IPO Readiness Starts with Compensation Governance

None of that is a policy problem. Policy can be written in a quarter, and most companies write a good one. The problem is evidentiary: every award already granted is a historical record somebody has to source, and you cannot write evidence backwards.

Reconstruction is possible while the people who made those decisions are still employed and can still remember why. That is the constraint nobody puts on the readiness plan, because it carries no deadline. It simply gets harder every quarter.

What actually happened

The business listed on the Toronto Stock Exchange in 2015: a consumer packaged goods group of roughly $2–3 billion in revenue, operating globally. Polymath’s founder sat in human capital planning inside Finance through that period, and between 2015 and 2017 he built the executive compensation plans for the listing, the equity dilution and RSU models underneath them, and the share-based compensation governance a listed issuer has to be able to demonstrate.

The dilution model is where the change announced itself. Privately, an award is a decision about one person, complete when the people in the room agree it. Modelled forward for a public register, the same award is a claim about how many shares exist in year three, what expense lands in which quarter, and what a reader is entitled to be told. Nothing about the decision changed. Everything about who had to be able to follow it did.

The forward half of the work was structure: the plan, the executive compensation design, and the authority underneath both — who may initiate a compensation or equity action, who reviews, who approves, at what threshold, with duties separated so the person modelling an award is not the person recording it. He owned that instrument for about eight years afterwards: the HR-related internal controls, the segregation-of-duties design, the General Authority Matrix, and the initiator-reviewer-approver chain across HR and Finance. He was the primary finance lead inside HR and worked with internal and external auditors to build, test and redesign those controls.

The backward half was harder and almost invisible. Awards had been made before any of that existed. They were sound decisions, taken by people who understood exactly why. What they had never needed was a form. So the work was establishing what had already been granted, on whose authority, on which date and on what basis, and making it reconcile. It was possible only because the people who made those decisions were still employed and could still be asked. That dependency carries no deadline, so nothing on the readiness plan ever flags it.

Reconstruction is a conversation with people who remember. Every one of those conversations has an expiry date that appears on no plan. Before a client’s listing plan hardens, Polymath settles the authority and then goes after the evidence behind every award already granted, in that sequence, because only the second half has a human deadline.

Why it happens

A private compensation decision is finished the moment it is made. Its audience is the people in the room, all of whom accept the reasoning because they participated in it. Documenting it adds nothing they need. That is a rational response to the audience, which is why the habit persists in companies that are otherwise well run.

A public compensation decision is finished only when it can be defended without those people. Its audience is an auditor testing a control, an accountant recognizing expense, a committee formed after the fact, a proxy reader, and eventually the successor of everyone involved. Each needs a document. None accepts a memory.

The gap between the two states is evidentiary rather than a matter of policy, and that decides how it can be fixed. Policy looks forward and can be drafted at speed. Evidence looks backward and cannot be drafted at all.

Which matters because grants persist. An award made three years before listing still vests afterwards. It carries expense into your first public statements, dilution into your prospectus, and terms the holder will quote back to you. The decision was private; the consequences are disclosed. So every outstanding grant becomes a record to be sourced — approver, authority, effective date, valuation basis, vesting, forfeiture — and sourced consistently, because the reconciliation is what gets tested, not the single case.

Two things then degrade at once, and only one is on anybody’s checklist. Documents can be located late; that is unpleasant and survivable. Memory cannot be located at all. The executive who set the band, the finance lead who agreed the valuation basis, the director who recalls why an exception was made: each is a single point of failure with no backup, in the years of fastest growth and highest executive turnover. Every quarter you wait, reconstruction gets cheaper to postpone and more expensive to perform, until it stops being possible at all.

A third effect shows up at the first committee meeting. A compensation committee does not begin from a blank page. It begins from the last private decision, because that is the benchmark every executive in the building already uses. If the prior structure cannot be explained, its opening move is either to ratify precedent it cannot defend or to reset it — and a reset is itself a disclosable change made by people in post a matter of weeks.

The strongest objection

The strongest case against this: counsel and the auditors handle compensation disclosure, they do it constantly, and they know the requirements better than management.

Correct, and not worth arguing with. Experienced equity counsel identifies a defective plan document quickly, and a good auditor tells you exactly which control is failing. Neither role is optional.

But both review what exists. Counsel does not decide the relationship between role level and award size, which population is eligible, or what counts as an exception; those are management design decisions with commercial consequences. The auditor tests the control you have, on the population you present, using the records you produce. If the record for a year-one grant is a recollection, the audit discovers that rather than resolving it.

The second objection is stronger: we will qualify for phase-in relief and can build this across our first public years. Phase-in relief is real, material and sensible to use. It governs what you must disclose, not whether your records support what you do disclose. The first year’s figures still originate in decisions taken before relief was a consideration. And relief runs on a clock set by regulation while memory runs on a clock set by your own attrition.

