Ask an owner why orders take eleven days to get out the door and you'll usually hear a name. "Sales is slow on quotes." "Accounting is backed up." "We need another person in ops." Then you watch the work move, and something uncomfortable shows up: nobody is slow. The quote took nine minutes to write. It sat for three days waiting for someone to notice it existed.
That gap — between how long work takes and how long it takes to get done — is where most mid-market capacity hides. You don't need more hands. You need fewer waits.
Touch Time vs. Wait Time
Pick one process that matters: quote to order, order to cash, lead to first call, ticket to resolution. Now split every step into two numbers. Touch time is how long someone is actually working on it. Wait time is how long it sits between steps.
In almost every process we look at, wait time dwarfs touch time, and it's not close. The work itself is minutes. The elapsed time is days. That ratio is the single most useful diagnostic in operations, because it tells you whether to hire, train, or re-plumb.
If touch time is high, you have a skill, tooling, or complexity problem. If wait time is high — and it usually is — hiring makes the queue longer, not shorter. You've just added another desk for work to sit on.
Where Work Actually Sits
Queues are boring and invisible, which is why they survive for years. They rarely show up in a process map because nobody draws a box called "sat in an inbox." Common places to look:
- An approval that only one person can give, who is also the busiest person in the company
- A handoff that happens by email, where "sent" and "seen" are different events
- A step blocked on information that exists, but in a different system
- Batch habits — invoices go out Fridays, POs get entered at month-end — that add days for no operational reason
- Rework loops, where something comes back because the first pass was missing a field
The cheapest way to find these is to take ten recent completed items and build a timeline for each from timestamps you already have: email dates, CRM activity, ERP record creation, ticket logs. You're not looking for an average. You're looking for the step where the clock stops moving.
Three Kinds of Bottleneck, Three Different Fixes
Approval bottlenecks are threshold problems. Someone approves everything because no one ever defined what doesn't need approving. The fix is a rule, not a reminder: under this dollar amount, with this margin, with this customer status, it goes automatically. Most approvals are theater for the 90% of cases that have never once been rejected.
Information bottlenecks are integration problems. The step is blocked because the person needs a credit limit from the ERP, a contract from the drive, and a last-touch note from the CRM — and getting all three takes longer than the task. The fix is bringing the data to the work instead of sending the person to find it. This is exactly what a unified hub is for: one screen where the context already lives next to the decision.
Single-point-person bottlenecks are documentation problems wearing a staffing costume. One person knows how pricing exceptions work, how the old customer records are coded, which vendor needs a special PO format. Everyone waits on them. The fix isn't cloning them. It's writing down the decision rules — the thresholds, the exceptions, who gets consulted when — so the work can route around them.
Fix the Queue Before You Hire
A new hire costs real money every month, forever, and adds coordination overhead. Removing a queue is usually a one-time project. Before you post a role, ask: if elapsed time on this process dropped by half, would we still need the person? Often the answer is no, and the honest version of the staffing request was "this process is painful and I want it to stop."
This is also where AI earns its keep, and where it doesn't. AI is good at the things that make work wait: drafting the first version, assembling context from scattered records, flagging the item that's been sitting too long, summarizing a thread so the approver can decide in thirty seconds instead of ten minutes. It is not good at owning the judgment call. Use it to shorten the wait, not to replace the decider.
Make the Clock Visible
The reason queues come back is that nobody measures them. Touch time feels like productivity, so that's what gets tracked. Pick two or three processes, measure elapsed time end to end, and put it on the same weekly report as revenue. The moment that number is visible, people start removing waits on their own.
This is the work we do with operators every week — mapping where the clock stops, then closing the gap with integrations, clear decision rules, and automation that fits how the business actually runs. If your team feels understaffed but nobody looks idle, the problem is probably a queue. We'd be glad to help you find it. Reach out and let's map one process together.