The machine nobody designed

Ask a technology leader what their estate costs, and you will usually get an answer within a few 

thousand. Ask what it delivers and the answer changes shape — it becomes a list of systems, or a 

list of projects, or an apology.

That gap is not a reporting problem. It is the symptom of something more ordinary and more 

expensive: almost nobody designed the thing they are now running.

Estates accumulate. They are not built.

Very few technology estates were planned as a whole. They arrived in pieces. A system came in with an acquisition. A tool was bought in a hurry because a process was failing that quarter.

A platform was chosen by a vendor who has since moved on, configured by a contractor who left, and inherited by a team who were told not to touch it. Each of those decisions was reasonable on the day it was made.

Put twenty of them end to end over ten years, and you do not have an architecture. You have sediment.

The interesting part is that this is rarely visible from inside the IT function, because IT is 

usually not where the problem sits.

The failure is in the joins, not the systems

When I look at a business properly, the individual systems are usually fine. Finance's platform 

does what finance needs. Operations chose sensibly. The CRM is a normal CRM.

What is broken is everything between them.

Work travels through the organisation and, at some point, it leaves a system. It becomes a spreadsheet. It becomes an email with an attachment. It becomes a person — often a specific, 

named person — who knows how to reconcile two sets of numbers that were never designed to be 

compared, and who has been quietly doing it every month for four years.

Nobody has ever costed that person's time as technology spend. It is technology spend. It is the 

price of an interface that was never built, paid monthly, in staff hours, by whoever is closest to 

the gap.

And because the failure is between departments, it belongs to nobody. Every department can defend its own technology on its own terms. Nobody owns the joins.

Governance is not a compliance register

The word governance has been damaged by association. For most people it now means a folder of 

policies, an annual audit, and a control matrix that is accurate for about a fortnight after it is 

written.

That is not what the word means, and it is certainly not where the value is.

Governance is how technology decisions get made. Who is allowed to buy something. Above what value someone else has to agree. Who owns a system after the project that delivered it has closed. What standard a new tool has to meet before it joins the estate. Which obligations apply, and to which system, in which country you operate in.

Written down, that fits on two pages. Undocumented, it is the reason a company ends up with three 

tools doing the same job and no way of noticing.

Alignment is the outcome people want. Governance is the mechanism that produces it. Buying the 

outcome without the mechanism is how organisations end up repeating the same expensive tidy-up every three or four years.

This is not a problem that starts at a thousand employees

I want to be precise about who this affects, because the assumption is usually wrong.

It is not about size, and it is not about how many countries you operate in. It is about how many 

distinct ways of working you have, and how those meet your technology.

A forty-person business with six departments, six tools nobody chose deliberately and one person 

holding the reconciliation together has exactly the problem described above. The difference is that 

at forty people it is still cheap to fix. At four hundred, the same tangle has contracts, headcount 

and political ownership attached to it.

Multi-country and multi-entity organisations are the harder version of this. They are not a 

different species. Everything above applies, with jurisdiction on top and with the estate diverged by 

location as well as by department.

The signals are consistent, whatever the size:

• Two departments solved the same problem separately, and the data now has to be reconciled by hand.

• A process is documented up to a point, after which somebody maintains a spreadsheet everyone depends on.

• Each team can describe its own systems, and nobody can describe how they connect.

• Technology cost has grown for two years without a matching change in what the business can do.

• Tools are chosen inside departments, and the effect on everyone else is discovered afterwards.

• An acquisition, a merger or a founder's departure left two of everything.

• Ways of working that suited fifteen people are being applied to fifty.

What looking at it properly involves

There is no clever instrument for this. It is done by taking the whole organisation as the unit of 

analysis rather than the IT department, and examining five things.

What the business actually does — functions, value chains, where revenue is earned, and cost is 

incurred, and what each team is genuinely accountable for rather than what the org chart claims.

What the technology actually is — every system, integration, licence and contract, with an owner 

and an annual cost against each one. A surprising number of organisations have never had this on a 

single page.

Where the two meet — and, more usefully, where they stop meeting, and the work becomes manual.

How decisions are made — spend authority, system ownership, change approval, standards, and the obligations that apply in each jurisdiction.

Where you are exposed — the skills that sit with one vendor, or one person, and what happens the 

week they are unavailable.

Out of that comes a target operating model and a sequenced roadmap. But the finding that changes the most minds is usually the least expected one.

The answer is often less, not more

A material proportion of what this work produces is subtraction. Systems to retire. Contracts to 

consolidate at renewal. Licences being paid for by two departments who each believe they are the 

only user. Processes that should be simplified before anybody automates them, because automating a bad process just produces the wrong answer faster.

That is worth saying out loud, because the reflex when a business feels inefficient is to buy 

something. Very often the estate does not need another platform. It needs someone to look at the 

whole thing at once — which is a thing almost nobody inside the business is positioned to do, 

because everybody owns a part of it.

⸻

A test you can run this week. Ask three people in three different departments to describe how a 

single piece of work — an order, a hire, an invoice, a claim — travels from the moment it arrives to 

the moment it is finished. Not the system it lives in. The whole path.

If you get three different answers, or if any of them contains the phrase "and then I send it to 

someone", you have found the thing that is costing you money.

⸻

Oghenetega Gharoro-Akpojotor is the founder of Toga EMEA, a boutique technology management 

consultancy working across Europe, the Middle East, Africa and APAC. Global Governance is Toga's 

assessment and operating model practice.