Nobody Fails a Digital Transformation Because of Bad Software

 

Here's a pattern I keep running into: a company spends months picking the "right" platform for its digital transformation, gets leadership sign-off, forms a steering committee, does the big kickoff. Six months later, almost nothing about how people actually work has changed.

The software was never the problem. The problem is that nobody wrote down who's actually allowed to make a call when something breaks — and something always breaks.

I went deep on this in a longer piece on governance models, but here's the short version for anyone who doesn't have twenty minutes to spare.

Four ways companies actually structure this

Centralised — one office owns every decision: budget, vendors, architecture, priority calls. Great for consistency. Terrible the moment eight people are approving requests for forty departments. If your team is waiting two weeks for something that should take two hours, this is why.

Federated — the center owns shared stuff like cloud and security, business units own their own execution. Most mid-size and large companies land here eventually, and for good reason: it puts decisions with people who actually understand the problem. Where it breaks is integration — every unit makes a locally smart call, and none of the systems talk to each other.

Agile governance — decisions get made as close to the work as possible, with narrow escalation paths. This only works if leadership genuinely lets go and accepts that team-level calls will sometimes be wrong. Fair warning: if your "agile governance" is really a monthly steering committee reviewing progress, that's centralised governance wearing a different shirt.

Hybrid — centralised for the cross-cutting stuff (security, data standards), federated for domain work, agile at the team level. This is what most real transformations end up running. It only works if the boundaries are actually clear — who owns what decision — otherwise you get constant escalation or people quietly going rogue.

The part almost everyone skips

Org charts get all the attention. Decision rights get almost none. Before any of this starts, someone needs a straight answer to:

• Who can approve spend over X?
• Who can change scope mid-project?
• Who can kill something that's clearly failing?
• Who actually owns the vendor relationship, in practice, not on paper?

One thing that helps: separate inform from approve from veto. A huge amount of governance drama is just someone with inform rights acting like they have veto rights, because nobody ever told them otherwise.

Change management is not a comms plan

Town halls, mandatory training modules, a nice email from the CEO — none of that actually changes behavior. What does: people watching someone they respect work differently. A department head pulling their own numbers from the new dashboard instead of asking an assistant to email the spreadsheet version. That's the whole mechanism. If your rollout is built on messaging instead of visible behavior at the top, you'll get compliance where it's mandatory and workarounds everywhere else.

The legacy system nobody wants to talk about

Every roadmap shows a clean future state. What it conveniently leaves out is the 12–24 months of legacy system work standing between here and there. Companies generally pick one of two paths: strangler fig (wrap the old system, migrate piece by piece, retire it slowly — slower, safer) or hard cutover (pick a date, run parallel for a while, cut over — faster, riskier, needs real preparation). Most mid-market companies do neither properly. They "integrate" the legacy system and never actually replace it, which just becomes a permanent maintenance tax.

Measuring this the right way

Most transformation dashboards track inputs — initiatives launched, people trained, processes "digitised." None of that tells you if anything got better. What actually matters:

• Operational — cost per transaction, time to onboard, error rates, before and after
• Adoption — active users vs. licensed seats, how often people use workarounds instead of the actual system
• Strategic — revenue tied to new capability, speed to market, customer satisfaction in the areas you transformed

"We finished the ERP rollout" is a milestone. It's not a result.

If you're picking a partner for any of this

A few questions that rarely come up but should: how do they handle scope creep once they discover complexity you didn't know about? What does knowledge transfer look like once they leave — are you left dependent on them, or genuinely capable on your own? Do they work inside your governance, or bring their own, and have you actually agreed on which before day one?

If you're at that stage, Agami's digital transformation services and automation solutions pages go into how they structure this, and their case studies are worth a look if you want to see what execution actually looked like for other companies rather than just a sales pitch.

---

None of the four models is inherently better. The transformations that actually land are the ones where somebody wrote down who owns what before the ambiguity turned into conflict, kept architecture debates from freezing delivery, and measured outcomes instead of milestones.

I wrote the full breakdown — decision-rights frameworks, the strangler-fig vs. hard-cutover tradeoff, ROI metrics, and how to vet a transformation partner — here, if you want to go deeper on any of this.