Imagine a small fit-out studio led by Mara. On an ordinary afternoon, the founder leaves her desk for lunch. No crisis follows per se, yet a quotation stays unsent, a designer waits to order a material that was already approved, and a client receives a promise that someone will call back. The work has not become harder in the founder’s absence. The decisions have simply stopped moving.
Mara’s team is capable enough to prepare drawings, manage suppliers and speak to clients. Mara built the studio by making the hard calls herself, and she still wants to see the final version of anything that might bear her name. That history makes her useful, but it also makes her the safest place for other people to put uncertainty.
When Mara returns, her inbox gives her a flattering account of the day: everyone needed her. She answers quickly, and two things move. The third turns out to be a question Daniel could have settled using information already in the project file. Mara calls that an easy win. Daniel learns a different lesson: if he waits, he will receive an answer that cannot later be challenged as his own mistake.
Where the delay actually comes from
A busy founder or a crowded inbox does not, by itself, establish a founder bottleneck. The bottleneck appears when someone has what they need to finish the work, including the authority to decide, but still has to wait for the founder’s attention. Three quite different causes can create that dependence.
An approval habit sends a routine choice back to the founder even though the person making it has already agreed on the standard.
An information gap sends it back because the founder knows a client preference, cost or limitation that nobody else can find.
And a reserved decision needs the founder for a legitimate reason: law, professional qualification, contract, safety or a potential loss that the firm has deliberately kept at the founder level. These requests can look much the same in an inbox, although they call for different responses.
In the studio, Daniel could order standard hinges within an agreed project budget, but the supplier offered a different finish. If that substitute meets a written acceptance condition and the purchase falls within his actual authority, Mara’s approval may add nothing. On the other hand, if it changes a client’s agreed specification, Daniel may be right to stop. Either way the inconvenience of waiting is not evidence which one is true.
The team may also lack the training to judge substitutions, or may not have the current drawing or price list needed to make the decision, and taking Mara out of the process would not fix that. The first useful move is therefore to inspect what the decision required, not reprimand the person who escalated it.
A better question than ‘Why do they always ask me?’
Take ten recent decisions that waited for you. For each one, write down what was being decided, what information was available, who held authority, what was at risk and why the wait occurred. A protected professional approval does not become waste simply because it took time. Neither does a request become evidence of incompetence because it arrived in your inbox.
The result may be uncomfortable. You might find that your team needs more training or that a genuine legal or client obligation accounts for the wait. You might also find that your own previous reversals have taught competent people to ask permission for decisions you told them to own.
Nothing in that inventory grants new authority. Instead, it gives the founder a more specific question worth answering: which single low-risk, reversible decision could move without me if its standard, information, limits and legitimate owner were made clear?
You could test this interpretation without withdrawing any approval. For one week, mark each incoming request when it arrives: did it concern a fact, a standard, a permission, or a risk? Then note whether the response required something only you could supply.
Recording requests as they arrive is more revealing than an end-of-week count because it retains the reason for each interruption. A request for a missing part number and a request to approve an irreversible contract change may both take two minutes, but they are not interchangeable.
The employee’s side of the evidence matters too. Ask what they believed would happen if they acted without checking. If the answer is ‘I would be corrected even if the outcome met the brief’, the firm may have created its own permission queue. If the answer is ‘I cannot see the signed specification’, then that points to a problem with document access. If the answer is ‘the contract names Mara’, the queue contains a real obligation.
All three findings must remain possible; your analysis should not force every request into an explanation that suits the founder’s plan to step away.
The aim is not to make the founder unavailable. It is to make availability stop being the operating system for ordinary work.



