ingridleiria Chief of Staff at Series D and Beyond
The role stops advising and starts acting. Years of accumulated organisational knowledge make it possible to decide in place of the chief executive, and that is either the most useful version of the job or the most dangerous one. The difference is entirely whether the authority is written down.
Unwritten proxy authority fails the same way every time. It works for a year, because the person exercising it is careful and well informed. Then a decision goes badly, or a functional leader who was overruled decides to test it, and there is no document saying what was delegated, by whom, within what limits. At that point the organisation concludes it has been operating a shadow executive team, and it is not entirely wrong. The role does not recover from this, and the chief executive's judgement is questioned alongside it.
The second failure is quieter. At this size everything reaching the chief executive has passed through three layers of summarising, each of which improved the news slightly, and the person nearest to them has both the best view of what is actually happening and the strongest incentive not to spoil it.
When to use this, and when not to
Use it at a company past roughly a thousand people, or one with separate business units or geographies that hold their own results, during preparation for a listing, a sale or a major transaction, when the chief executive is asking the role to decide rather than to recommend, or when the diagnosis in chief-of-staff-by-stage returned the proxy archetype.
Do not use it when the role still carries a delivery mandate and advises rather than decides; that is cos-at-series-c. Do not use it at a smaller company where a chief executive simply wants decisions taken off their desk, because delegated authority without organisational depth behind it is resented rather than accepted, and the person exercising it will not have the context to use it well.
Two adjacent cases belong elsewhere. The mechanics of the transaction itself, the bankers, the diligence room and the equity story, sit with finance and the advisers; fundraise-readiness covers the company's side of preparation. Governance mechanics, packs and committee calendars are board-and-investor-management.
What you need before starting
The chief executive's willingness to write the authority down. The whole role rests on this. Missing: draft it yourself, in their voice, narrow, and ask them to correct it. A refusal to write any version of it is important information, and the honest reading is that they want the work done without the authority attached, which is the arrangement that ends badly.
A current readiness assessment, if a transaction or listing is in view. Close speed, control documentation, board composition, cap table hygiene, disclosure discipline. Where a connected finance system is available, take close timing from the last six closes rather than from anyone's recollection, which is reliably optimistic. Missing: build the assessment from the last audit management letter and the last two board packs, which between them name most of the problems already.
The metric definitions in use in every business unit. Missing: collect them before anything else. A company that cannot add up its own numbers across units cannot be run, cannot be diligenced and cannot be listed, and the discovery always comes later than it should.
The statutory and committee calendar for the next eighteen months. Missing: get it from whoever is responsible for corporate governance. That is a company secretary in the United Kingdom and much of the Commonwealth, a general counsel or corporate secretary in the United States, and a governance officer or clerk in a charity or a public body. At this size the calendar has legal consequences and nothing on it should ever be a surprise.
The current succession picture. Who would step into each executive role, internally, tomorrow. Missing: it will be uncomfortable and it is the board's question before it is yours. Assemble the honest version privately with the people leader first.
What currently reaches the chief executive, and from whom. Missing: audit two weeks of their meetings and decisions and categorise by whether a leader should have handled it. That audit is the baseline the role is measured against.
The method
Draft the authority instrument before exercising any authority. Six fields, below. Narrow beats generous: authority can be widened in a line and cannot be narrowed without embarrassment.
Have the chief executive communicate it to the leadership team themselves, in writing. Authority the organisation has not been told about is not authority, it is a claim, and the first time it is used it will be read as one. This step is regularly skipped because it is awkward, and skipping it is the single most common cause of the failure this skill exists to prevent.
Log every decision taken in that name, on the day. Date, decision, which clause it falls under, who was consulted, and when the chief executive was told. The log is what makes the authority auditable, and auditable authority is what makes it survive a bad decision.
Tell the chief executive within a fixed window, always. Same week, in a standing summary, including the routine ones. A decision they hear about first from someone else costs more trust than the decision itself is worth.
Review the instrument every six months and sunset anything unused. Authority granted for one situation quietly persists into others; that drift is how a proxy becomes a shadow executive without anyone deciding it should.
Build the readiness register and fix the boring items first. Close speed, controls documentation, entity hygiene, contract assignment clauses, share register accuracy. The items that stall a transaction are almost never the interesting ones, and every one of them takes longer to fix than a twelve-month window allows.
