Should the COO own AI strategy in an SMB?

If the business has a COO, they are often the strongest natural owner - the function already runs change management, owns operational metrics, and lives inside the workflows that AI will touch. If the COO does not have time, the CFO is the next-best owner. Either way, AI ownership benefits from being assigned to an existing leadership role rather than spread across the team or treated as a project for IT.

Which operational metrics does AI actually move?

Five with the most consistent evidence at SMB scale. Cycle time on routine knowledge work drops 30-50% after a proper Claude rollout. First-response time on customer communications drops 60-80% with AI triage. Labour cost per unit of output drops 10-25% on AI-augmented workflows. Error rate on routine data entry and classification drops by half or more. Senior-staff time spent on routine work drops 40-60%, freeing time for higher-leverage activity.

What is the 90-day adoption sprint?

The COO-owned change-management programme that makes AI rollout stick. Week 1-2: name the workflows in scope, train the team, ship the first two. Week 3-6: weekly review of usage and outcomes per workflow, fix what is not working, kill what no one uses. Week 7-12: lock in the workflows that landed, measure the metric movement, write up the case study for internal and external use. Without this 90-day discipline, adoption fades and the rollout produces little measurable gain.

How does the COO know if AI is actually working?

Three signals. Usage rate (the AI workflows are being used regularly by the people they were built for - measurable in platform analytics). Cycle-time metrics (the relevant operational metric is moving in the right direction quarter over quarter). Team feedback (the team can name specific workflows that have changed their day). Without all three, the rollout is either superficial or not really rolling out.

What does AI failure look like for a COO?

Licences distributed with no rollout plan, three months later usage data shows half the team has not logged in, no measurable metric movement, and a creeping sense of cynicism about the next AI initiative. The fix is not buying more tools - it is owning the change-management work that should have happened on day one. The COO is the function that should never have been off this hook in the first place.

← All insights

Playbook·8 min read

AI for the SMB COO: where it actually moves operational metrics

If the CFO is the natural owner of AI strategy and governance, the COO is the natural owner of operational deployment and change management. Two complementary roles, both critical, and most SMBs only need to fill the role they actually have. Here is the COO's version of the playbook.

Why COOs are natural AI owners

Three reasons. First, the COO already owns operational metrics, which is where AI’s impact shows up. Second, the COO already runs change-management work, which is the harder half of any AI rollout. Third, the COO lives inside the workflows AI will touch - not in the strategy abstraction layer above them.

In SMBs without a dedicated COO, the founder or operations lead fills this role. The key is naming the owner explicitly. AI adoption without a named operational owner produces tools purchased and forgotten.

The five operational metrics AI consistently moves

1. Cycle time on routine knowledge work

Drops 30-50% after a proper Claude rollout with Projects, Skills and training. The metric is days or hours per common task (proposal turnaround, customer-comms draft, monthly report). The improvement is measurable inside one quarter.

2. First-response time on customer comms

Drops 60-80% with AI triage and draft-response workflows. The metric is hours from inbound to first reply. Particularly large impact in customer service and inbound sales functions.

3. Labour cost per unit of output

Drops 10-25% on AI-augmented workflows. The metric is $/unit of output. Easiest to measure on repetitive workflows: invoices processed, listings drafted, tickets resolved, transcripts summarised.

4. Error rate on routine data entry and classification

Drops by half or more on structured-extraction workflows. The metric is errors per 1,000 transactions. Particularly relevant in finance, billing and operational data flows.

5. Senior-staff time spent on routine work

Drops 40-60% when juniors get AI-augmented and stop interrupting seniors with routine questions. The metric is hours per week per senior on routine work versus higher-leverage work.

The 90-day adoption sprint

The single most important thing the COO does for AI rollout is run the 90-day adoption sprint. Without it, licences are distributed and most of the productivity uplift evaporates.

Weeks 1-2: Name and ship

Name the workflows in scope (specific, not abstract). Train the team. Ship the first two workflows. Set the metric baseline.

Weeks 3-6: Iterate

Weekly review of usage and outcomes per workflow. Fix what is not working. Retire what no one uses. Add two or three more workflows as confidence builds.

Weeks 7-12: Lock in

Lock in the workflows that landed. Measure the metric movement (versus the baseline). Write up the case study for internal use and for any future board, buyer or partner reporting. Plan the next quarter’s expansion.

Signals it is working

Usage: the AI workflows are being used regularly by the people they were built for. Measurable in platform analytics. If the usage data shows half the team has not logged in, the rollout has failed regardless of what the tooling looks like.
Metrics: the operational metrics targeted are moving in the right direction quarter over quarter. Movement does not need to be dramatic; it needs to be visible.
Team feedback: the team can name specific workflows that have changed their day. Not “AI is helpful in general” - specific, named workflows.

Signals it is failing

Licences distributed but usage data shows minimal engagement
No measurable metric movement after 90 days
Team cannot name specific workflows AI has changed
The rollout becomes “the AI project” rather than a permanent operational change

When these signals appear, the fix is not buying more tools. It is owning the change-management work that should have happened on day one.

How XLev helps COOs

For SMB COOs picking up the AI mandate, our typical engagement delivers the AI strategy workshop, installs the workflows identified, trains the team, runs the 90-day adoption sprint alongside the COO, and stays through to the next quarter’s expansion plan. The COO owns the change; we provide the implementation muscle.

Book a free 30-minute discovery call via the Contact page.

Frequently asked questions

Should the COO own AI strategy in an SMB?: If the business has a COO, they are often the strongest natural owner - the function already runs change management, owns operational metrics, and lives inside the workflows that AI will touch. If the COO does not have time, the CFO is the next-best owner. Either way, AI ownership benefits from being assigned to an existing leadership role rather than spread across the team or treated as a project for IT.
Which operational metrics does AI actually move?: Five with the most consistent evidence at SMB scale. Cycle time on routine knowledge work drops 30-50% after a proper Claude rollout. First-response time on customer communications drops 60-80% with AI triage. Labour cost per unit of output drops 10-25% on AI-augmented workflows. Error rate on routine data entry and classification drops by half or more. Senior-staff time spent on routine work drops 40-60%, freeing time for higher-leverage activity.
What is the 90-day adoption sprint?: The COO-owned change-management programme that makes AI rollout stick. Week 1-2: name the workflows in scope, train the team, ship the first two. Week 3-6: weekly review of usage and outcomes per workflow, fix what is not working, kill what no one uses. Week 7-12: lock in the workflows that landed, measure the metric movement, write up the case study for internal and external use. Without this 90-day discipline, adoption fades and the rollout produces little measurable gain.
How does the COO know if AI is actually working?: Three signals. Usage rate (the AI workflows are being used regularly by the people they were built for - measurable in platform analytics). Cycle-time metrics (the relevant operational metric is moving in the right direction quarter over quarter). Team feedback (the team can name specific workflows that have changed their day). Without all three, the rollout is either superficial or not really rolling out.
What does AI failure look like for a COO?: Licences distributed with no rollout plan, three months later usage data shows half the team has not logged in, no measurable metric movement, and a creeping sense of cynicism about the next AI initiative. The fix is not buying more tools - it is owning the change-management work that should have happened on day one. The COO is the function that should never have been off this hook in the first place.

Where this fits

AI Strategy Workshops

Half-day or full-day workshops with leadership. Walk out with a 12-month plan, not a slide deck.

See AI Strategy Workshops →Book a free 30-min discovery call