What does the back office actually cost — books, audit, tax, legal — and when do I upgrade each piece?

You are going to spend money on things no customer will ever see, and the question is not how little you can spend. It is what triggers each step up. The back office upgrades in discrete steps. A step taken early is a fixed cost with no work behind it, and a step taken late is a cleanup with a deadline. We print no prices here, because they change every year and by city, and the shape is what lasts. One thing is different for a security company: your buyer reads your books the way it reads your product, so the audit, the insurance certificate, and the clean cap table are sales documents before they are finance documents.

Books climb a ladder, one trigger at a time.

You keep the books with an outsourced bookkeeper until something makes that impossible, then with a fractional controller or CFO, then with an in-house controller. Each step is triggered by size and complexity, never by preference. The triggers in our portfolio were the first priced round, which turns the cap table and the financials into documents an investor reads; the first enterprise contract, which brings revenue recognition a bookkeeper cannot handle; and the point where the fractional person's share of the week passes half, which is when the fraction costs more than the seat. Take each step when the trigger fires, and not on the calendar.

Audit, tax, and credits are three engagements.

You will be quoted one line for the tax work, and it is three separate engagements with separate providers and separate fees. The audit, which you do not need until a buyer, a lender, or an investor asks for it, and which costs less scheduled off-season than in season. The tax filing, which you need every year from the first. And the tax-credit work, which most founders do not know about.

The research and development credit can be taken against payroll taxes, which makes it real cash in the year rather than a benefit deferred to some future profit, and it is available to most companies at your stage doing engineering work, up to an annual cap, while revenue is still small. Your tax preparer will tell you whether you qualify. A security company at seed is almost all engineering. Claim it from the first year, with a provider who does only that, and treat it as a line in the plan.

Payroll and benefits are a fork.

You choose between a bundled employer organization, which charges a higher fee per employee and gives a small company group health rates it could not otherwise get, and separate payroll processing plus benefits you source yourself, which costs less to process and more to insure. That trade is the whole decision, and it flips at a headcount you will feel before you count it. Start bundled, because the benefits are what a candidate compares, and unbundle when the fee line outgrows the rate advantage.

Legal is priced by the document.

You budget legal by the document, because each one is roughly predictable and the total is not. A valuation opinion, a priced round, a convertible round, the standard document set, a patent search, and a patent application each carry their own budget. Independent lawyers are cheaper than firms for project work and are not a substitute for a firm at a financing, because the lead investor's counsel will be a firm and the negotiation runs at their pace. Ask for a fixed fee per document, and ask for the list of documents you will need in the year before you sign with anyone.

The credits are underclaimed.

You will leave cloud credits on the table because nobody told you the tiers exist. The cloud providers run startup programs tiered by stage, and an affiliation with your investor unlocks the larger tier. Ask your lead which programs it is affiliated with in the week the round closes, because the credits are worth a quarter of infrastructure and they expire.

Two afternoons keep the company operating.

You split payroll from reserves across two institutions before you have a reason to, and you hold what you are not spending this month somewhere off a single bank's balance sheet. In a banking crisis the emergency is access to your cash, not solvency, and the companies that had done this were the ones that made payroll without a phone call. It is an afternoon of work.

The second afternoon is collections. Payment terms are runway, and they are a bigger lever than most expense cuts. Negotiate terms as hard as you negotiate price, invoice on signature, and ask your happiest customer for a prepaid multi-year renewal before you ask an investor for a bridge. It is cheaper capital, it costs no equity, and it sends no signal of distress. The one caution is that a prepaid contract delays your churn signal by the length of the term, so mark it.