Force common metric definitions across units, and accept the restatement. One definition per term, applied historically, with the change disclosed rather than absorbed silently. A restatement made voluntarily and explained is a sign of control; the same restatement discovered in diligence is a discount on the price.
Decide what must be common and what may differ, with a rule. Standardise where it materially reduces risk or cost, or where a number has to add up across units. Allow variation where local reality demands it, meaning law, language, customer behaviour or market structure. Reporting convenience is not a reason to standardise, and it is the reason most often given.
Schedule contact with the actual work. Customer calls listened to, a week in a regional office, a support queue read. Put it in the calendar as a recurring commitment, because it is the first thing dropped and it is the only correction available for the distance the role creates.
Run succession properly, including the uncomfortable version. For each executive, who steps in tomorrow, who is ready in two years, and where the honest answer is nobody. A company at this size that cannot name a successor for each executive has a governance problem the board will raise, and it is better raised internally first.
Plan the ending from the beginning. Say to the chief executive what you are aiming at and by when. A role built entirely on proximity is fragile, and everyone in the organisation knows it.
The delegated authority instrument
Six fields, on one page, signed and dated by the chief executive.
Scope. The specific areas, named. Not general management.
Thresholds. Financial limits per decision and in aggregate per quarter, and the reputational test that sits alongside them, usually whether the decision would be uncomfortable if reported externally.
Always excluded. Anything affecting an executive, anything with regulatory or disclosure consequence, anything that changes strategy, anything the board has reserved. These do not move for speed, and speed is exactly the argument that will be made.
Notification. Who is told, and how fast. The chief executive within the week; the affected leader before anyone else.
Escalation. What a leader does when they disagree with a decision made under the instrument. There must be a named route to the chief executive that does not run through you, and it must be published, or the authority will be experienced as unappealable.
Review date. Six months, with the default being lapse rather than continuation.
The instrument fails in three recognisable ways: it is written and never communicated; it is written generously and grows by precedent; or it is used at the edge of the threshold repeatedly, which is the signal that the threshold is wrong and should be renegotiated rather than stretched.
Readiness as a continuous state
Listing or transaction readiness is a state the company is either in or not, sustained over years. Treating it as a project that starts twelve months out is the characteristic failure, because every item takes longer to fix than that window allows.
What being in the state means: financial reporting that closes fast and reliably, month after month; internal controls documented and tested rather than described; a board with appropriate independent composition and functioning committees; disclosure discipline, meaning the company says the same thing internally and externally; equity and share register hygiene; and a management team that can survive diligence and answer consistently on the same questions in different rooms.
Keep the register live, with an owner and a date against every item, and report it to the audit committee on a rhythm. The value of the register is that it converts a diffuse anxiety into a list where the boring items are visibly first.
Decide alone, or escalate
Decide alone: everything inside the written authority, governance operations, the shape and content of board material, the sequencing of the mandate, and what reaches the chief executive.
Escalate: anything outside the written authority, without exception, including when speed argues for it and including when you are certain you know the answer. Anything affecting an executive. Anything with material legal, regulatory or disclosure consequence.
And one judgement that improves with time and should be exercised generously: anything the chief executive would want to hear from you before hearing from someone else. The cost of over-informing is a few minutes; the cost of under-informing is the authority itself.
How the role ends
At this stage the role resolves in one of three ways: into an executive role with a function attached, into the transformation it carried becoming a permanent business, or out of the company. All three are normal and all three are better than the fourth, which is remaining indefinitely as a proxy with no accountability. In that fourth case the role stops developing, the organisation stops respecting it, and the person becomes unemployable at the level their experience should command, because the last five years cannot be described in terms of anything they owned.
Name the intended ending in a conversation with the chief executive, put a date on it, and revisit it annually. The conversation is easier than it sounds and it materially improves how the role is treated in the meantime.
Worked example
Situation. A payments infrastructure business, Tarrant Group, 2,400 people across eight countries, two years after a 260 million round, revenue 310 million growing at around 30 percent. Three business units with their own leadership and results. The board had asked management to be ready to list within roughly two years, without committing to a date. The Chief of Staff had been with the company six years and had been deciding things in the chief executive's name for at least two of them, with nothing written down.
