You come back from a long weekend and there are eleven messages waiting. None of them are emergencies. Every single one is a request for permission. “Can I tell the client we’re pushing the deadline?” “Should we take this new enquiry on?” “The supplier has changed the price. What do you want to do?”
Your team didn’t fail you. They worked while you were away. But every time a decision appeared, they stopped and waited. That pattern is costing you more than the holiday.
What does it actually mean to delegate a decision?
Delegating a decision is different from delegating a task. When you assign a task, you hand over the work. When you delegate a decision, you hand over the authority to choose the outcome within defined limits. The limits are what make it safe. Without them, your team isn’t really deciding anything. They’re executing under the assumption that you’d approve, which is a different thing entirely.
The confusion between the two is where many delegation attempts stall. “Can you handle the client complaint?” is task delegation. “You can resolve any complaint under £300 without asking me, and flag it in the weekly update” is decision delegation. One takes work off your plate. The other takes the decision off your plate too. Founders who feel they’ve delegated but are still being asked everything have usually only done the first.
Why do founders resist handing over decisions?
The hesitation is understandable. You built this business by making good calls, often quickly. You know what “right” looks like in a way your team doesn’t yet. Delegating a decision feels like giving up the thing that makes the business work. The worry is that someone will get it wrong, and you’ll spend more time fixing it than you would have spent making it yourself.
The problem is that model doesn’t scale. The National Cyber Security Centre, writing about how to govern AI-enabled decisions, notes that delegating without defined escalation paths can turn a fast mistake into a fast breach. The same logic applies to human teams. The absence of a system routes all mistakes through you, with a delay and at a cost to speed.
The FCA fined Starling Bank £29.9 million in 2024 for failings in its financial crime controls. Growth doesn’t dilute accountability. Designing your delegation system is how you put accountability in the right place, without keeping everything on your desk.
Where does delegation break down in practice?
In services businesses with five to fifty staff, delegation tends to break down in the same places. There are no written decision rights, so people don’t know what they’re authorised to decide alone. There is no escalation path, so when something doesn’t fit the pattern, the only option is to wait for the founder. And the boundaries that do exist live in the founder’s head, not on paper.
This creates a loop. Your team learns to ask rather than decide, because asking is safe and guessing wrong is punished. The more they ask, the more you become the bottleneck. The more you become the bottleneck, the harder it is to step back.
The UK Government’s Sourcing Playbook, which sets out how public bodies should delegate work to third parties, identifies the same failure mode: when roles, outcomes, and accountability are not written down, responsibility drifts upwards. Its companion, the Consultancy Playbook, applies the same discipline to professional services delegation. The principle works equally well inside a small firm.
When oversight is diffuse and accountability isn’t documented, errors compound rather than self-correct. The ICO’s £325,000 fine against Uber in 2022 stemmed partly from weak internal controls around who had access to data and on what basis. The specific context is data protection, but the governance failure is identical to what happens in any business where no one has written down who decides what.
Which decisions can safely leave your desk?
The clearest guide for sorting delegable decisions from ones to hold is reversibility. A reversible decision made slightly wrong can be corrected without much damage. An irreversible one, or one with a large potential cost, needs tighter approval. Working through a typical week’s decisions with that lens usually shows that the majority of calls landing on your desk are fully delegable within defined limits.
Safe to delegate with clear authority and a defined limit: routine client communications, supplier queries under a spending threshold, scheduling and resourcing decisions within agreed parameters, and first-draft recommendations that a senior person reviews before they go out.
Hold for now: pricing exceptions outside an agreed range, complaints beyond a defined financial or reputational threshold, any commitment that binds the business in a way that hasn’t been pre-authorised, and decisions that carry legal or regulatory exposure.
The ICO and FCA are both clear that accountability for outcomes stays with the firm, even when the decision is made by a team member or a system. The EU AI Act takes the same risk-based approach for high-stakes automated decisions, requiring human oversight and documentation at every stage. Writing that expectation into your own decision rights framework applies the same logic at your scale, without the compliance overhead.
What does a working delegation system look like?
A delegation system for a small services firm has four components: written decision rights, a named escalation path, periodic checkpoints, and a sample audit. You don’t need a consultant or a framework product to build it. A shared document and a half-day with your management team is enough to get a working version in place.
Written decision rights state, for each role or team, what the person can decide alone, what needs a colleague’s input, and what needs sign-off from you. Be specific about limits. “Handle client enquiries” tells someone what to do. “Approve scope changes up to £2,000 without asking me, and include it in the weekly update” tells them what authority they have.
Escalation path: define a named route for exceptions. “If it doesn’t fit the above limits, speak to [role] before proceeding” is enough. The NCSC’s guidance on governing AI-enabled decisions stresses this point: teams need a specific, named route for anomalies, not a general instruction to escalate. A team that doesn’t know who to call will call you.
Checkpoints: schedule a monthly review of decisions your team made. The goal is to check whether the framework produced the outcomes you’d have chosen, and to catch gaps in the decision rights before they become habits. This is a review, not a retrial.
Sample audit: quarterly, pull ten decisions from across the business and walk through who decided, on what basis, and what happened. The CMA’s guidance on AI oversight recommends this pattern for automated decision chains. It works equally well for human ones. You’re looking for drift, not fault.
Delegation transfers work, not accountability. The system is what keeps both in the right place. If you want help designing it for your firm, Book a conversation is the place to start.
