Operations · 8 min read · June 11, 2026

Back-office automation: where the savings actually come from

McKinsey says about 30% of the activities in most jobs are automatable. Deloitte found a 32% average cost reduction where intelligent automation scaled beyond pilots. How those headline numbers translate, and don't translate, to a real mid-market back office.

Two pieces of research anchor every serious conversation about automation economics. McKinsey Global Institute's landmark analysis found that with technology already demonstrated, about 30% of activities in most occupations are automatable: not 30% of jobs eliminated, but 30% of the hours inside jobs. Deloitte's 2022 survey of 479 executives, meanwhile, found that organizations that scaled intelligent automation beyond pilots reported an average cost reduction of 32%.

Both findings deserve careful reading. They describe what disciplined programs achieve in the workflows they target, not a guaranteed haircut on your total payroll. The savings live in specific places: data entry, document handling, status communication, scheduling, reconciliation. The judgment calls, relationships, and exceptions stay human.

The mid-market translation

In a dental group, the automatable 30% looks like insurance verification, intake re-keying, and confirmation calls. In an accounting firm, it's document chasing and status updates. In a home health agency, it's referral intake, onboarding checklists, and report assembly. The pattern is always the same: high-volume, rule-based, multi-system workflows that someone currently does by hand because no one had time to fix them.

The sound methodology is baseline-first: document the hours and error rates of a workflow before automating it, automate the highest-volume workflow first, then measure against the baseline. If a vendor proposes automation without measuring your current state, they're selling software, not savings.

Why this is now a mid-market conversation

Intelligent automation used to require enterprise budgets, which is why the Deloitte numbers historically came from large shared-services organizations. Modern AI tooling has sharply reduced the cost of building these workflows. Savings that once required a consulting army are now within reach of a mid-market company or a multi-site group, if the work is scoped honestly and measured against a real baseline.

Sources

Want the next analysis first?

Keep reading

Next step

Find out where performance is being lost, and what it is worth to fix.

The Performance Diagnostic runs for two to four weeks at a fixed fee and ends with a baseline, three priced improvement plays and a 90-day plan. The findings are yours whether or not we continue.