Should you pay for an audit before you raise? Only if its cost is smaller than the discount it removes. That discount is a number you can estimate this week, from your ticket size, your counterparty, and how much of your value sits in figures a stranger cannot yet check.
Most founders file the audit under compliance. Something the regulator wants, or the accountant nags about, filed next to the annual return. That framing loses money. An audit is not a compliance chore. It is a purchase of someone else’s verification, priced in basis points you can add up.
So price it. The investor across the table applies a haircut to unaudited numbers because checking them is expensive and the pool you sit in contains people who lie. African founders already struggle to attract investors, local or international,8 and a number nobody can check widens that gap. A credible audit moves the checking cost off the investor’s desk. What comes back to you is a lower cost of capital, a faster close, or both. Every one of those is measurable before you commit.
Why these three models
The audit decision is a pricing decision wearing a governance costume. Three lenses carry it, and they disagree in useful places.
The first explains why the discount exists at all: when quality cannot be verified, the market pools you with the worst plausible version of yourself, and only verified disclosure separates you out. The second turns that into arithmetic: how large the discount is, and the exact conditions under which the audit fee is worth paying. The third asks the question the first two skip: whose signature you are buying, because a certificate is only worth the trust the certifier already carries in your investor’s network.
The model-thinking method would normally add a behavioral layer and a governance layer. Both fold in here rather than earning their own cards. The behavioral layer, why founders treat a priced asset as a chore, sits in the closing section. The governance layer, the regulator and the development financier as certifiers, sits inside the third model, because in African deals the licence-holder and the DFI are nodes in the same trust network as the auditor.
Two of the three lenses live in the same family, equilibrium pricing. That is honest, not lazy. A computable discount is fundamentally a pricing story, so the span here runs across two outcome types, equilibrium and complex, rather than three. The third lens is what keeps the set from collapsing into one voice.
1. The unraveling: why silence gets priced as a lie
Start with the mechanism that creates the discount.
An investor cannot see inside your numbers cheaply. So they reason about the pool. If good founders and exaggerating founders both show unaudited figures, the investor cannot tell them apart, and prices everyone at the pool average. That average is dragged down by the exaggerators. You, the honest one, subsidise them.
Verified disclosure breaks the pool. When one type can prove quality and the other cannot afford to fake the proof, disclosure unravels: the provable separate themselves, and whoever stays silent is priced as if they had something to hide. The audit is that costly, hard-to-fake proof. An audit makes your numbers expensive to doubt, and cheap doubt is the whole reason your terms are worse.
The evidence is direct. Studying privately held firms where audits are not mandated, the information environment most like an African startup’s, audited firms carry a significantly lower cost of debt, and lenders put more weight on audited figures when setting the rate.1 That is the mechanism in one sentence: the audit changes how heavily a lender leans on your figures when pricing the deal.
Verifiable disclosure and unraveling. Assumes proof is costly enough that a liar would not buy it, and that investors reason about the pool, not just about you. Fits because unaudited African numbers are exactly the unverifiable claim the model prices down. Breaks when nobody can tell a real audit from a bought opinion, so the proof stops separating types. Counteracts the cohort haircut that one exaggerator imposes on everyone who looks like them. May reinforce a two-tier market where only those who can afford proof get read.
2. The comparative statics: how big, and when it clears
Now make it a calculation.
The discount has a floor you can look up and a ceiling you have to estimate. The floor is the cost-of-capital effect. On the debt side, one instrumental-variable study of private firms puts the average saving from an audit at 0.47 percentage points on the cost of debt.2 On the equity side, the presence of a top-tier auditor is associated with a cost of equity roughly 55 basis points lower for US firms and 23 lower outside the US.3 Those are the mechanical, provable parts of the discount.
Run the debt case first, because it is the cleanest. Take a working-capital facility of USD 200,000. A 0.47-point rate saving is about USD 940 a year, near USD 2,800 across a three-year facility. A statutory audit for a small company in Kenya starts around Ksh 50,000, a few hundred dollars, and a venture-grade audit with real complexity runs several times that.4 Even at the top of that range, the facility-life saving covers the fee. On anything above roughly USD 150,000 of debt, the audit clears on the interest line alone, before you count the doors it opens.
The equity case is larger and less tidy, because the real discount is not the 23 basis points. It is the haircut the investor puts on the numbers themselves. If they privately treat your reported revenue as, say, a quarter softer than you claim until it is checked, that haircut moves your valuation by far more than the fee. You cannot look that figure up. You can estimate it: take the revenue an investor cannot independently verify, apply the discount you have watched investors apply to founders like you, and multiply by the valuation multiple in play. That product, minus the audit cost, is your return. If it is positive, buy the audit. If it is not, do not.
