What Does a Fractional COO Actually Do? The First 90 Days, Week by Week

"Streamlines operations and builds scalable systems." That is what most descriptions of a fractional COO say, and it is true in the way a map of the world is true: accurate, and useless for finding your way home. Founders ask me what the role actually does because they are about to pay for it, and they want to know what will be different on a Tuesday in month two.

So here is the role as I have actually run it, in companies between ten and a hundred and fifty people, week by week.

What the role is, in one sentence

A fractional COO owns how the company turns decisions into results — the operating model — a few days a week, with real accountability for the numbers, and without the cost of a full-time executive. It is not a consultant who writes a report, and it is not an operations manager who runs the schedule. It is an executive seat, filled part-time.

The part-time part matters less than founders expect. Most operating problems in a company this size are not a shortage of hours; they are a shortage of decisions. Two focused days a week from someone who has made those decisions before usually beats five days from someone who has not.

Weeks 1–2: one map of how work actually moves

I do not start with an audit of every department. I start by following one unit of value through the company — one order, one client project, one lead that became a customer — and writing down every handoff, every tool, every place someone re-types something, every approval that waits for the founder.

Three things come out of this:

• A list of processes that depend on a person rather than a system. In every company I have done this, there are three to five that would stop the business if that person were on holiday.
• The real cycle time of the core process, which is almost always two to three times longer than leadership believes.
• The founder's hidden job: the decisions and approvals that route through them because nobody else has been given the authority. This is usually the biggest bottleneck and the one nobody has named.

Weeks 3–4: the numbers that get reviewed every week

Before fixing anything, I set up the weekly review: five to seven numbers, one owner each, one meeting that looks at them. Not a dashboard with forty metrics — a short list where a change in any number means somebody does something.

For a DTC brand that was cost per sale, contribution margin per order, fulfilment lead time, return rate and cash conversion. For a services company it was lead-to-close time, utilisation, delivery margin per project and the number of open client escalations. The list differs; the principle does not: if a number can move and nobody has to act, it does not belong on the list.

This is also where automation starts. At one company the weekly report was assembled by hand from four systems and took days. Once attribution, KPI analytics and logistics data were wired together, it took hours, and the meeting started being about decisions instead of about whether the numbers were right.

Weeks 5–8: the first fixes, chosen by cost

With the map and the numbers, the priority list writes itself. I pick the two or three processes where the gap between "how it should work" and "how it works" costs the most money or the most founder time, and fix those first. Typical examples:

• Ownership. Each process gets one named owner and a written definition of done. This sounds trivial and changes everything; most stalled work is stalled because two people each think the other owns it.
• The approval bottleneck. Decisions get thresholds: below this amount or risk, the owner decides; above it, the founder does. The founder's inbox shrinks by half within a month.
• Tooling and automation. The first automation is always the most boring one — moving data between the CRM and the fulfilment or delivery system, so nobody re-types it. Boring automations pay back in weeks. Ambitious ones pay back in quarters, if at all.
• What to stop. Every company has reports nobody reads and meetings nobody needs. Cutting them is the cheapest capacity you will ever add.

Weeks 9–12: making it survive without me

The test of a fractional COO is not what happens while they are there; it is what happens when they leave. By the end of the first quarter the company should have:

1. A written operating rhythm — which numbers, which meetings, which owners — that runs whether or not I attend.
2. Documentation for the processes that were rebuilt, short enough that people actually use it.
3. A next-quarter plan with two or three priorities, each with a number attached, rather than a list of initiatives.
4. At least one person on the team who has grown into owning a function that used to route through the founder.

What a fractional COO does not do

Set expectations honestly, and the engagement goes better for everyone.

• They do not run the day-to-day. That is the operations manager's job; the COO builds the system the manager runs.
• They do not fix a broken business model. If unit economics are negative, operations can make the loss smaller, not turn it into profit.
• They do not replace the founder's judgement about the product or the market. They make sure that judgement, once made, turns into results.

How to tell it is working

Three signals, in order of appearance. First, within a month, the founder's calendar changes: fewer approvals, fewer "quick questions". Second, within two months, the weekly numbers become boring — they stop surprising anyone, which means the company can finally see itself. Third, within a quarter, one of the numbers you chose at the start has moved in the direction you chose, and everyone can explain why.

If none of those three has happened by day 90, the problem is either the diagnosis or the person. Either way, that is the moment to say so.