The bottleneck is almost never where you think it is

Most conversations about automation start in the same place. Someone describes the part of their business that hurts most, and it is almost always the complicated end.

The reporting that takes two days. The invoicing reconciliation. The spreadsheet four people maintain and nobody fully trusts.

Those are real problems. But they are rarely where the trouble starts. They are where it collects.

The unglamorous fix that unlocked everything else

We recently worked with a growing business that had exactly this shape of problem. The fix that made the biggest difference was not the clever one. It was a website enquiry form.

Their enquiries landed in a shared mailbox that several people watched. It had worked for years, until a handful of enquiries a week became several a day.

Nobody owned an enquiry, so most were answered quickly and some quietly were not. Occasionally two people replied to the same one. And every enquiry worth pursuing was retyped by hand into their work management system, then partly retyped again to raise a proposal.

What we put in its place looks modest on the page. The form now writes straight into their system as a structured record, that record is assigned automatically to the manager for the region, and that person is notified by name. When they decide to pursue it, the client and the proposal are created from the data already captured. Nothing is retyped.

The build was the quick part. Working out that this was the thing to build took longer, and it is the part that actually mattered.

Nobody in the business had raised the enquiry form as a problem. It only surfaced because we had traced the entire path a job takes before we touched anything.

It was the highest value change we made all year.

Value comes from position, not cleverness

Everything downstream depended on that form. The project naming convention we built later assumes the client name is correct. The folder automation assumes the project name is correct. The accounting codes assume the folders. The monthly forecast assumes all of it.

Every one of those is more technically interesting than a web form, and every one was sitting on top of data a tired person had typed out of an email at four o'clock on a Friday.

Had we automated the reporting first, we would have built something impressive that produced confidently wrong numbers, faster and more consistently than the manual version it replaced.

An automation inherits the quality of everything upstream of it. So the value of a fix depends far more on where it sits in a workflow than on how much work it took to build.

The skill is not in the building. It is in knowing which change to make, and a small change at the right point will outperform an ambitious one in the wrong place every time.

Which is why you map the whole thing first

None of this is visible when you look at problems one at a time. Examine the monthly report in isolation and the answer looks like a better report. Follow the whole path a job takes, from first enquiry to final invoice, and you notice the same information being re-entered at four separate points, each time with a fresh chance to differ from the last.

So before automating anything, we map the workflow end to end. Not the org chart and not the software, but the actual path of the work and the information travelling with it. It is a slow, slightly tedious exercise involving a lot of questions that sound obvious, and it is where nearly all of the useful findings come from.

Four questions surface most of what matters:

• Where does information enter the business, and does it arrive structured or as prose someone has to interpret?

• Where is the same thing typed more than once?

• Where does an item sit without a named owner?

• Where does someone apply judgement that has never been written down?

  
  

Processes survive on quiet human corrections. Someone notices the wrong client was selected and fixes it without mentioning it, because it took them four seconds.

Automate around that person and the corrections stop happening, and the errors they were absorbing surface all at once, six weeks later, in a report to the board.

The simple answer is often the hard sell

The best automation work we do tends to look underwhelming written down. A form. A naming rule. One field that gets populated automatically instead of by hand.

The payoff comes from choosing the right one, and you only find the right one by understanding the whole process before touching any part of it.

If you are weighing up where to start, resist the urge to begin where it hurts most.

Follow the pain upstream and find where it originates. Usually it started somewhere quiet, a long way from where you noticed it, and the fix is smaller than you expected.