The static that flips the whole calculation is your counterparty. Whether the audit pays turns on who reads it. The same private-firm study found that a well-known auditing name added no extra cost-of-debt benefit, and that in one sector, where a single lender held 74 percent of the market and a century of lending history, the audit barely moved the rate at all.5, 2 A counterparty who already verifies you cheaply, a bank that has watched your account for five years, a distributor who sees your stock turn, pays you almost nothing for an audit, because you are removing a checking cost they were not paying. Do not buy verification for a counterparty who is not charging you for doubt.
Comparative statics. Assumes the discount and the fee move independently, so you can compare them at each stage. Fits because both sides are estimable: fee from a rate card, discount from the haircut you have seen applied. Breaks when the haircut is unknowable, early stage with no revenue to verify, so there is nothing for the audit to certify. Counteracts the reflex to audit on principle. May reinforce under-auditing by founders who cannot yet estimate their own haircut and so assume it is zero.
3. The network: whose signature you are actually buying
An audit is a signature, and signatures carry only the trust the signer already holds in the reader’s network.
This is where the two equilibrium lenses part company. The cost-of-equity benefit of a top-tier auditor is not universal. It is stronger in countries with better investor-protection institutions and weaker outside the US, precisely because part of what a global audit name sells is implicit insurance that thin-institution markets do not price.3, 6 Read together with the debt study, where the auditor’s name added nothing, the lesson is precise: a certifier is worth exactly the number of trusted edges connecting it to the counterparty you need.
Position that against your actual reader. A development financier will not disburse against unaudited accounts, and its own rulebook names the requirement: annual statements from a competent independent auditor, full stop.7 For that reader, an auditor already inside the DFI’s recognised set is a short, trusted path. For a local relationship bank, that same globally-networked firm is a longer, more expensive path that buys you little. The central node is the name your specific investor already trusts, one edge away, which is rarely the biggest name on the letterhead.
So the lever is a matching problem, not a status problem. Name your next three counterparties. Pick the least expensive auditor whose signature reaches all three inside the trust network they already use. Then get, in writing, the specific figures that reader will verify against, because an audit read by nobody is a cost with no discount attached.
Network centrality. Assumes trust flows along existing edges, so a certifier’s value depends on its distance from your reader, not its size. Fits because DFIs, regulators and foreign funds trust different certifier sets. Breaks when your reader trusts no third party and insists on checking directly, which collapses the audit’s value to zero. Counteracts overpaying for a famous name a local counterparty does not price. May reinforce incumbents if only network-central firms are ever believed, freezing out capable local auditors.
GEER: the cheapest moves toward a computable discount
Order the levers by cost and reversibility. Cheapest and most reversible first.
Estimate your haircut before you spend anything. Ask two investors who have passed on peers what discount they mentally apply to unverified revenue in your category. That number is the whole decision, and the conversation is free.
Fix what an audit would flag anyway, at near-zero cost. Reconcile your bank statements to your revenue claim. Get IP assignments and director loans documented. Half of a small audit’s friction is disorganisation you can remove yourself.
Buy the smallest credible verification that clears. A review or agreed-upon-procedures engagement costs less than a full audit and, for a debt counterparty, often carries most of the discount. Reserve the full statutory audit for the reader who requires it by rule.
Match the signer to the reader, last, because it costs the most to change. Only once you know your three counterparties do you choose whose signature reaches all three.
RADAR: what to decide before this financial year closes
An audit is dated to a financial year. Miss the year-end and you wait twelve months, so the calendar sets the deadlines, not you.
Do now, T+3 days. Estimate the haircut and multiply it out. If the computed discount beats the fee for the counterparty you will actually approach, commit to auditing this financial year. This is reversible in principle and dominant across every scenario where you hold verifiable numbers.
Hedge, T+14. If you are unsure you will raise, engage a review rather than a full audit, and keep clean monthly reconciliations. Cheap insurance: it preserves the option to upgrade to a full audit at year-end without having lost the year.
Defer and trigger, T+28. If you have no revenue worth verifying yet, do not audit. Pre-commit the trigger instead: the first month you cross the revenue level at which an investor’s haircut on your numbers exceeds the audit fee, you start the engagement. Write that number down now, because the irreversible cost is a missed year-end, not the fee.
CHAIN: what usually happens after a founder audits
Match the reference class on structure, not surface. The right comparison runs to other founders who bought third-party verification into a market where checking was expensive: exporters getting product certified for a buyer who cannot inspect the factory, borrowers commissioning a valuation before a lender releases a facility. In those classes, the base rate is clear. Verification moves terms when the counterparty was paying to check, and moves nothing when they were not.
Modify for your present state. A first-time founder with no track record has a larger haircut to remove, so a larger discount to capture, than a repeat founder investors already read cheaply. Thin local institutions cut both ways: they raise the value of a globally-recognised signature to a foreign fund and lower it to a local bank.
