A six month plan for the CFO role, built around getting the Series C done and putting the right operating discipline in place behind it.
Fixing that gap is the highest leverage thing Finance can do before the Series C.
Employee Intelligence has gone from one customer to 28, we have lost none of them, and average contract value has tripled since the Series B. The financials do not make any of that obvious today.
A chart of accounts that does not separate what we sell, a history assembled from billing, and the largest variable cost uninstrumented.
Get the numbers investor ready before September 8, manage cash and the raise alongside that, then build the finance capability we need after the round.
This is not a normal onboarding. We need to compress the first few weeks around the raise.
Rebuild revenue from contracts and restructure the financial data model. Begin on contract ahead of the formal start if that helps accelerate timing.
The round starts. By then the key definitions are locked and the core diligence exhibits are ready.
Run a tight process toward a close around election day. Finance owns the numbers, the story behind them, and the diligence workload.
Then move into the audit, revenue policy, reporting and planning cadence. The goal is to operate like the company we just raised the money to become.
Kicking off the round will not wait for all financials to be ticked, tied, and rebuilt. That is why the revenue rebuild and the chart of accounts move in front of everything else, including the audit, so that we can tackle the highest leverage items first.
These are all numbers you shared with me. The issue is that an investor cannot see this story clearly in the financials today.
Every customer on the line has bought more, other than the one still working through its pilot rollout. One three thousand person account went from a first deal to over a million dollars inside a single quarter.
Eleven million in late stage pipeline is carried by four sellers with no marketing function behind them, against a mid market motion with a one to two month cycle that is barely explored.
Rates, comparable multiples and investor appetite for application layer AI. We do not control those.
The product story is already strong and the team knows how to tell it. I would not spend much time trying to reinvent it.
This is in our control, it currently understates the business, and we can materially improve it before the raise.
At any revenue multiple, every million of ARR correctly stated is that multiple in enterprise value.
There are three underlying design problems. A faster close or a better controller does not solve them.
The chart of accounts is the data model for the financial statements, without correcting this layer, inputs are slow and cumbersome to get correct, and outputs do not make sense.
Quarterly ARR moves down in a business that has never lost a customer. That is a recognition artefact and a billing lag, not a demand signal, but the statements cannot currently say so.
Model spend is not attributed by customer, environment or agent, and the ceiling can be pushed through. Gross margin therefore cannot be defended line by line.
The September 8 deadline drives the order: get the numbers credible first, manage capital in parallel, then build the longer term finance capability.
The round gets harder if investors do not trust the numbers. So the first job is to get the financial model, revenue history and metric definitions to a place we can stand behind before September 8.
If the underlying structure is wrong, the inputs are more difficult slowing information readiness and the output is funky anyway. Fix the structure once.
The numbers are not telling the story because, amongst other things, the accounts do not distinguish subscription from consumption from delivery.
Better commentary will not fix that. Changing the chart of accounts – the financial data model - underneath will.
Revenue by product line and by revenue type, so subscription, agent consumption and true services never touch again.
COGS broken out with inference spend as its own account, so gross margin is real.
A department dimension, so the research versus go to market split is an easy query
Historically a six-to-twelve-week exercise. Using AI to accelerate mapping and reclassifying can compress the timeline effectively. A full mapping table from old model to new needs to be retained, so history restates cleanly and the change is auditable.
Build one structured contract dataset and use it to drive every revenue view we publish.
Parties, effective date, term, total contract value, billing schedule, ramp, escalators, renewal terms, and any go live or acceptance gate.
Contracted ARR. GAAP revenue under 606. The deferred position. And the reconciliation between them. Today those are three separate exercises that disagree with each other.
We don't have so many contracts that tying the contract summaries back to the schedule is impossible. This will hold up to diligence and subsequent audit.
GAAP revenue matters, but it is likely not the best way to show growth in this business. I would show both, with contracted revenue as the lead metric and a written definition that does not move during the process.
Defined once, published in the room, never adjusted mid process. One number the board, the model and the market all read the same way.
Agent usage counts only where a contracted minimum sits under it. Everything else is shown separately as upside. Fold it in and diligence strips it out, possibly costing credibility.
Total contracted value and remaining performance obligation. Multi year deals with ramps make bookings momentum invisible in a quarterly ARR line, which is an issue I think we have today.
No churn since inception and near universal expansion should produce the best single statistic in the business and possibly in the market. It’s possibly so strong it could look ridiculous. It currently appears nowhere.
Four columns, tied out for every period, available from day one of diligence.
What customers have committed to, from the contract schedule.
What has been invoiced, under the new policy and its effective date.
What 606 allows in the period, including the deferred position.
What has landed in the bank, tied to the balance sheet.
Every awkward question about the historical revenue line resolves once the four columns sit beside each other. The same facts then read as a billing discipline issue that has already been fixed, with a date attached to the fix. This exhibit is also what lets us lead with contracted revenue without anyone suspecting the GAAP number is being avoided.
The whole asset is a graph assembled from public and partnered sources. Lineage means proving, for any field, where it came from and under what right.