Task. Two outcomes in eighteen months: an authority arrangement that would survive scrutiny, and a company genuinely in the readiness state rather than preparing for it.
Action. The instrument was drafted first and deliberately narrow: capital expenditure to 2 million per item and 6 million a quarter, customer dispute settlements to 500,000, hiring approvals below executive level, and any decision needed inside 48 hours where the chief executive was unreachable, in every case with the exclusions and a published escalation route that did not run through the role. The chief executive sent it to the leadership team himself, which took two drafts because the first read as an announcement of status rather than a working arrangement.
It was used twenty-nine times in the first year. Four were escalated by a leader under the published route, three were upheld and one was reversed, and the reversal was minuted, which did more for the instrument's credibility than the twenty-five uncontested decisions.
The wrong turn ran about four months. Readiness work started with the visible items: board composition, the equity story, a second independent director, adviser selection. Then the auditors' management letter arrived with eleven control observations, three of them repeats from the previous year, and the month-end close was still taking 24 days. The re-sequencing rule was written down as boring items first, on the reasoning that diligence finds control weaknesses and slow closes with certainty and forms a view of board composition with judgement.
The metric problem, found in month five, was the largest single item. The three units defined an active subscriber differently: one counted any account with a transaction in the period, one counted contracted accounts regardless of activity, the third excluded accounts in dispute. The group number had been assembled by addition for three years. One definition was adopted and applied backwards over eleven quarters, reducing reported active subscribers by about 6 percent and group recurring revenue by roughly 4 percent. That restatement went to the audit committee voluntarily, with the history and the reason, and was uncomfortable for one meeting.
Result. At eighteen months the close was at 9 days, nine of the eleven control observations were closed with evidence, and the two remaining sat on the register with dates and owners. The subscriber definition had held for four quarters across all three units, and unit reports reconciled to the group number monthly without manual adjustment for the first time.
The listing did not happen on the original timetable; the window moved and the board chose to wait. The readiness state held anyway, which is the point of treating it as a state, and it made a later approach from a strategic acquirer manageable rather than chaotic. The authority instrument was renewed twice with one narrowing, removing hiring approvals, after a hire made under it in a business unit was read locally as bypassing its own leader. That correction was only visible because every use had been logged.
A second scenario, where it goes differently
The same company profile, but with a sale agreed to run on a four-month timetable rather than an open-ended listing horizon. Readiness could not be built inside that window, so the method inverted.
Instead of a register worked in order of importance, the work became triage against what diligence would certainly find. Three categories: fix now, meaning anything cheap and material such as unsigned contract assignments and share register corrections; disclose early, meaning anything expensive to fix and certain to be found, including the subscriber definition inconsistency, which was put in the data room with the history attached rather than discovered by a buyer's analyst; and accept, meaning anything a buyer would price rather than walk away from, listed for the chief executive so that nothing was a surprise in a negotiation.
The authority instrument changed shape as well. Under a live transaction, disclosure consequences attach to ordinary decisions, so the thresholds were tightened rather than loosened and a clause was added routing anything with disclosure implications to the chief financial officer and legal counsel regardless of value.
What changed the method was the timetable, not the company. Where there is time, build the state and fix the boring items first. Where there is not, stop trying to build it, decide what will be found, and choose deliberately between fixing, disclosing and accepting each item. The failure in a short window is attempting the long-window method and arriving at diligence with half of it done.
Output
DELEGATED AUTHORITY INSTRUMENT Version: [n] Date: [date]
Scope: [named areas]
Thresholds: [per decision, per quarter, plus the reputational test]
Always excluded:[executives, disclosure, strategy, board-reserved matters]
Notification: [who, how fast]
Escalation: [named route to the chief executive, not through this role]
Reviewed by: [date] Default on that date: lapse
Communicated by the chief executive to the leadership team on: [date]
DECISIONS TAKEN IN THE CHIEF EXECUTIVE'S NAME
| Date | Decision | Clause | Consulted | CEO told | Contested? Outcome |
READINESS REGISTER
| Item | Category (close / controls / board / cap table / disclosure) | Owner | Due | State | Found by |
METRIC RECONCILIATION
| Term | Unit A | Unit B | Unit C | Agreed definition | Restated from | Effect |
SUCCESSION
| Role | Ready now | Ready in 2 years | Gap, stated honestly |
Failure modes
Authority exercised before it is written. Recognise it when you can describe what you decide but cannot show a document. Stop deciding until it exists, even for a fortnight.
