单一负责人的复利效应

The short version: A one-owner advantage does not stay a fixed size. It widens. Each decision made with full memory of the last one becomes the base the next decision starts from, while a rotating cast of vendors restarts near zero at every handoff. Left to run long enough, that widening gap is the entire moat.

Why does the advantage widen instead of holding steady?

Under one owner, a given decision is made with the full, untranslated context of every decision before it: what was tried and quietly dropped, what a client actually meant the first time they mentioned something offhand on a call, which fix held and which one had to be reversed six months later. None of that context is a deliverable. It never gets written down in full, because writing it down in full was never anyone's job. It simply stays in the one mind that made the decisions.

Split that same work across a rotating cast of vendors, specialists, or internal hires, and most of that context does not survive the first handoff. What arrives at the next stage is a brief, a deck, or a handover document: a compressed, partial record of what was decided, rarely of why. The next decision gets made against that thinner record, which is a fine start once. Repeated over twenty decisions, the two paths do not stay parallel. One keeps compounding on a full base. The other keeps compounding on whatever fraction of the original signal survived translation, and that fraction shrinks a little at every additional handoff.

What actually gets lost at a handoff, and what survives?

A handover document is not nothing. It is also not the thing that actually compounds. What tends to survive a handoff, and what tends to disappear with the person who made the decision, look different in practice.

• The reasoning behind a decision, not just the decision, is usually the first casualty: a document can record what was chosen far more easily than why the other options were rejected.
• The near-misses and discarded options rarely make it into any document at all, so the next owner sometimes retries an approach that already quietly failed once.
• The specific language a client actually uses, and what it really means to them, degrades a little with every retelling.
• Trust built through repeated, direct contact does not transfer with a file. It has to be rebuilt, slowly, by whoever inherits the relationship.

Take one small, concrete decision: writing alt text for an image. Google states plainly that the same alt text does two jobs at once, describing an image to someone who cannot see it and helping Google's own systems understand what the image shows (Google, to re-verify). Written once, by someone weighing both jobs at the same time, it is a single decision that pays twice. Written by an accessibility hire this quarter and quietly overwritten by an SEO hire next quarter, neither aware the other had a reason for touching the same sentence, it becomes two competing decisions fighting over one line of text. Multiply that pattern across every small decision on a site, over years, and the difference stops being one bad edit. It becomes which direction the site's decisions are accumulating in.

Is this the same argument as choosing one owner over three vendors?

Close, but not quite, and the difference is worth being precise about. The one-mind advantage makes the case for that choice directly: three specialists each optimizing their own scorecard, a full-service agency that moves the same seam inside one building instead of removing it, or a single owner accountable for all three pillars at once. That is an argument about a single moment, the way work gets staffed on day one.

This is an argument about what happens after that choice, across the years that follow it. Choosing one owner is the entry price, paid once. What compounds, or fails to, is everything that happens on every day after that choice, and that part of the argument has a shape of its own, worth examining separately from the staffing decision itself.

What does the gap look like once entropy is left unmanaged?

Entropy, not stability, is the default state of a website nobody actively owns across time. The WebAIM Million census of the top one million home pages found 95.9 percent carry at least one automatically detectable WCAG failure (WebAIM, to re-verify). Most of those pages were not built broken on day one. Something passed a check once. The person who understood why moved to a different project, or a different company, and nobody who inherited the site carried the same reason forward with it.

The comparison holds at a smaller scale too, one cycle at a time.

Read down either column and the pattern is not about talent. It is about which base each cycle starts from.

Why does none of this show up on a scorecard until later?

In any single quarter, a split team and a single owner can look statistically similar, because compounding is invisible at the scale of one cycle. The gap only becomes visible once enough cycles have passed for one curve to have pulled ahead of the other, which is exactly why organizations chronically underrate the cost of restarting context.

One widely cited estimate, credited to Demandbase's own sales leadership and reported by Demand Gen Report, puts the cost of poor sales-and-marketing misalignment at 10 percent or more of annual revenue (Demand Gen Report, to re-verify). That figure is a single-source estimate, repeated widely because it matches practitioner experience, not a peer-reviewed study, and it should be read as exactly that: directional, not precise. What it points at is real regardless of the exact number. Misalignment between two functions that should share context is not a cost paid once and settled. It is a recurring tax, paid again every cycle the context stays split, and a recurring tax is precisely the kind of thing that compounds in the wrong direction.

What does this argument actually call for?

Not an argument against documentation or process; both still matter, and neither is being dismissed here. It is an argument that documentation and process are a floor, not a substitute for the same mind carrying the reasoning forward. The orchestration playbook lays out the actual mechanics: how one owner sequences visibility, accessibility, and conversion so that each decision serves the next one instead of quietly undoing it.

This piece is the reason that mechanic is worth running at all. Not because it looks tidier in the moment it is chosen, but because held long enough, the gap between a decision that compounds and a decision that resets is the only moat that does not erode on its own.