Five Hands Down

A brand officer at a large company gets a budget approved. Say it is a serious number, and say the brief is good, because it usually is. That person knows their market, has thought carefully about what the business needs, and has fought internally to get the money released.

Then the work leaves the building.

The agency of record takes it, because they hold the relationship. They keep the strategy and the client contact and pass the build to a production partner, because they do not employ engineers. The production partner takes their share and passes the difficult part to a specialist studio, because the platform is not one they know well. The studio is busy this quarter, so they route it to a network they trust. That network gives it to one person, or two, who are good, who are available, and who are being paid a fraction of a number they will never see.

Nobody in that sequence did anything unusual. Every step was defensible in the room where it was decided. And the person who ends up writing the software has never spoken to the person who wrote the brief, will never speak to them, and is working from a description that has now been retold five times by people optimising for different things.

1. The money, first, because it is the easy part

Give every hand in the chain a thirty percent margin. That is not a scandalous number in this industry. It is roughly what a firm needs to cover its people, its overhead, its risk on the engagement, and its profit.

Five hands at thirty percent each leaves 16.8 percent of the original budget for the work itself.

At a gentler twenty percent per hand it is 32.8 percent, which sounds better right up until you say it in plain language: two thirds of the budget was spent on the act of passing the work along.

Now notice what the last party in that chain is being asked to do. They have been handed the entire technical risk of the project, on a sixth of the money, with no access to the person who can answer a question, and no upside if it goes well. There is exactly one way to deliver profitably in that position, and it is to do the minimum that will pass whatever inspection is coming. Tests, documentation, error handling, security review, and the second opinion on the data model are the first things to go, because they are the things nobody in the chain is going to check.

This is not a story about bad engineers. It is a story about what any rational engineer does when handed a sixth of a budget and a third of a brief.

2. The brief decays too, and nobody audits that one

Money is at least counted. The brief is not.

Assume each handoff preserves 85 percent of the original intent, which is generous. It assumes every intermediary is competent, attentive, and honest, and that nothing is lost because a document was written by someone who had not been in the room.

Five handoffs at 85 percent leaves 44 percent. At 80 percent it leaves 33 percent.

What goes missing is never the feature list. Feature lists survive handoffs perfectly, because they are easy to copy. What dies is everything that made the brief good: why this matters now, which of these requirements is load bearing and which was a nice idea in a meeting, what the officer is actually being measured on, which failure would be embarrassing and which would be catastrophic, and the particular thing about this business that a competitor got wrong last year.

That is the context which turns a specification into a decision. It is the first thing lost at every hop, and its absence is invisible in the deliverable.

The feature list arrives intact. The reason for it does not.

3. And the feedback path is the same chain, backwards

Here is the part that turns a bad structure into an undetectable one.

When the engineer at the end hits a problem, the question has to travel back up five links to reach somebody who can answer it. Every link is a queue, and every link is staffed by someone whose interest is to resolve it locally rather than escalate, because escalating means telling the party above you that something is wrong.

We have written elsewhere about what a slow decision does to a schedule. Sixty decisions at three days each is forty five days of pure waiting. Through a five link chain the wait is not three days, and most questions never make the trip at all. They get resolved by a guess, at the bottom, by the person with the least context in the entire arrangement.

So the brand officer's status report is green. It is green at every layer, and each layer is being sincere, because each layer only knows what the layer below chose to tell it. The officer finds out at launch, or later, when something quietly does not work and nobody can explain why.

Then the conclusion drawn is almost always the wrong one: that the vendor was bad, or that digital work is unpredictable, or that the budget was too small. The budget was not too small. Five sixths of it went somewhere else.

4. What we do instead

One contract, and the people doing the work sit at the table with the person who owns the outcome.

That means engaging with the office that actually holds the problem, which in practice means the CXO function rather than a procurement layer three floors below it, and it means the engineers and designers being present when the intent is explained rather than receiving a summary of it. Not a kickoff appearance. Present for the argument, because the argument is where the real brief lives.

We should be precise about what we are not claiming, because the honest version is more useful than the heroic one.

We are not claiming we work alone. We run a delivery network, and specialists in it do work for our clients. The difference is not the absence of partners. It is that the network is disclosed, governed by one standard, priced once, and contracted for by us. You know who is in it, we carry the accountability for it, and nobody in it is reselling your work to somebody you have never heard of at a margin you were never shown.

We are also not claiming this is always cheaper on the day. A direct engagement with the people who will build the thing usually quotes higher than a chain does, because the chain's number was arrived at by discounting the work, not the margin. What it is, reliably, is cheaper by the time it is finished, and the difference shows up as the thing working.

5. The quorum

The second half of this is about who is in the room, and it is the part we care about most.

Our teams are deliberately mixed. Small pods of young engineers work directly under people who have genuinely done the thing before, on real client work with real consequences, with a serious quorum held around every significant decision. Not a review at the end. Present while the decision is being made.

The reason is not sentimental. Judgement is the one thing in this profession that has never successfully been written down. You can document a process, a standard, a checklist, an architecture. You cannot document taste, and taste is what decides which of nine plausible options is the one that will still look right in three years. It transfers by proximity or it does not transfer at all.

