Customer due diligence for buyers.

The book of business is the asset every disciplined buyer is actually purchasing. The agency entity, the lease, the office furniture, the AMS license — none of these carry meaningful enterprise value relative to the book. The renewable commission stream from the book of business is the cash flow the multiple is applied against; the question that determines whether the buyer pays a defensible price is not "what is the book today" but "what will the book be twelve, twenty-four, and thirty-six months after close." Customer due diligence is the forensic work that answers that question.

The discipline matters because the seller's retention number — the one in the management presentation, the one in the LOI conditional anchor — is rarely what the buyer's diligence team will eventually verify it to be. Not because sellers are dishonest, but because the standard agency reporting tools produce headline numbers that systematically overstate the book quality the buyer is actually buying. A seller reporting "94% retention" can be reporting an accurate number that nevertheless conceals churn the buyer will absorb post-close. The first DD move is to separate the headline from the operational reality.

This Pillar is the map for that work. It pairs especially closely with financial due diligence (where the retention number feeds pro-forma modeling and the earnout structure) and carrier due diligence (where customer concentration intersects carrier appointment risk). If you take one thing away, let it be the posture: the book is what you're buying; verify the book first, model the financials second, structure the deal third.

§ 01 · The book is what you're buyingVerify the asset first.

The independent buyer's most consistent strategic error in customer DD is treating the seller's retention number as a starting point for negotiation rather than as a starting point for verification. The retention number in the management presentation is a marketing artifact — produced for the listing, calibrated to defend the LOI multiple, presented in the framing most favorable to the asset's defensibility. The verification work is the forensic counter-frame that produces the buyer-relevant number.

The customer-DD architecture has three layers, each answering a different question. Layer one: book quality. The output is a Normalized, Defensible, Sustainable retention number — distinguished from the headline by the methodological discipline that separates policy retention from premium retention, accounts-rate from revenue-rate, and includes-new-business from same-store renewal. Layer two: book defensibility. The output is a legal-defensibility map — whether the agency owns the book contractually, whether producers hold renewal rights, whether non-compete and non-piracy covenants will enforce against a producer who leaves twelve months post-close. Layer three: verification and protection. The output is the deal-structure adjustment — earnout triggers anchored to the book's residual risk, holdback escrowed against the retention assumption, non-piracy covenant drafted with enforcement mechanics the buyer's counsel signed off on.

Skipping any layer leaves unbounded risk in the Purchase Agreement. The financial-modeling exercise produces an EBITDA number; without book defensibility verification, that number could evaporate on Day 1 of post-close integration if a producer walks with their book. The structural protection only works if the buyer has verified the legal mechanics that make the protection enforceable. The three layers are sequential, and the sequence matters.

§ 02 · Policy retention vs premium retentionThe diagnostic split.

The single most common headline-vs-reality gap in agency retention reporting is the conflation of policy retention with premium retention. A seller reporting "94% retention" can mean either of two very different things, and the lower number is the buyer-relevant one.

Policy retention is the percentage of policies in force at the start of the measurement period that were still in force at the end of the period — a per-count measurement that treats every policy as equal. A book with 1,000 policies that loses 60 small personal-lines policies and retains every commercial account would report 94% policy retention. The number is accurate; it also obscures whether the lost policies represented meaningful revenue or whether the retained policies represent the dollars that anchor the deal.

Premium retention is the percentage of premium dollars at the start of the period that renewed at the end of the period — a dollar-weighted measurement that captures the revenue-relevant signal. The same book with the 60 lost personal-lines policies might report 94% policy retention but 91% premium retention if the lost policies skewed small. The four-point gap looks small until the math compounds: a 3% per-year premium-retention gap relative to policy-retention is a 9% gap in the book's revenue base over three years. The buyer's pro-forma EBITDA modeled against the policy-retention number will systematically overstate the cash flow over the integration window.

The diagnostic move is mechanical. The buyer's DD team requests both numbers from the seller's AMS — usually a three-year history of policies-in-force at start and end of each twelve-month period, plus the corresponding premium-dollar totals. Compute both ratios. If the seller's reported number doesn't match the buyer's recomputed number, the conversation that follows is informative regardless of the cause — it's either a definitional difference (in which case the premium number is the conversation), or it's a data-quality issue (in which case the data quality is the conversation).

The gap is also the input to the earnout sizing in §07. A book with a 4-point policy-vs-premium gap carries different residual risk than a book where the two numbers align. The structural protection should reflect the gap explicitly — the earnout's retention trigger anchored to the premium-retention number, not the policy-retention number, with the threshold set above the historical premium-retention baseline by enough margin to give the seller skin in the game but below the headline policy number by enough margin that the buyer is not paying full price for retention the diligence work could not verify.

