The numbers on automation are genuinely good. UK research puts the average return at £6.30 for every £1 spent on workflow automation over three years. Document-heavy processes pay back in three to six months. Across the mid-market businesses we have worked with, we have found an average of £280,000 in recoverable operational value sitting in plain sight.
So why do so many automation projects stall, overrun, or quietly get switched off six months later?
In our experience it is almost never the technology. UiPath, Automation Anywhere and Power Automate all work. The bots do what they are told. The project fails earlier than that, in a step most teams skip: checking whether the process was ready to automate in the first place.
Here is the trap. A process is slow and painful, so it feels like an obvious candidate for automation. The instinct is to point a bot at it and move on. But if the process is broken, undocumented, or full of exceptions that live in one person's head, automation does not fix any of that. It just produces the wrong outcome faster, at scale, and now nobody remembers how it used to work.
We have walked into engagements where a previous automation attempt had encoded a workaround someone invented years earlier. The bot ran perfectly. It also did the wrong thing several thousand times before anyone noticed. The lesson is not "automate less." It is "understand the process before you automate it."
The mistake is assuming the most painful process is the most ready one. They are often the opposite. The processes that automate well share three unglamorous traits.
They are stable. The steps are the same most of the time, and the rules do not change every quarter.
The data is clean and accessible. Ideally twelve months of it, so the automation has something reliable to act on.
They are documented. Well enough that a new joiner could follow the process within a week. If a process only exists in someone's memory, you are not ready to automate it, you are ready to map it.
Get those three right and the economics look like the headline figures. Skip them and you join the long list of projects that delivered a working bot and no value.
When we map a mid-market operation, the highest-value opportunities are rarely the ones leadership names first. Invoice and document processing is a reliable winner, with cost reductions of 75 to 90 percent and payback inside half a year. Approvals, reconciliations, and the small daily handoffs between systems add up quietly. None of it is glamorous. All of it pays.
The point is that the opportunity has to be found and ranked, not guessed. The difference between a 200 percent first-year return and a stalled pilot is usually which process you picked, not which tool you bought.
If you are weighing up automation this year, resist the urge to start with a platform demo. Start with two questions. Which of our processes are genuinely stable and documented enough to automate now? And where is the recoverable value actually sitting?
That is the work we do before a single bot is built, because it is the work that decides whether the project pays for itself. The firms getting six-to-one returns are not running better software than everyone else. They are automating the right processes in the right order.
Not sure which of your processes are ready? Answer 8 questions and get a personalised Intelligent Automation Readiness Score across four areas: where your highest-value opportunities sit, which processes are genuinely ready right now, and what your first move is worth. We have found an average of £280,000 in recoverable operational value across the mid-market businesses we have assessed.
Tell us what you are rolling out and where adoption, automation or AI is sticking. We will come back with a clear plan for the first steps, what success looks like, and what it costs. No fifty-slide pitch.
Book a call