Why ROI Is the Wrong North Star for Workflow Automation

here's the thing: the CX Today piece, “Measuring the ROI of Workflow Automation,” asks the only question that consistently gets you invited back to the boardroom — where’s the money. Any leadership team that won’t talk ROI is basically leaving the door unlocked and hoping the budget fairy shows up. But treating “what’s the return?” as the only metric that matters is how a smart conversation about automation quietly becomes very dumb.

Let’s start with the part I think the ROI crowd gets right: hard numbers do change behavior. When leaders see ticket-handling costs drop, or handle times compress, people pay attention. Projects that once lived in “innovation labs” suddenly get real funding. That discipline is valuable. It also creates a kind of tunnel vision.

Yeah, no, most automation business cases collapse into labor-hours-saved math because it’s tidy. You know the ritual: “We’ll save X hours a week, which equals Y salaries, so the project pays back in Z months.” That looks great in a deck. It also misses what a lot of automations actually buy you.

Some automations are throughput machines — process more invoices, route more tickets, close more cases. Fine. But others are capability builders: they let you test new service models, stitch data across systems, or give frontline staff context they never had before. Those second-order effects show up as fewer escalations, smarter cross-sell moments, faster product iteration cycles — all real, all meaningful, and all annoyingly hard to cram into a single ROI cell.

Treating both kinds of automation with the same spreadsheet calculus is like valuing a bridge only by the scrap price of its steel. You ignore the town it connects to, the businesses that pop up around it, the traffic patterns that change because it exists. Platform bets — whether that’s UiPath, Microsoft Power Automate, or whatever your preferred acronym soup is — are closer to infrastructure than tools. Their payoff is messier, slower, and often indirect, which makes every quarter-obsessed CFO a little itchy.

Industry context warps the ROI lens too. In retail or customer service, workflow tweaks can visibly hit churn and loyalty metrics. You get quicker customer responses, more consistent experiences, and some of that does show up in revenue lines. Meanwhile, in healthcare or government, automation might be about compliance, safety, or more equitable access to services. Try capturing “fewer life-destroying paperwork errors” in a single ROI percentage and see how far you get.

So yes, ROI is a useful language, but it’s a dialect with regional accents.

Now to the part that tends to be missing from the glossy ROI diagrams: the costs nobody wants to put on the slide.

Data privacy is the obvious one. Shipping a workflow into production often means routing personally identifiable information through a daisy chain of connectors and third-party tools. That isn’t just a checkbox for legal; it’s a real operational risk. A misconfiguration or over-permissioned integration can turn into remediation work, regulatory pain, and reputation damage that will never be fully reflected in that “year one savings” bar chart.

Integration complexity is the other silent tax. Legacy ERPs and custom CRMs don’t greet new automation tools with a friendly handshake; they respond with weird exceptions, brittle APIs, and rate limits that cut your shiny workflow off at the knees. What was pitched as a quick pilot turns into a quarter of engineering triage, and suddenly the ROI that looked stellar on a pre-sales slide now looks suspiciously like a rounding error.

Vendor lock-in? Same story. When you commit to a proprietary orchestration layer, you’re not just buying functionality; you’re renting your future optionality. A few years ago, plenty of enterprises raced into single-cloud commitments because it made procurement easier and dashboards prettier. Then the bills showed up, and the word “egress” entered everyone’s nightmares. Automation platforms can follow the same arc if you don’t bake exit ramps into the design.

There’s a quieter cost, too: culture and institutional memory. Replace a human step with a bot and you don’t just free up time; you delete a daily touchpoint where someone notices patterns. The person who used to handle exceptions often knows all the edge cases the system was never designed for. When that knowledge atrophies, the long tail of weird failures gets more expensive — especially when something breaks in a way the original automation never anticipated.

Look, the counterargument is obvious: without strict ROI scrutiny, nothing gets funded and everyone hides behind “strategic value” until the money runs out. I get it. Hard dollars are a necessary discipline. Pilots should show quantifiable savings and some credible velocity metrics, or they’re just theater.

But strict doesn’t have to mean narrow.

Treat automation ROI as portfolio math, not a single bond. Run staged pilots with dual metrics. Track direct cost savings, absolutely — but pair them with leading indicators: time-to-resolution, escalation rates, customer satisfaction, the number of new experiments the platform makes feasible. You’re not just buying efficiency; you’re buying option value.

We’ve seen this movie before. When Amazon rolled out its internal tools for automating deployment and infrastructure, the immediate cost story was helpful but not the headline. The real payoff was that teams could ship faster, experiment more, and treat operations as code instead of ceremony. The spreadsheets couldn’t fully explain why that mattered; the compounding effect over years did.

I’ll be honest — there is no clean, universal formula here. Any vendor pitch that hands you one neat ROI number is selling confidence as much as software. The smarter buyers I talk to obsess less over the exact percentage and more over governance: clear data-handling rules, planned integration work, exit clauses that actually get tested, and metrics designed to surface unintended consequences as quickly as they surface savings.

Philip K. Dick loved pointing out how the tools we build quietly rewrite the worlds we live in; workflow automation is doing that in slow motion across back offices everywhere. CX Today is right to push on measuring returns — but the companies that actually win will be the ones that treat ROI as a starting point, not a finish line.