‘Our Business Case Was Wrong by 40%’ — An AP Leader's Postmortem on Automation

AP automation rarely fails outright — it underdelivers in ways few buyers anticipate. In the discussion that follows, Steve Iannarone, a former P2P global process owner at a large CPG organization, speaks candidly about what the business case got wrong, what implementation truly cost, and how much longer the promised savings took to arrive. The answers form a practical reality check for anyone evaluating an automation investment. 

IOFM: What did you believe about AP automation before you bought it that you now know was wrong? 

Iannarone: Three assumptions were proved wrong. First, I expected users to respond to workflow notifications because they made exception resolution easier. In reality, notifications were often ignored, filtered, or viewed as email clutter. Second, I overestimated invoice capture accuracy and how quickly the technology would “learn” vendor behavior. That maturation can take months or longer. Third, I assumed ERP integration would be relatively seamless. Instead, monitoring interfaces, resolving failures, and establishing ownership sometimes created entirely new responsibilities. Automation works, but it requires disciplined adoption, governance, and ongoing support. 

IOFM: Where did the ROI you put in the business case actually come from — and which line item(s) didn’t materialize? 

Iannarone: The biggest lesson was that ROI projections were often built on assumptions rather than measured baselines. Cost per invoice, touches per invoice, exception rates, and processing effort had not always been tracked before developing the business case. That forced us to estimate benefits that should have been supported by historical data. As a result, projected reductions in cost per invoice and manual touches did not materialize as quickly or consistently as expected. I now believe establishing accurate baseline metrics before selecting a solution is essential. You cannot credibly demonstrate automation ROI if you cannot accurately measure where you started. 

IOFM: What did the true cost look like once you added integration work, internal headcount, and change management to the license fee? 

Iannarone: In my experience, the true cost ultimately ran approximately 20–40% above the initial business case. ERP integration sometimes required dedicated IT support that had not been budgeted. Change management was another underestimated expense; in one implementation, we trained Procurement and Receiving location by location to achieve the behaviors necessary for successful matching. Headcount savings also took considerably longer than projected. A reduction expected within six months was not fully realized until roughly two years after implementation. The lesson is that license fees are only one component of total cost technology support, process change, training, adoption, and stabilization must all be included. 

IOFM: What broke that no one warned you about? 

Iannarone: Almost anything can break, particularly when requirements and data quality are underestimated. Accurate, detailed business requirements significantly reduce that risk, but implementation timelines often pressure teams to move too quickly through requirements gathering. The consequences then surface during testing, implementation, or after go-live. I recommend creating a comprehensive readiness checklist covering approval routing, coding, tax logic, matching rules, interfaces, exception handling, and, especially, vendor master data quality. Even seemingly minor configuration differences can create major problems. Something as simple as an incompatible date format between the automation platform and ERP can disrupt an otherwise well-designed interface. 

IOFM: How honest were the reference calls you took, and what question do you now wish you’d asked them? 

Iannarone: Vendor-provided reference calls were useful, but they often presented a polished version of the implementation experience. What was missing was a candid discussion about how long implementation, vendor onboarding, stabilization, and organizational adoption actually took and how much change management was required. Today, I would ask much more specific questions, such as: What did your original business case underestimate? What additional resources were required after signing the contract? How long before you achieved the promised ROI? What failed during implementation or after go-live? And most importantly: If you were implementing the solution again, what would you do differently? That question often produces the most valuable insight.

Access IOFM's Automation Readiness Checklist