§ 03 · The Leaky BucketHow new business hides churn.

The Leaky Bucket Syndrome is the structural pattern that lets a 92% headline retention number conceal a book that is operationally leaking. The mechanism is simple. The agency's gross premium grows in two ways: existing accounts renew (retention) and new accounts come in (new business). If the agency's retention rate is 88% but new business adds 4% of premium per year, the headline "net retention" number reads as 92%. The bucket is leaking 12 points and getting refilled with 4 — the headline says 92% but the book quality is degrading by 8 points per year of underlying customer relationships.

The pattern matters to the buyer because new business does not transfer cleanly with the close. A producer's recent new-business wins are typically the most relationship-dependent revenue in the book; they have not yet renewed at the agency, the relationship is owned by the producer's recent outreach, and they are the most likely cohort to follow the producer if the producer leaves post-close. The buyer who pays for the agency's headline retention is paying for renewal stickiness the new-business component does not yet possess.

The diagnostic is to separate same-store renewal retention from net retention. Same-store renewal retention measures only the policies that were in force at the start of the period AND still represented the same client relationship — excluding new business added during the period, excluding lost accounts replaced by new ones from the same client. It is the cleanest measure of book stickiness, and it is the number that should anchor the buyer's pro-forma. The seller will not produce it on the first request — sellers don't typically run their AMS reports this way — but the buyer's diligence team can recompute it from the policy-level AMS export.

The gap between same-store renewal and the headline net retention number is the most informative single ratio in the customer-DD work. A 1-2 point gap is healthy book maintenance; a 4-6 point gap is a leaking bucket; a 7+ point gap is a book whose operational quality is being concealed by new-business velocity that may not survive a change of control. The buyer prices the gap directly: into the earnout trigger, into the holdback amount, or into a multiple compression if the gap is large enough that structural protection cannot fully cover it.

§ 04 · The 15% Rule + the Existential DozenConcentration thresholds.

Customer concentration kills more agency acquisitions than any other single diligence finding. The mechanism is direct: if a single client represents 18% of agency revenue and that client elects not to renew with the new owner six months post-close, the buyer absorbs an 18% revenue hit on the agency's largest source — and the EBITDA hit is disproportionately larger because the agency's fixed cost structure does not contract proportionally. A book that loses its top three accounts loses something like 40% of its EBITDA contribution, even if the policy-count retention number reads as 93%.

The two operational thresholds the disciplined buyer enforces:

The 15% Rule. No single client should represent more than 15% of agency revenue. Books that exceed the threshold carry concentration risk that has to be priced into the deal — either through a multiple compression, through a retention-anchored earnout sized against the concentrated account specifically, or through a structural carve-out that excludes the over-concentrated account from the purchase consideration entirely. The 15% number is operational, not regulatory; it reflects the threshold at which a single client's departure can move the enterprise EBITDA by more than the deal's normal absorption capacity.

The Existential Dozen. The top 12 accounts by revenue. Mapped by carrier (which carrier appointments do these accounts depend on), by line of business (are they all commercial property, or diversified), by relationship owner (which producer or staff member is the day-to-day contact), and by renewal cycle (when does each policy renew). The Dozen exists because, across most independent agencies, the top 12 accounts represent 35% to 55% of revenue. The book's first-year defensibility is mostly the Existential Dozen's defensibility. The diagnostic playbook for the Dozen is the most operationally informative customer-DD artifact the buyer can produce, and it should be built before the LOI is countersigned, not after.

The structural implication of both thresholds: when either is violated, the deal needs a structural protection layer that the standard earnout sizing does not provide. A concentrated-account-specific retention covenant ("the earnout reduces by X percent if Account Y does not renew at terms equivalent to the trailing-twelve-month relationship") protects the buyer at the account level. A holdback escrowed specifically against the Dozen's retention protects the buyer at the cohort level. Both mechanisms are documented further in financial due diligence-K earnouts/holdbacks and the legal architecture indemnification — the customer-DD findings are the input to the deal-structure mechanics those clusters specify.

§ 05 · Producer-owned booksThe legal-DD overlay.

The most catastrophic single finding in customer DD is not concentration or retention — it is producer ownership. If the agency's producer agreements grant the producers (not the agency entity) contractual ownership of the book, the asset the buyer is purchasing is structurally hollow. The financial model is anchored on revenue the agency does not own. The retention math is computed against a book that can walk out the door with the producer who built it. The structural protections in the Purchase Agreement only matter if the underlying contractual ownership flows to the agency entity that the buyer is acquiring.

