The Four Approval Gates You're Still Holding

Your team can produce. That's not the constraint and it hasn't been for a while.

The constraint is that nothing goes out until you've looked at it, and you look at things in the gaps between client work, which means your visibility runs at exactly the speed of your inbox and slows down in precisely the months you'd most want it to speed up.

You know this. What you probably haven't done is separate the review into the parts that genuinely need you from the parts that don't — because from the inside it feels like one thing called approval, and one thing can't be partially delegated.

It isn't one thing. It's four, and they have different answers.

Gate one: is this true?

Somebody has to check that the claim is accurate, the number is right, the client detail is anonymised properly, and nothing overstates what you actually do.

This gate should stay, and it should stay with someone senior — but notice that it isn't a taste question. It's a factual check, it's fast, and it's the same check every time. Which means it can be written down as three or four criteria that anybody senior enough can apply, and it means it doesn't have to be you specifically.

The tell that you're conflating this with the others: your feedback on drafts contains both that number's wrong and I'd have said this differently. Those are two gates in one pass, and fusing them is what makes review expensive.

Gate two: does this sound like us?

This is the one founders hold longest and defend hardest, and the defence is usually genuine. Voice is real, it's a differentiator, and a business whose content sounds like everybody else's has lost something that took years to build.

It's also the gate that's cheapest to hand over, for a reason that's slightly deflating: voice is mostly imitable from examples. Not from a style guide — nobody has ever successfully written a style guide that transferred a voice — but from twenty pieces you've approved, read together, with three or four rules attached about what you never do.

Voice transfers through examples and rules, not through review. Every round of "I'd have said it this way" is a lesson that doesn't accumulate.

Which is the real cost of holding this gate. Every correction you make is knowledge that stays in your head, so the tenth draft needs as much correction as the first, and you've built a review loop that runs forever without converging.

Gate three: is this good enough?

Quality. Structure, argument, whether it's worth publishing at all.

Here's the uncomfortable thing about this gate: it's the one where your involvement most reliably slows the business down for the least return. Not because your judgment isn't better — it usually is — but because the marginal difference between good and very good is much smaller than the difference between published this week and published next month, and consistency is what the market is actually buying.

Consistency feels boring to creators but magnetic to consumers. We get tired of our own message long before the world ever hears it enough to remember it.

The rule that resolves this in practice: set a floor, not a ceiling. Anything above the floor ships. Your energy goes into raising the floor over time — through examples, through the occasional teaching edit — rather than into lifting individual pieces over it.

Gate four: should this go out now?

Timing and context. Is there a reason this particular thing shouldn't publish this week — a client situation, something happening in the market, a conflict with something else going out.

This gate stays with you, and it should. It's the only one of the four that requires information the rest of the business genuinely doesn't have.

It's also nearly free. It's a scan of a schedule, not a review of a draft — thirty seconds a week rather than an hour, and it can happen against a queue rather than against individual pieces.

The order matters more than the decision

You can't hand over all three at once. Try it and you'll have one bad week, pull everything back, and be more entrenched than before.

Hand them over in this order.

Quality first. Counterintuitive, and it's right. Set the floor explicitly — three or four criteria, written down, derived from pieces you've rejected and why. Then let anything clearing the floor ship without you. The first month will produce two or three pieces you'd have made better. That's the price, it's smaller than you think, and it buys you the throughput to find out whether the floor is set correctly.

Voice second, and only once there's a corpus. Twenty approved pieces plus a short list of rules — never these words, always this structure, this is how we refer to clients. Then a defined period where drafts come to you after publication rather than before, so you're teaching from live work instead of gating it. That inversion is the actual mechanism; a post-hoc note lands as coaching and a pre-publication note lands as a veto.

Accuracy last, and possibly never. It's genuinely the highest-stakes gate and it's also the one that most obviously can be written as criteria. Hand it to a specific senior person, not to the team generally, with a defined escalation for anything they're unsure about.

Timing stays. Keep it. It costs you thirty seconds and it's the only one that needs your context.

The conversation to have with your team

None of this works if it's announced as a process change, because from the other side it reads as either a loss of your attention or a new set of hoops.

Say the actual thing. The bottleneck is you, it's costing the business visibility in exactly the months that matter, and you're going to fix it by writing down what you've been checking for rather than by caring less. That framing is true and it lands completely differently.

Then be specific about what changes and when. Which gate is moving this month, what the criteria are, and what happens when somebody's unsure — because the failure mode of delegated approval isn't people publishing bad work, it's people freezing on the edge cases and quietly routing everything back to you anyway.

And name what you're keeping. Holding the timing gate and saying so is a much stronger signal than pretending you've handed over everything, and it means the one thing you do need to see actually reaches you.

What actually stops this

Two things, and neither is about trust.

The first is that there's nowhere for the corpus to live. Handing over voice requires twenty examples somebody can read, and if approved work is scattered across four platforms with no record of what was approved and why, the handover has no substrate. This is a small fix and it's a prerequisite.

The second is that the queue is empty. A business with no material a month out can't afford to let anything ship without a look, because each piece is the only piece. Depth in the queue is what makes delegated approval survivable — it means a bad piece is one of twelve rather than the week's entire output.

That's the connection to the Outreach pillar generally. The gates and the queue are the same problem viewed from two ends: if every piece starts from zero, publishing is a fresh act of creation every time, and a fresh act of creation is exactly the thing a founder can't help reviewing.

How to tell it's working

Two signals, and neither of them is that you feel more relaxed.

The first is that output holds through a busy month. That's the whole point, and it's the only measure that matters — a business whose publishing rate is flat across its heaviest delivery quarter has fixed something structural. If output still drops when you get busy, a gate is still with you regardless of what the process says.

The second is that the corrections converge. Six months in, the notes you'd make on a draft should be fewer and smaller than they were, because the examples and the rules are doing work that the review used to do. If you're making the same correction in month six that you made in month one, the knowledge is still in your head and the loop hasn't started accumulating.

If neither signal appears, the likely cause isn't the team. It's that the criteria were never actually written down — that the floor lives in your judgment and the handover was permission rather than specification.

What you're actually buying

Not time, though you'll get some back.

You're buying the property that visibility continues during the months when you're least available — which are, reliably, the months right after you land good work. That's the oscillation that costs founder-led businesses most: the busy quarter goes dark, the drought arrives ninety days later, and it gets read as a market condition rather than as the predictable output of a review bottleneck.

Your visibility is throttled by your inbox. That isn't a discipline problem or a capacity problem — it's a permissions problem you can solve in a fortnight.

Nothing goes out without you is a sentence about architecture, not about standards. Founders who fix it don't lower their standards; they move them from a person into a set of criteria, which is the only form standards can take if they're going to survive a busy quarter.