Operations are never transformed once and finished. They are continuously re-balanced — and the balance that was right three years ago usually isn't right now.
Every enterprise operation runs on some mix of three things: human talent, digital workers — ERP, SaaS, RPA, AI agents, workflow, automation — and as-a-service providers: BPO, GCC, centres of excellence, managed services. All of it powered by a common set of technological platforms, work methods such as SOPs, and standards.
That mix is company-specific, and it never settles. It shifts as technology matures, as geographical cost structures move, as talent markets tighten. Operations aren't transformed once and finished — they are continuously re-balanced across those three levers.
BizBlocz exists to help you navigate that shift with evidence instead of conviction.
Right now the force moving the mix fastest is AI, which is why most of what's here concerns where AI belongs in the balance — and where it doesn't. But the question underneath is older and will outlast it: for this specific piece of work, who or what should be doing it, and what is that worth?
Most of those decisions get made without an operations baseline. “Should we apply AI here?” can't be answered until someone has said what here is — the actual work, who does it today, at what cost, with what exceptions.
Vendors answer that with their own product. Consultancies answer it with a bespoke study. Neither leaves you anything you can check.
Two layers.
The BizBlocz business operations process taxonomy blends the universally accepted APQC Process Classification Framework with the process language and frameworks of the top ERP and enterprise SaaS systems — SAP, Oracle, Salesforce, ServiceNow — adjusted by our founder's experience across all corporate business processes.
That blend is the point. APQC is vendor-neutral by design: it tells you what the work is, never what runs it. The ERP layer tells you how it's actually implemented. Thirty years of delivery tells you where both are wrong in practice.
127 subprocesses across 11 business areas. None of them invented — we didn't build a private language for work that already has one. Business Operations →
Against every subprocess, two dimensions:
The AI mix — which kinds of AI actually apply to this work. Most processes need a blend, and the blend differs for almost every one of the 127.
The benchmarks — the target range of improvement you can expect from implementing that mix. Not a single number. A range, with a stated confidence.
A number is only as good as the question nobody asked about it. So:
Sources are not equal, and we score them before they count. A study from a major consultancy carries more weight than a technology vendor's own case study — the vendor has a product to sell. Each source is scored by class, then adjusted again for what kind of document it is: a named customer case study with real figures counts differently from a marketing page.
Nothing rests on one source. Findings are triangulated across independent publishers. Where a figure appears in only one place, it is graded down, not hidden.
Outliers are detected and set aside rather than allowed to drag an average.
And when subprocess-level evidence is thin, we say so and widen carefully — falling back to process level, then business area, then cross-functional research, at progressively reduced weight. That waterfall is visible in the output, not buried.
The result is a confidence grade, not just a number. Strong, Solid, Moderate, Thin or None — reflecting how much independent evidence actually exists for that specific subprocess. Most sit at Moderate. Some are Thin. A handful are None, and we publish that too.
That last part matters more than it sounds. Publishing where the evidence runs out is the difference between a research base and a sales deck.
The evidence base is built in rounds, not scraped once — each a systematic sweep across the taxonomy rather than a search for confirming numbers.
Consultancies, research firms, universities, industry associations and vendor case studies — each weighted by class, not treated alike. Every source carries its publisher, document title, publication date and URL. You can trace any figure on this site back to where it came from, and judge for yourself whether a study still holds.
The corpus favours primary documents over aggregators. Where a figure could only be reached through a secondary republication, the weighting reflects that. How these profiles are built →
The taxonomy itself. How a modern company runs, area by area, down to the subprocess: what the work is and what executes it today.
Enter your own cost base for a process and get a range, a confidence grade, and the sources behind it. Free, no signup.
The research and methodology written up in full, including where we think the field is wrong.
Diego Navia has spent 30+ years modernizing global operations by making those transitions work.
At Accenture, some of the world's first and largest global ERP transformations and shared services. At KPMG, standardizing global operations and using technology to power M&A activity. At PwC, he co-founded the global Service Delivery Network and built it to 12,000 people across nine hubs — site selection, legal structures, practice acquisitions, technology stack, service architecture — generating over $1B in billable revenue at double-digit CAGR across seven years. It now employs around 70,000. Since then, advising leading global organizations and consulting firms on the next generation of modern operations and the trends shaping them: GBS, BPM, RPA, workflow platforms, AI, GCCs.
He has worked in more than half of US states, 30+ countries and across five continents. MBA, Duke's Fuqua School of Business. Industrial Engineering. Operations certificates from MIT and the University of Pennsylvania.
The conviction underneath all of it is about data: the most dangerous data point is a confident-sounding one nobody has questioned.
BizBlocz is a brand of Navteva Services LLC. Questions: get in touch →