The broker chain destroys exactly this. The senior people are at the top of the chain, holding relationships and writing decks. The junior people are at the bottom, building the thing. The two ends of the chain never meet, so the seniors stop learning what is actually hard to build and the juniors never learn why one option is better than another. Both halves get worse every year, and the industry calls this normal.

Concentrating strong people and putting everybody else close to them is not a nice culture. It is the only known transfer mechanism for the thing we are actually selling.

6. Speed and taste come from the same structure

We say we are about speed and taste, which sounds like an aesthetic claim. It is a structural one.

Speed comes from the chain being short. A decision made in the room takes hours. The same decision made through five intermediaries takes weeks, and often is not made at all. Nothing else about our engineers is faster than anybody else's. The path their questions take is.

Taste comes from the senior person being close enough to intervene while it still costs nothing to change. Taste applied at the end is called rework, and rework is why projects overrun. Taste applied in the room is invisible and free, and it is the entire difference between a thing that works and a thing that merely functions.

Both properties are consequences of the same decision, which is to keep the distance between the person who decides and the person who builds as close to zero as the work allows.

7. Where the talent goes when nobody builds this

The last part of the argument is about the people, and it is the reason we are building a company rather than a practice.

Skilled people leave the places that trained them, at rates most people would not believe. Across sub-Saharan Africa, the Caribbean and the Pacific, someone with a tertiary education is around thirty times more likely to emigrate than someone without one. Roughly 18 percent of everyone from a low income country holding a bachelor's degree now lives in an OECD country. In many small island states, high skill emigration exceeds 70 percent. And the share of all international migrants who hold a tertiary degree rose from 28 percent to 38 percent between 2000 and 2020.

The usual explanation is pay, and pay is real. But it is not what the people themselves report as the thing that decides. Stack Overflow's 2025 developer survey found the leading drivers of job satisfaction were autonomy and trust, competitive pay, and working on real problems, in that shape, with three quarters of respondents describing themselves as complacent or unhappy where they are. People do not primarily leave for a number. They leave environments where their capability is used well below its level, and they will cross an ocean to find one where it is not.

That is the loss worth naming properly. An engineer working five links down a broker chain, on a sixth of a budget, on a brief they cannot question, is a national asset deployed far below its capability. The country paid to educate that person. The industry then employs them in a structure that wastes most of what they can do, and is then surprised when they take the first opportunity to work somewhere serious.

Our position is that the environment is the retention mechanism, and that it is buildable. Real problems for real clients, proximity to people worth learning from, standards that are enforced rather than described, and a record of what somebody achieved that a third party can verify without asking us. Do that in a place, and the calculation people are making when they consider leaving changes, because the thing they were leaving to find is now available where they already live.

We are not sentimental about this either. Whoever builds that environment ends up holding the supply of people everybody else is short of. It is the most commercially valuable thing in this industry and it happens to also be the right thing to do, which is a rare and pleasant alignment.

8. What this costs us

Four things, so this reads as a position rather than an advertisement.

Direct engagement removes our hiding place. When there is no agency of record above us, there is nobody else to be blamed for a bad quarter, and there is no brand to borrow credibility from. We have to be the name on the door, which is harder and slower than being somebody's subcontractor.

Refusing to be a broker costs revenue we could easily have. Work arrives that we could take and pass along at a margin without touching it. That is free money and we do not take it, because the moment we do, we are the third hand in somebody else's chain and every argument in this paper stops applying to us.

The quorum is expensive. Senior people sitting in rooms with junior people is time that could have been billed at senior rates. It is priced into what we charge, and we would rather explain that than pretend it happens for free.

And pods do not absorb shocks. Six people cannot become sixteen in a week without becoming the thing described in our first paper. When scope doubles, we sequence it and say so, which some buyers experience as inflexibility right up until the point where it turns out to be the reason the date held.

9. What to do with this, even if you never hire us

If you are the person whose budget this is, three questions will tell you almost everything.

Ask who is actually going to write it, and ask for their names in the contract. Not the account team. The hands.

Ask how many organisations sit between your signature and those hands, and what each of them takes. A firm that cannot answer that quickly is not being coy. They usually do not know, which is worse.

And ask how a question from the person building it reaches you, and how long that takes. If the answer involves more than one intermediary, you already know what the quality will be, and it will not be because anybody involved was bad at their job.

The chain is not a conspiracy. It is just what happens when nobody designs the alternative. We think designing the alternative is the whole business.

References

1. 01Frédéric Docquier and Hillel Rapoport, and subsequent literature summarised in The brain drain from developing countries, IZA World of Labor, wol.iza.org
2. 02World Bank KNOMAD, Brain Drain, Gain, and Circulation, Working Paper 19, on tertiary emigration rates by region, worldbank.org
3. 03Stack Overflow, 2025 Developer Survey, on satisfaction drivers and how developers describe their current roles, stackoverflow.blog
4. 04Tigotek, Speed Is a Design Decision, on decision latency inside a schedule, /papers/speed-is-a-design-decision
5. 05Tigotek, The Coordination Tax, on what happens to a team that grows instead of splitting, /papers/the-coordination-tax
6. 06Tigotek, Nobody Owns the Gap, on the eighteen months between a graduate and an engineer, /papers/nobody-owns-the-gap