Business Process Improvement Consulting
Most process work fails because it documents the process someone describes rather than the one people actually run.
Book a callThere are two versions of every process in a growing company: the one people describe in a meeting, and the one they run on a Tuesday when something has gone wrong. Business process improvement that only captures the first produces documentation nobody uses.
The work itself is not complicated. Watch how a thing is actually done, find the steps that exist because of a decision nobody remembers making, take them out, and put a number on what remains so you can tell when it drifts. It is unglamorous, and between $5M and $20M in revenue it is usually worth more than any system you could buy instead.
How the work runs
Five steps. The order matters more than the steps do.
- Observe before mapping. We watch the work happen — sit with the people doing it, follow real records through the system. The gap between described and actual process is where the improvement usually is.
- Map the current state honestly. Process mapping that omits the workarounds is fiction. A workaround is information: it marks a place where the official process does not fit reality, and it is often the best idea anyone has had about the job.
- Remove before optimising. The cheapest step is the one deleted. Most processes at this size have accumulated approvals, checks and confirmations that made sense once and were never revisited.
- Then redesign, then document. Documentation last, so you are not writing up a process that is about to change.
- Then instrument. Cycle time, error rate, exception frequency — whatever will tell you it has drifted back.
Where this pays
Process problems cluster in the same handful of places in almost every company we see at this size. These are the findings, not hypotheticals.
| Area | Typical finding |
|---|---|
| Quote to cash | Quote, contract and invoice disagree; manual reconciliation absorbs finance time |
| Customer onboarding | Every new customer handled differently; delivery cannot forecast capacity |
| Sales handoff | No defined handoff, so context is lost between sold and delivered |
| Approvals | Thresholds that made sense at $2M and now block routine work |
| Data entry | The same record retyped into three systems because none of them is authoritative |
| Exception handling | Everything unusual routed to one person, who has quietly become the process |
| Reporting | Numbers assembled by hand each month, so they arrive too late to act on |
Two of these have specific homes. Where approvals and pricing exceptions are the bottleneck, the fix is a deal desk rather than a general process review. Where the finding is that the system nobody trusts is the CRM, the process work and the CRM work have to happen together — configuring a system around an undefined process just makes the confusion faster.
Where the problems concentrate in revenue systems specifically, this usually becomes a revenue operations engagement instead — the same work, applied to a narrower and higher-value surface.
The gain is rarely one dramatic saving. It is a few hours a week returned to people who are expensive, a class of error that stops occurring, and a business that can take more volume without adding headcount in proportion. That is what operational efficiency improvement looks like at this size. Anyone describing it as a transformation programme is describing a different company.
Methodology, honestly
The question we get asked first is which business process improvement methodology we use. The honest answer is that the named frameworks were built for environments with far more repetition, and far more data, than a $5M–$20M company has.
- Lean. The useful half travels well. Find the waiting, the rework and the handoffs; delete what does not serve the customer. We use this constantly, without calling it a lean programme.
- Six sigma. Built for statistical control over high-volume repetition. A process that runs a few hundred times a year does not generate enough events to support it, and the training overhead lands on people who have day jobs.
- Formal notation. BPMN and its relatives are good at communicating between analysts and expensive to keep current. A one-page diagram someone will actually update beats a formal model that is wrong within a quarter.
- Tooling. Process improvement software is almost never the constraint at this size. A whiteboard, a spreadsheet and the CRM you already pay for will carry the whole engagement.
What you get
One page per process, showing how it actually runs — workarounds, rework loops and offline steps included.
What to change, in what order, with the value and the effort attached to each item so you can stop partway and still be ahead.
Running with real work going through it, not a proposal. We stay until it survives a normal week without us.
Written after the change, short enough to be read, owned by a named person. The version a new hire can follow on day three.
Two or three numbers per process, an owner for each, and a standing review that catches drift before a customer does.
Engagements normally start with a diagnostic across the processes causing the most friction, then take on two or three of them properly rather than touching all of them lightly. Where the answer turns out to be that somebody needs to own operations continuously rather than fix a defined set of processes, that is a fractional COO engagement, and we will say so before you have paid for a review you did not need.
Who this fits
Good fit
- ✓Revenue between $5M and $20M, growing faster than the way the business is run
- ✓The same problems recur and get solved individually every time
- ✓One or two people are load-bearing, and work stops when they are on holiday
- ✓A system was bought to fix a process, and the process is still broken
- ✓A sale, a raise or an audit is coming and operations will be looked at closely
Poor fit
- ×You want the current process documented and left as it is
- ×Manufacturing floor or clinical process work, which needs a specialist we are not
- ×The real problem is one specific person, and process is being asked to solve it
Start where it costs most
A short diagnostic across your highest-friction processes, with a prioritised view of what to fix, what it is worth, and what to leave alone.
Book a callCommon questions
- What is business process improvement?
- Changing how work actually gets done so it takes less time, produces fewer errors, or stops depending on one person being available. In practice it is three things: finding out how a process really runs rather than how it is described, removing the steps that no longer earn their place, and attaching a measure so you notice when it drifts back.
- Do you use lean, six sigma or another methodology?
- We borrow from lean where it helps — removing waste and shortening handoffs is nearly always the right instinct. Six sigma needs a volume of repetitions most companies this size do not have; control charts on a process that fires 200 times a year tell you very little. Most problems at $5M–$20M do not need a framework. They need someone to watch the work happen and remove the steps that stopped making sense two years ago.
- How do you measure whether operational efficiency actually improved?
- Baseline before changing anything, because a process with no before is not improved, it is just different. Then track the measure that matches the failure mode: cycle time where the complaint is slowness, rework and error rate where the complaint is quality, exception frequency where the complaint is that nothing is standard. Two or three numbers reviewed monthly beat a dashboard nobody opens.
- Will this produce documentation nobody reads?
- It does if you document first. We fix the process before writing it down, so what gets documented is worth following — and it is short. Most process documentation fails on length before it fails on accuracy.
- Should we hire a process improvement specialist instead?
- If you have enough process work to keep someone busy permanently, hire. Below roughly $20M in revenue most companies do not — the work arrives in bursts, and a full-time specialist with too little to do starts generating process for its own sake. Where the need is ongoing operating ownership rather than process work specifically, the role you want is closer to a fractional COO.
- How long does an engagement take?
- A diagnostic across your highest-friction processes takes two to three weeks. Redesigning and handing over one significant process is typically four to eight, depending on how many systems it crosses and how many people have to change what they do. Anything promising a full operational overhaul in thirty days is selling documentation.
Related
- Fractional COOYou have outgrown the way you run the business, but a full-time COO is a $300k commitment you cannot justify yet. A fractional operator closes that gap.
- CRM ConsultingA CRM is only as good as the process it encodes. Most of what looks like a CRM problem is an undefined process problem.
- What a Fractional COO CostsNobody publishes real numbers for this, so here they are. What fractional operators charge, how engagements are structured, and what moves the price.