Most teams automate the loudest step, not the slowest one. The phone rings constantly, so you automate the phone — and nothing downstream gets faster. The real constraint is usually a silent queue nobody complains about.
This is a triage problem, not a software problem. You need a way to rank bottlenecks by impact and risk before you build anything.
According to the two-axis framework, which quadrant should you automate first?
Select one answer.
Why automating the loudest step fails
A bottleneck is the single slowest step in a chain, not the noisiest one. When clinics automate a visible step ahead of an unchanged slow one, work just moves into the queue faster. The fix is to measure elapsed wait at each handoff rather than volume: booking to reminder, form sent to form returned, visit to charge posted, result received to patient told. Whichever gap is longest is the constraint.
That reframing matters because it changes what you build first. Bottlenecks are commonly caused by process, system, and software issues rather than staff effort, so the answer is usually a redesign, not a faster human.
The two-axis triage framework
Score every candidate process on two numbers.
- Axis 1 — Volume: how many person-hours per month the process consumes today. This translates directly into money and sets your savings ceiling.
- Axis 2 — Judgment: how much human judgment each case requires, scored 1 (fixed rules, always the same) to 5 (expert judgment, every case is different). This measures risk.
Cross the axes and you get four plays:
- Many hours + low judgment → automate now. The winning quadrant: invoice reconciliation, order registration, recurring reports, repetitive queries.
- Many hours + high judgment → assist, don't replace. Let the system prepare most of the work and keep a human on the decision.
- Low hours + low judgment → automate later or selectively.
- Low hours + high judgment → avoid or redesign before automating.
A three-factor variant adds error cost to volume and time per execution, which is useful when quality problems are the real pain.
Run a bottleneck triage in one week
- List candidates. Ask the team to name what wastes their time. Expect 40–60 items; that is normal.
- Measure, don't guess. Pull elapsed wait times at each handoff from your existing tools. If you cannot get the data quickly, that reporting gap is your first bottleneck.
- Score each item on hours per month and judgment level.
- Rank by quadrant, not by who complained loudest or what is easiest to wire up.
- Standardize before you automate. Document how the process runs today and define what decisions get made and by whom.
- Ship one boring win. Pick the top high-volume, low-judgment process and automate it end to end.
- Re-measure the same gap you started with. If it did not shrink, you automated the wrong step.
Common bottlenecks worth triaging first
The recurring culprits are manual approval chains, document version chaos, data re-entry across disconnected systems, email-based routing, and tool silos. Cumbersome approval processes are a classic case: a single person on vacation can stall the whole chain. Client intake and scheduling, data entry and document processing, reporting, and invoice follow-up are the categories that most often sit in the winning quadrant.
Signs you picked the wrong first project
- The process runs only occasionally.
- It touches too many departments at once.
- It needs expensive custom integration before it works.
- It is more impressive to describe than it is painful to run.
The right first automation is almost boring: it saves real hours immediately, breaks rarely, and makes someone's day measurably better.
How the Featured Expert Can Help
EF Operations, led by Emily Falk, provides strategic operations consulting that aligns people, systems, and technology. The firm focuses on bringing structure to growth while preserving what makes a business unique. You can reach Emily directly at emily@efoperations.com or visit efoperations.com to discuss how they work with growing businesses.
Think you have found your bottleneck? Test your understanding before you start scoring.