Not every cost can be re-earned

A client I had argued with about scope got their migration in the end. Their records moved from the old CRM to the new one, the project closed, and the invoices cleared. What they received was an export from one system loaded into another, along with the job of working out what to do with it. They got a migration. They did not get a working system.

It didn't go wrong. It just didn't go right.

There is no decision in the sequence I can point to as the mistake. What there is instead is a set of people behaving sensibly on the basis of what the previous round had taught them, and one cost that never made it into the teaching.

What happened first

An earlier phase of migration work at the same organisation went out with scope that was never locked or properly articulated. Requests kept arriving, and because nothing had been named at the start, all of them counted as changes. The timeline held anyway. A development team worked fourteen-hour days and into nights, and the client received close to everything they had asked for, on the number they had originally agreed.

From where the client sat, that was not a rescue. It was simply what happened. They asked for a great deal on a fixed budget and they got it.

What that taught each seat

The client learned that the budget was sufficient. That is not a naive conclusion. It is the correct inference from the only evidence they had, and the supplier had provided the evidence. Nobody told them the number had been met by people working nights, because that is not the sort of thing that appears in a status report.

The business learned that effort absorbs estimating error. If a shortfall can be closed by the team rather than by a conversation, the conversation becomes optional, and the cost of an optimistic number never appears anywhere it can be seen.

The developers learned what the job is. Not as a theory about the industry, but as a practical expectation about how the next one will go.

The second phase

By the time the next piece of work was being sized, I was saying that what the client described as an installation was a process rebuild, and that the scope was larger than the number in their head. The price moved to match. The client was not pleased. The organisation I was consulting to trusted my commercial instincts less afterwards, because I had made a deal harder at a point when they wanted it easier.

Everyone in that exchange was consistent with what they had learned.

The client rejected the figure because the last phase had demonstrated that figures of that size were unnecessary. The business, facing a year well short of target and the real possibility of getting none of it, took a reduced piece of work that would close, which is what the role is for. The developers, watching it descope, were asking the narrow and repetitive question: am I going to have to do a bad job again. Last time the constraint was the timeline. This time it was the budget. The specific reason changes and the experience does not.

The work went ahead in its reduced form and produced the export.

It should be said that this was not cheap for the client either. A CRM migration of that shape is a several hundred thousand dollar undertaking, and they had set out to have it done for a fraction of that. What they hold now is something the next vendor will assess, accurately, as not fit for use, and the second attempt will cost more than the first one would have, because it will include undoing this one. They did not save the money. They postponed a larger version of the same bill.

The argument I lost

I did raise the burnout from the first phase. It was not a revelation to anyone. They knew the budget had been miscalculated and they knew how the gap had been closed. Nobody was pretending otherwise.

The answer was that revenue could not be allowed to suffer over it. That is a serious position rather than a callous one. A business short of its number is dealing with something immediate and quantified, and set against that, a period of overwork on a phase that had already shipped registers as an inconvenience that has passed.

What I could not do was make the asymmetry visible. Revenue that is missed this year can be earned next year, possibly from a client who does not exist yet. That is not a comfortable argument to make, because the replacement is hypothetical and the shortfall is real. But it is recoverable in principle, and there are many routes back to it.

What people spend of themselves to close an estimating gap does not work that way. Sustained overwork does not get returned when the quarter improves. The health, the years, the accumulated tiredness that changes how somebody feels about the work: none of that has a route back. It is the only cost in the discussion that is genuinely permanent, and it was the only one without a figure attached to it.

A trade-off conducted in a single currency will always resolve towards the side that has a number. The irreversible cost had no number, so it functioned as a consideration rather than as a cost.

That is why it lost. Not because anyone dismissed it. I did not manage to change that, and I still think it was the right argument to be making.

What the answer probably looked like

I put it as a choice between funding the work properly and living with what happens when you don't. That framing was too narrow, and the narrowness was mine.

There was a third version nobody built. A scope smaller than the one I was defending, but coherent enough to produce something usable. Shaped with the development team in the room rather than handed to them. Explicit about what was being left out and why, so the gaps were decisions on the record instead of discoveries later. And if a stretch was still needed at the end, one that people agreed to with the reasons in front of them, rather than one that arrived as the only remaining way to keep a promise they had not made.

That version is harder to construct than either of the positions being argued, which is probably why neither of us reached for it. It requires revenue, quality and the people doing the work to be held at the same time, and most commercial conversations are set up to hold two of the three.

On the cost that had no number, I do not think the answer is to give it one. A figure on a sheet gets traded like any other figure, at whatever price the sheet will bear. Noting it and moving on is not enough either. It needs to be held consciously, by people who understand that the effects run further than the phase being discussed: into how someone feels about the next project, and the one after that, and sometimes into things that have nothing to do with work at all.

What it cost

The engagement ended not long afterwards. There was no falling out and nothing was said about not working together again. It was a quiet parting, with something left in the room. When the phase came up later, the shortfall was attributed, gently, to a consultant who had taken a stand that did not come off.

From where they were standing, that reading was fair. I had made a deal harder, the work went ahead in a smaller form regardless, and the client had been irritated in the middle of it. From outside, that looks like commercial misjudgement, and being right about the size of the work does very little to change the picture. Nobody audits an estimate that was never used.

The part that stays with me

The developers' nights in the first phase became the client's budget in the second. The client's budget became the descope. The descope becomes the bill the client has not yet received. Every step is defensible on its own, and the direction is set by a cost that nobody recorded.

That is the mechanism worth naming. Costs without numbers do not enter the record, and costs that do not enter the record teach nobody anything. The client could not learn from the nights because the nights were never reported. The business could not weigh them because they had never been priced. The only people who held that information were the ones with no say in the estimate.

So the useful question at the start of a piece of work is not what the requirements are. It is what the last one taught everybody, and which of the costs was carried by people who never got to put a number on it.

Client and organisation details left out deliberately.