Subtract the counterfactual before you credit the audit. Some of the better terms you would have won anyway, because your revenue grew or a warm introduction pre-verified you. The audit’s true contribution is the discount that remains after you strip out what other proof already delivered. If a long banking relationship was going to get you the rate regardless, the audit added little, and the honest calculation says keep the fee.
Matrix-break flag. If verification becomes near-free, real-time payment data an investor can pull directly, bank feeds, mobile-money ledgers open to a lender’s API, the audit’s separating power falls. When the investor can check you continuously for nothing, the annual audit stops being the cheapest proof and becomes a lagging one. Watch for the counterparty who asks for a data connection instead of a signed opinion.
Where this calculation goes blind
The arithmetic assumes you can estimate your own haircut. Many founders cannot, because the discount is invisible from your seat. You never see the deals that died in silent diligence, or the term that was quietly a notch worse because your numbers could not be checked. The cost of being unverified is paid in outcomes you never witness, which is exactly why it gets filed as a chore rather than priced as a loss.
The model also cannot see a corrupted certifier. If audits in your market can be bought, the signature stops separating honest from dishonest, and the whole discount collapses for everyone. That is a market-quality problem no single founder can solve by spending more.
Neither blindness changes the move. Name the three counterparties you will approach in the next twelve months. Ask two of them what they discount unverified numbers by. If that discount, times the value sitting in figures they cannot check, beats the fee, book the audit before this year-end. If it does not, keep the money and put the year-end in your calendar as the trigger to run the number again.
Sources and notes
- Michael Minnis, “The Value of Financial Statement Verification in Debt Financing: Evidence from Private U.S. Firms,” Journal of Accounting Research 49(2), 2011. Abstract and finding that audited private firms carry a significantly lower cost of debt and that lenders weight audited information more heavily, via IDEAS/RePEc: https://ideas.repec.org/a/bla/joares/v49y2011i2p457-506.html
- “Do audited firms have a lower cost of debt?”, International Journal of Disclosure and Governance, 2021. Instrumental-variable estimate that audited firms save on average 0.47 percentage points on the cost of debt versus unaudited firms: https://link.springer.com/article/10.1057/s41310-021-00133-1
- Sadok El Ghoul, Omrane Guedhami and Jeffrey Pittman, “Cross-country evidence on the importance of Big Four auditors to equity pricing: The mediating role of legal institutions,” Accounting, Organizations and Society 54, 2016. Cost of equity falls on average 55 basis points (US) and 23 (non-US) in the presence of a top-tier auditor, via EconPapers/RePEc: https://econpapers.repec.org/article/eeeaosoci/v_3a54_3ay_3a2016_3ai_3ac_3ap_3a60-81.htm
- Audit fee floors under the Kenyan Accountants (Remuneration) Order, 2021 (ICPAK), summarised with the schedule: minimum audit fee for a small company (revenue under Ksh 5 million) of Ksh 50,000, and an hourly partner rate of Ksh 14,500: https://alphacap.co.ke/audit-accounting-fees-kenya/
- Same study as note 2: a well-known (big-name) auditor added no additional cost-of-debt benefit, and in agriculture, where one lender held a 74% market share and a century of lending history, audit barely reduced the rate, indicating that a counterparty with cheap alternative information substitutes for the audit: https://link.springer.com/article/10.1057/s41310-021-00133-1
- Same study as note 3: the equity-pricing benefit of a top-tier auditor is stronger in countries with better investor-protection institutions and weaker outside the US, where the implicit insurance auditors provide is priced lower: https://econpapers.repec.org/article/eeeaosoci/v_3a54_3ay_3a2016_3ai_3ac_3ap_3a60-81.htm
- African Development Bank / African Development Fund, “Guidelines for Financial Reporting and Auditing of Projects.” Annual financial statements of all Bank-funded projects must be submitted for auditing by a competent independent auditing firm to certify their reliability: https://www.afdb.org/fileadmin/uploads/afdb/Documents/Procurement/Project-related-Procurement/GUIDELINES%20F0R%20FINANCIAL%20REPORTING%20AND%20AUDITING%20OF%20PROJECTS.pdf
- Practitioner context on African SME financing, Ronalds LLP (Kenya): SMEs “find it difficult to attract investors, both local and international,” with limited access to funding, one reason credible reporting is treated as a route to capital: https://ronalds.co.ke/how-adoption-of-ifrs-can-unlock-your-sme-funding/
Note on figures: the cost-of-capital effects in notes 2 and 3 are averages from cross-firm studies and are used here as the estimable floor of the discount, not a promised outcome for any single company. The working-capital and revenue-haircut arithmetic in section 2 is illustrative, built to be re-run with your own ticket size, counterparty and haircut estimate. Currency conversions are approximate.