Automate the boring 20% first

When business owners get interested in automation, they almost always arrive with the same target in mind: the impressive thing. The customer-facing portal. The AI-assisted estimate. The end-to-end system that will change how the whole company works. And when those projects stall — as they often do — automation itself gets the blame. Our experience says otherwise. Automation fails not because the technology is weak but because the sequencing is wrong. The reliable path starts with the boring 20% of work that nobody finds interesting and everybody wishes would do itself.

What the boring 20% looks like

In every business we assess, the same family of tasks shows up: assembling documents from the same fields, copying data between two systems, chasing approvals that follow a known path, compiling the weekly report, sending the reminder that someone always forgets to send. These tasks share four properties that make them ideal first candidates. They are high-volume, so the savings multiply. They are rule-based, so a machine can actually do them. They are low-risk, because a mistake is caught downstream rather than sent to a customer. And they are unloved — meaning automation here costs no one their favourite part of the job, which matters more for adoption than any technical factor.

Why the boring first, in earnest

Three reasons, learned from watching the other order fail. First, certainty of payoff: a document generator that saves twenty minutes per quote will save exactly that, starting this week — no adoption curve, no behavioural change, no dependency on customers learning anything. Second, credibility: the fastest way to poison a broader automation programme is a flagship project that over-promises; the fastest way to build support is three unglamorous wins that everyone can count. Third, learning: the boring builds teach your organisation how automation behaves — its failure modes, its maintenance needs — at low cost, before anything critical depends on the lesson.

“The measure of a good first automation is not how impressive it looks in a demonstration. It is whether, six months later, anyone can imagine going back to the manual way.”

Where we draw the line

Not everything boring should be automated, and our opinion here sometimes disappoints: if a task happens twice a month and takes ten minutes, automating it is a hobby, not a project. If the process is unstable — changing shape every time it runs — automation just makes the instability faster. If the task involves genuine judgement — which customer to call back first, whether this job needs the senior crew — the machine has no business deciding, though it can still prepare the information. And if the person doing the task is the only one who knows how it works, document it first; automating an undocumented process is how businesses become hostage to a script nobody understands.

The practical sequence we use

1. Inventory the repetitive. List the recurring manual tasks with their frequency and minutes — the numbers are usually surprising.
2. Score for volume × stability × risk. High volume, stable steps, contained failure modes: that is the shortlist.
3. Build one, run it in parallel. Nothing goes live until the automation has matched the manual process for a full cycle.
4. Bank the win, then repeat. Each success funds credibility and reveals the next candidate. Three boring wins will do more for your operations than one ambitious flagship ever has.

If your team could name its most tedious task right now — and they can, immediately — you are looking at a candidate. We scope these as small, fixed-price builds precisely because the first one should be a decision you never have to defend.

Three first builds that paid for themselves

To make "boring 20%" concrete, here are three first automations from real engagements — none dramatic, all still running. The document assembler: a services business whose quotes were assembled by hand from a price list; a generator now builds them from the job record in seconds, ending version errors and cutting quote turnaround by two days. The invoice chaser: a distributor whose overdue accounts were chased manually, unevenly, depending on the week; a scheduled routine now sends graduated reminders automatically, and the owner learned that cash flow improves more from consistency than from firmness. The onboarding checklist: a firm whose new-client setup lived in one partner's head; a structured checklist with automatic reminders now guarantees nothing is missed, and the knowledge survives the partner's holidays. None of these took more than a few weeks to build. All three are boring. All three are exactly why the second automation is always easier to justify than the first.