half of respondents, meanwhile, said they’re fielding more than 5,000 invoices each month. The survey also found that 38% of respondents take five or more days to process a single invoice.
“While that timeline may not seem particularly problematic, it quickly becomes unmanageable when multiplied across thousands of invoices,” Ottimate’s report stated.
This all comes on top of the increasing prevalence of fraud these days, some of it driven by new artificial intelligence tools. The survey found that four in 10 respondents had experienced either invoice fraud or overpayment in the last year.
And for midsize firms, manual reviews of invoices before payment remain the top choice of financial controls in place to stem fraud. To wit, 52% of respondents cited that as their financial control tool, while 50% of respondents said they require two or more approvers before releasing a payment.
“As Al evolves, fraud is getting more sophisticated and harder to detect,” Ottimate’s report said. “The manual processes and disconnected systems that define partial automation make organizations especially vulnerable.”
Among the reasons that mid-market businesses have been slow to adopt automation in AP?
Half of respondents cited cost, while a similar share cited “difficulty integrating with existing systems.” Forty percent cited an “unclear return on investment.” And indeed, the jury is still out on the ROI of some of the latest generative AI tools, with one recent report by Workday finding that almost half of the time saved using them is spent correcting faulty outputs.
Over the years ahead, finance chiefs doubtless will be taking a closer look at the promise and peril of AI tools across several functions. For some, finance-specific functions by Anthropic’s Claude model appear promising, though those, too, come with risks.