Once output drives hire, fire and promote decisions there is an argument the data is a consumer report. Prospect legal teams are already asking.
European residents processed without consent needs a documented legitimate interest test, plus erasure and objection handling against a graph.
Employment related AI is classified high risk, carrying conformity, documentation and human oversight obligations.
New York City bias audit rules, Colorado, and the California automated decisionmaking regulations.
Inferred proficiency is heading toward promotion, succession and a skills based pay foundation. That needs testing.
Collection methods viable in 2017 may not be viable now, and the terms behind historical data may have changed since.
Built with counsel and complete before the room opens. This is the sort of thing that stalls rounds when it surfaces late and could also be a lever inside nine-month enterprise cycles.
These questions are likely to come up anyway. If we address them first, we control the discussion and build credibility.
The eighty two percent reduction in unwanted attrition is at once our strongest asset and our largest diligence risk. It is an extraordinary result for a single intervention across a seventy five thousand person workforce, where attrition is normally driven by pay, management quality and the labour market.
What I would establish in September: the baseline, the denominator, whether it is a cohort or the full population, and what else changed in that period. Same treatment for the twenty nine million in operating savings, separating hard cash from imputed employee time.
If it holds, we lead with it. If it does not hold as stated, we need to know in September rather than during dilligence.
A large share of the balance invoiced in one month immediately ahead of a raise reads as a pull forward. Answer it with a dated policy change and a collections curve.
Twenty eight customers with one anchor logo carrying the case study. Name it, show the cohort expansion behind it, show the mid market pipeline that dilutes it.
A twenty six percent win rate and eleven million in late stage pipeline carried by four sellers and no marketing. Built by segment, that is the use of proceeds argument rather than a gap.
Keep enough cash in the business to get the round closed, and put proper ownership around the decisions that matter.
The immediate job is simple: make sure we have enough liquidity to run the process properly.
Resolve the backstop letter or move to the bank facility. Take more capacity than the immediate need where terms allow, since it can be repaid from round proceeds.
Built in week one, published weekly, and the format does not change. This is the single artefact that answers the predictability question fastest.
The receivables pattern reads as a pull forward unless it is collected and unless the new billing policy is documented with an effective date.
Model spend is part of cost of revenue. Today I’m not sure we can properly attribute it and we do not have a hard cap around it.
Spend broken down by environment, by customer and by agent. A single invoice total cannot tell us whether an account is profitable or whether an internal experiment is quietly expensive.
The limiter is soft today and a handful of people can push through it. That is an uncapped cash exposure with no owner. A real ceiling plus a documented route to raise it takes days, not weeks.
From what I’ve seen gross margin is where this market currently separates infrastructure from resold intelligence. If COGS cannot be defined precisely, covering hosting, inference, data acquisition and delivery, then the margin profile that justifies the multiple cannot be defended. Getting the definition right is worth more in the room than any operating saving it produces.
Once the round closes, the job changes from proving the story to running the business better. This is the work that should make the next raise much easier.
Over months four to six I would put six core pieces in place. Each one should remove a recurring surprise or a recurring debate.
Audit kicked off against a clean finance data model, a documented ASC 606 policy, and numbers we feel good presenting.
Warehouse plus semantic layer, ingesting CRM, product usage and billing. This gives us operational dashboards for faster decisioning.
Monthly close on a published date, a rolling forecast, and quarterly planning that produces owned commitments.
What the go to market build actually buys, modelled by segment, so hiring against the round is a calculation rather than a bet.
The framework for the agents. Jo raised it first and it is the highest leverage margin lever in the business. This might be more ongoing effort.
What we build internally and what stays outsourced, decided deliberately rather than by drift.
The board should spend its time debating a few things that matter a lot, not listening to us present a pack. The format should guide that.
The three or four things we actually need from this board, written as decisions rather than as updates.
A board pack goes out early containing the facts a director needs to form a view, so discussion starts from a shared base of understanding rather than from questions.
Named introductions, specific asks, owners and dates.
Financials, metrics and the key reports available in full, but not read aloud.
Assume every director puts the pack through an AI tool before the meeting. Write it so the summary they get is the summary we want them to have.
Do the highest leverage work first, and be explicit about what I need from the team.
Before the round kicks off · I need contract and billing access early
This is the underlying data model for finance. Within one week of start · I may need outsourced capacity here
From week three – chart of accounts dependant
Before the room opens · needs agreement that definitions are then fixed
Before the room opens · likely needs counsel engaged in September
Second board cycle after start · no dependency
The goal is that neither of you should have to chase me for status. The cadence should make it obvious.
13 week cash flow, collections position, late stage pipeline movement. Same format every week.
What moved, what slipped and why, and what I need from you. Keep it to roughly 200 words instead of adding another meeting.
Set the close date in advance and hold it. If we miss it, that is useful information in itself.
Contracted ARR, retention, the bridge, gross margin including inference. Same definitions every month.
Decision agenda up front, homework out the door, the full pack available behind it.
Every commitment in this plan should have an owner, a date and a status that anyone can see without asking.