Corporate Functions of the Future Won't Look Like Functions at All
B · Useful, limited BCG organisation-design partners set out a 'functions of the future' model for AI-run G&A; a coherent, practical framework with useful benchmarks, but the numbers come from unnamed clients and BCG's own surveys with no method shown.
Summary based on the full text·Original: about 15 minutes to read
In short
A May 2026 BCG article by Matthew Marchingo, Emmanuel Sissimatos, Kevin Kelley, Pragya Maini and Tarun Kajeepeta. It argues that finance, HR, legal, compliance, procurement and IT should stop running as separate departments and become platforms that run end-to-end workflows (hire-to-offer, source-to-pay, record-to-report) with AI agents doing routine work and small human teams handling judgment, exceptions and governance. Adding AI to unchanged functions, it says, is why most AI programmes show no value.
Key numbers
- 3% to 5%G&A cost as share of revenue at best-in-class companies, vs near 8% for most others
- 25% to 40%reduction in functional headcount over time that BCG says many organisations see, mostly in middle layers
- 0.3% to 1%of annual revenue invested over 18 to 36 months, with payback in 12 to 30 months
- roughly 80% vs 25%CEOs saying they are well positioned to adopt AI vs those with the foundational prerequisites in place (BCG survey of 200+ executives)
- as many as 95%companies adopting AI that BCG observes failing to capture meaningful value
As reported by the source.
English·Web page·opens bcg.com
What it reports
- Benchmark: leaders run G&A at 3% to 5% of revenue vs about 8% for most; for a $10 billion company that is $300 million to $500 million a year.
- One technology company requires "AI justification" before new hires; functional headcount stayed flat while revenue grew about 25%.
- Claimed effects: G&A cost halved in some cases as AI automates 60% to 70% of routine jobs; cycle times cut by up to 90% when periodic reporting becomes always-on.
- Causes of failure: silos optimising their own goals, weak central governance, and change management without KPIs or redefined roles and decision rights.
- CEO survey: upward of 80% say 40% or more of their direct reports sit in G&A; 70% say those roles should be consolidated, often under a CFO plus a chief administrative officer.
- Seven building blocks: AI-first functional strategies; global business services as an execution engine handling 30% to 40% of activities; lean centres of excellence; modular ERP and a single data backbone; digital fluency as a hiring and promotion requirement; leadership for human-plus-agent teams; outsourcing as an accelerator.
- New metrics: cost per exception resolved instead of cost per invoice; combined human-plus-AI capacity instead of FTE utilisation.
- Sequence: 4-6 week diagnostic, 4-5 months of first automations, then scale-out in the second half-year.
- Cases: a global bank cut compliance costs 35% to 45% and breaches by 40%; an oil and gas major automated about 70% of transaction clearing and cut invoice cycle times by 80%.
How they got it
Consultancy synthesis from BCG client work and a BCG survey of more than 200 corporate executives. No sample details, definitions or baselines are given for the outcome figures.
Limits
- All client results are anonymous and self-reported.
- The 95% failure figure is described as a BCG observation, not a measured study.
- Investment and payback ranges are BCG's own estimates and support its service offering.
- Workforce consequences are acknowledged but treated briefly.
Why it matters for companies
For HR and other support functions, the useful parts are concrete: a G&A cost benchmark to test against, the "AI justification" rule for new hires, and the metric shift from cost per transaction to cost per exception. The model also implies an HR function organised around end-to-end workflows rather than sub-functions, and fewer coordination-heavy middle roles. Treat the percentages as targets to validate in a pilot, not as expected results.
Summary based on the full text. Read the original for the full argument and data.Read the original on bcg.com
- Tags
- Operating model