Written but never announced. Recognise it when a leader is surprised that a decision was yours to make. Have the chief executive send it, and accept the awkwardness of asking twice.
Authority growing by precedent. Recognise it when a decision is justified by a similar one made last quarter rather than by a clause. Fix at the six-month review by narrowing rather than by codifying the drift.
Deciding at the edge of the threshold repeatedly. Recognise it as a pattern of decisions just under a limit. The threshold is wrong; renegotiate it rather than stretching it.
Distance from the work. Recognise it when everything you know arrived as a summary and every number is presented rather than observed. Schedule the contact and protect it, because it will be the first thing cancelled.
Readiness as a project. Recognise it by a start date twelve months before a target listing date. Fix by treating it as a state, with a live register and the boring items first.
Standardising for reporting convenience. Recognise it when the justification for making a unit change is that it makes the group report easier. Reserve standardisation for risk, cost and numbers that must add up.
Edge cases
A leader refuses to accept a decision made under the instrument. Do not escalate it yourself. Point them to the published route and stand aside. The route existing and being used is what makes the authority legitimate.
The chief executive uses the role to avoid a conversation they should have. Recognise it when a message about an executive's performance is being delivered by proxy. Decline it, once, plainly. This is the boundary that most defines whether the role is respected.
The founder chief executive is the problem. At this size a founder may be doing a job they neither want nor suit, and the board is thinking about it before anyone says it. The role cannot lead this and must not brief the board against its own chief executive. What it can do is say the observation to the chief executive first, honestly, and be predictable about what it will say if the board asks directly.
A transaction collapses late. The readiness state is retained, the register stays live, and the people who worked eighteen-hour weeks for four months are told what happens next within days. The most damaging period is the fortnight of silence afterwards.
Authority in a jurisdiction where it has no legal standing. Delegated authority is an internal arrangement, not a legal one, and some decisions require a director of a specific entity. Keep a list of what must be signed by whom, per entity, and never rely on the instrument to cover it.
Quality bar
- The delegated authority is written, signed, communicated by the chief executive, and reviewed every six months with lapse as the default.
- Every decision made in that name is logged the day it is made and reported to the chief executive within the week.
- A published escalation route exists that does not run through this role, and it has been used at least once without consequence for the person using it.
- Readiness is treated as a continuous state, with a live register where the unglamorous items are first.
- Metric definitions are common across units and reconcile without manual adjustment.
- Contact with the actual work is in the calendar, not in the intention.
- A succession picture exists for every executive role, including the honest gaps.
- The end of the role is planned and dated rather than allowed to drift.
Adapting this to your context
The defaults come from a late-stage private company of about two thousand people preparing for a listing. Most of it survives translation; the vocabulary does not.
- "Series D" as a label. It means separate business units or geographies holding their own results, decisions routinely made in the chief executive's name, and outside obligations on fixed dates. A group of trading subsidiaries, an NGO with country offices or a family holding company is here whatever it has raised.
- The readiness state. Written for a listing or a sale. The same register serves a regulatory inspection, an accreditation review or a generational handover, unglamorous items first.
- The authority thresholds. Two million per item and six million a quarter came from one company's balance sheet. Set yours as a fraction of monthly cash movement rather than by copying a figure, and keep the reputational test beside it.
- Delegated authority and the law. The instrument is internal. In many jurisdictions certain acts must be signed by a named director, trustee or officer of a specific entity, so keep the list of what must be signed by whom before relying on a threshold.
- What not to change. The authority is written, communicated by the chief executive, and every decision under it is logged the day it is made.
Related skills
cos-at-series-c precedes this and produces the mandate discipline that late-stage programmes still run on. chief-of-staff-by-stage re-diagnoses when a transaction, a listing or a large reduction changes the company underneath the role.
board-and-investor-management and board-deck carry the governance and committee mechanics referenced throughout. fundraise-readiness covers the company's preparation for a raise or a sale process. decision-memo is the format for everything escalated outside the instrument, and its decision log is the institutional memory this stage depends on. crisis-and-incident-comms is what a transaction collapse or a disclosure failure turns into within a day. program-management runs the readiness register when it needs a team rather than a person.