The method

We run six steps. We treat the first three as time-critical in a way the others are not, because they depend on people who may leave.

1. We set the authority for compensation actions before we refine the plan. We name who may initiate, review and approve a compensation or equity action, at what threshold, for which population. The trade-off we accept: it constrains founder and chief executive discretion at the moment that feels least necessary. Skip it and every award ever made needs a named person to vouch for it, which is a governance model with a resignation risk.
2. We reconstruct the existing grant record while your decision-makers are still employed. Approver, authority, effective date, valuation basis, vesting and forfeiture, for every award outstanding — and we carry the dilution treatment through, because your first public statements are assembled from it. The trade-off we accept: unglamorous archaeology that surfaces defects you would rather not have found. Skip it and you do it later without the people, or you disclose an estimate.
3. We separate the modelling from the recording from the approval. We arrange it so the person who builds the award model does not enter it, and neither of them approves it. The trade-off we accept: more steps in a small team where everyone already holds three jobs. Skip it and your first control test finds it in your first public year, the most expensive place to find anything.
4. We write the grant logic as rules rather than as a set of decisions. We set out levels, ranges, eligibility, the relationship between role and award size, and what counts as an exception. The trade-off we accept: rules refuse cases the chief executive wants to accept, and we treat that friction as the product rather than a side effect. Skip it and the committee’s first meeting opens on precedent nobody present can explain.
5. We reconcile the plan administrator, payroll and the share register on a fixed cadence, starting now. We work in individual records, not aggregate values. The trade-off we accept: dull work in a period when nothing appears wrong, competing with everything urgent. Skip it and an auditor finds the differences at the worst point in the calendar, when resolving them includes a delay you cannot afford.
6. We convene the compensation committee before it is legally required to exist. We give it the plan, the rules, the reconstructed record and real decisions to make. The trade-off we accept: you take on oversight before you have to, and some decisions get slower. Skip it and you list with a committee whose first act is to inherit decisions it cannot source.

We hold that sequence deliberately. We take authority first, because it sets what the record has to show. Reconstruction second, because it is the only step with a human deadline. Note what kind of work we are describing. Compensation governance looks like an HR matter and behaves like a Record-to-Report one: its outputs are an expense line, a dilution figure and a disclosure, all assembled from decisions taken years earlier in another function. We build the authority and the evidence structure before the reporting obligation exists, and that is Foundational Architecture. Writing a policy after it exists is not.

Evidence and measures

What has to be settled by when, working backwards from the listing:

Five measures worth tracking over ninety days. None needs a new system:

— Proportion of outstanding grants whose approver, authority, effective date and valuation basis can be produced from a document rather than a person.

— Number of people who are the sole source for any historical compensation decision. Target zero; the first honest answer is usually two or three, and they are senior.

— Difference between the plan administrator’s record, payroll and the share register, counted in individual records rather than dollars.

— Number of compensation actions last quarter where the same person modelled, entered and approved.

— Days taken to answer one question end to end: why did this executive receive this award, in this amount, on this date.

If those five move, the record became defensible. If the only improvement is the policy document, it did not.

How Polymath solves this

Our first pass is an inventory of what cannot be evidenced. We pull every outstanding award and try to source approver, authority, effective date, valuation basis and vesting terms from documents alone. The proportion that can only be answered by asking a person is the finding, and it is usually the first time a chief financial officer sees the size of it. We then write the authority framework against what that exposed rather than against a template, and we run the work as Foundational Architecture.

We do the reconstruction with your equity administrator, your controller and whoever will face the auditors, because they are the people who have to defend it afterwards. We do not hand the authority framework over as a policy document: we configure thresholds, approval routing and separated duties into the HCM and equity platforms you already run, test them on a live compensation action, and then hand the cycle back to the people who own it.

What compounds is the authority structure, and it is reusable in a way a plan document is not. The same thresholds and separated duties govern your merit cycle, off-cycle adjustments, severance and the control evidence the external audit tests each year. A new plan, a new country or an acquired population is added to a structure that already holds, rather than triggering a fresh governance build under time pressure. The work that took a year before the listing is never done twice; it is inherited, and the audit tests it instead of management reconstructing it.

What it costs to do nothing

Reconstruction has a price curve that moves one way. Done now it is internal time. Done in the quiet period it is advisory fees, executive attention when it is least available, and an estimate where a fact should be. The larger exposure is timing. Equity records that will not reconcile surface in underwriter and legal diligence, and they surface late, because that is when anyone looks properly. A few weeks of remediation is trivial against a two-year plan and fatal against a market window.

There is a quieter cost meanwhile. Employees holding awards whose terms cannot be confirmed discount them, so you carry full expense for compensation that is not doing its retention work — and the gap widens exactly as those people become most portable.

The listing date is negotiable and moves all the time. The date your longest-serving compensation decision-maker resigns is not, and nobody tells you it has passed. That is the date we start against.