Trending Articles
Trending Topics
Trending
This article is part of our Opinions section, where we invite industry professionals to share their views on the most pressing technology questions of our time.
AI was supposed to be a revolution for service providers. Expectations for faster operations and lower costs were sky high, but the reality has brought many back down to earth.
PwC’s 29th Global CEO Survey found that 56% of CEOs have seen no revenue growth or cost reduction from AI, and only one in eight report both. That backs up what we’ve seen across managed service providers. There’s been no shortage of investment in AI, but the return just hasn’t materialised.
The knee-jerk reaction is to point the finger at the technology. In most cases, that misses the point entirely. The problem is that most are trying to automate operations they don’t fully understand. AI gets bolted onto chaotic operations and only succeeds in scaling the inefficiency.
If you want AI ROI, you need to get the foundations right first. It comes down to three things.
If you don’t have a clear picture of how work moves through your business, you can’t answer basic operational questions. Do you know how much work is in progress? Who owns it? Where does it get stuck?
Say a client submits a last-minute change before a pay run. The request lands in a shared inbox, gets forwarded to one team, then sits waiting on another for approval. Everyone assumes it’s being handled. It isn’t. The problem only gets noticed when the pay run goes out unchanged, and the client escalates. You talk about how to stop it from happening again, but the same pattern always repeats because no one has identified the root cause. This is what happens when work is scattered across inboxes, spreadsheets, and disconnected systems.
In that environment, teams spend most of their time dealing with exceptions and chasing updates. Prioritisation becomes guesswork because without a clear view of workload and capacity, everything feels urgent. McKinsey estimates that employees spend up to 28% of their working week on email. In service businesses, much of that time is spent triaging requests and following up on work that’s already been done.
AI won’t fix that on its own. Before you automate anything, you need a transparent, reliable view of your operations.
Once you have visibility, the next step is mapping the threads that hold the work together. This is where many organisations run into problems. There is a documented version of a process, but that rarely reflects what’s happening on the ground.
Take client onboarding. On paper, it’s a clean, linear sequence: receive documents, run checks, approve, activate the account. In practice, documents arrive incomplete or with errors. Someone chases missing information over email, while compliance flags an issue that doesn’t get formally logged anywhere. The process gets held up, and teams are improvising to move things forward. That neat four-step process becomes a series of manual interventions that vary in sequence and duration every time.
You need data on that kind of variation to build a sustainable AI business case. Without it, most AI projects struggle to scale. An initial pilot may show promise because the scope is controlled and edge cases are limited. But when the same approach is applied across the full operation, inconsistencies multiply. The volume of exceptions means teams are still manually intervening almost as much as before.
To avoid this, you need to map who is doing what, how long it takes, and what happens when it doesn’t go to plan. That level of detail lets you identify which processes are stable enough to automate and which need to be simplified or redesigned first.
Even with visibility and a solid understanding of your processes, you’ll still struggle to get results if your objectives for AI are too vague. Goals like “improve operational efficiency” or “raise customer service standards” are fine in general, but they aren’t specific enough to guide implementation.
That lack of clarity is a big reason why many AI investments fall flat. A recent MIT study found that 95% of businesses are getting zero ROI from AI. If you want to create real value, you need clear, measurable targets. That could mean reducing onboarding time for new clients, shortening invoice processing cycles, or fully automating the triage of inbound emails. The more defined your objective, the easier it is to build a solution and measure whether it’s working.
It also becomes clear that not everything should be automated. Some processes need to be simplified first, some standardised, others removed altogether. This is where process orchestration helps. It gives you the structure to coordinate work across people and systems, so you can apply AI in a controlled and targeted way. Instead of automating isolated tasks, you’re improving how work flows across the operation. That’s when the real gains show up.
The gap between AI expectations and reality has little to do with the technology itself. The businesses getting value from AI have absolute clarity over how their operations actually work. They build from that foundation instead of automating first and fixing later.
The truth is, you can invest in the most advanced tools on the market, and still see no ROI if the foundations aren’t there. The businesses that sort this out will pull ahead. The ones that don’t will keep wondering why their AI spend isn’t paying off.