The diagnostic is a legal-DD overlay on the customer-DD work — three documents in three folders of the VDR have to align before the financial model means anything. The producer employment agreements have to grant the agency ownership of the producer's book during employment and after termination. The non-compete and non-piracy covenants have to be enforceable in the relevant state (some states will not enforce non-competes; non-piracy is generally more enforceable but the drafting has to be precise). The renewal-rights documentation has to confirm that the agency, not the producer, is the contractual broker of record at renewal — even if the producer ostensibly "owns" the day-to-day client relationship.

Three findings flip the diagnostic from green to red. Producer-owned books. If any producer's agreement says the producer owns the book in any form, the affected revenue is at risk on Day 1 of post-close. Unenforceable non-competes. If the agency operates in a state where non-competes are not enforceable (California is the canonical example), the producer can walk to a competitor the day after close with the book in tow. Missing or weak non-piracy. Even where non-competes don't enforce, non-piracy covenants generally do — but only if the drafting includes the agency's clients by name or by definition and includes a damages mechanic the agency can actually invoke. A non-piracy covenant that requires the agency to prove the producer "induced" a client to leave is structurally weaker than one that prohibits the producer from servicing any of the agency's client list for a specified period.

When the diagnostic flips red, the deal is not necessarily dead. The structural response is to require the seller to amend the producer agreements pre-close (often with a portion of purchase consideration conditioned on the amendments being executed), to size the earnout against the at-risk revenue specifically, and in the most concentrated cases, to require the seller to introduce the buyer to the affected producers during the LOI window and confirm verbally that the producer intends to remain post-close. None of these are bulletproof; all are better than discovering the producer-ownership problem post-close.

§ 06 · VerificationFile review, triangulation, sample audit.

The verification work that closes the gap between the seller's reported retention numbers and the buyer-defensible numbers has three operational components. Each requires VDR access that goes beyond the summary-level retention reports the seller will instinctively share first.

File review. The buyer's diligence team — or a contracted outside firm — reviews a representative sample of client files in the agency's management system. The sample should be stratified: top-10 accounts (all of them), random sample of mid-tier accounts (20 to 30 typically), and random sample of small accounts (10 to 20). For each file, verify the contractual broker of record, the carrier appointment that services the policy, the renewal-date cadence and history, the producer of record, and any open service-failure or complaint records. The sample reveals patterns the summary reports conceal — books where the headline numbers look clean but individual files show three-year-old service tickets, broker-of-record changes the seller didn't disclose, or producer-of-record shifts that signal ownership ambiguity.

Triangulation. The retention numbers the seller reports should match three independent data sources: the AMS, the trailing P&L (commission revenue line-by-line), and the carrier statements (premium volume by carrier per period). When the three sources reconcile, the buyer's confidence in the underlying data is high; when they diverge, the divergence itself is the diagnostic. A common divergence pattern: the AMS retention number reads cleanly, but the commission revenue in the P&L shows a step-down in a recent year that the AMS doesn't explain. The cause is usually a carrier-specific commission-rate change, a class of business that runs lower-commission, or — occasionally — uncollected commission payable that the agency's books captured but reality didn't.

Sample audit. The buyer's diligence team contacts a small number of clients directly — typically the top 5 to 10 accounts — under cover of a "transition introduction" framing during the LOI window. The conversations serve two purposes: they validate that the client knows the agency, the agency knows the client, and the relationship is operationally what the seller represented; and they begin the post-close relationship before the close occurs. Sellers will sometimes resist the sample audit, citing relationship-protection concerns. The buyer's posture is that the audit is non-negotiable on a deal of any meaningful size — the largest accounts' confirmation of relationship reality is structurally required by lenders, by investment committees, and by the buyer's own risk tolerance.

The verification output is a written attestation — typically attached as a Purchase Agreement exhibit — that the buyer has reviewed the file sample, triangulated the AMS against the financial records, and conducted the sample audit. The attestation is the basis for both the buyer's offer multiple and the structural protections drafted into the agreement.

§ 07 · Structural protectionEarnout, holdback, non-piracy.

Risks the customer-DD verification could not fully clear get priced into the deal structure. Three mechanisms — earnout, holdback, non-piracy — work together to convert residual book-quality risk into structural protection that the buyer can actually invoke if the verified retention assumption fails post-close.

The retention-anchored earnout conditions a portion of the purchase consideration on the book's actual retention performance in years one and two post-close. The mechanic is binary or sliding: binary structures pay the earnout in full if a defined retention threshold is met and nothing if it is not; sliding structures scale the payment proportionally to the retention outcome. The disciplined buyer uses sliding, anchored to the same-store renewal retention number (§03) computed by the buyer's diligence team, not by the seller's AMS report. The threshold is set above the historical baseline by enough margin to ensure the seller has skin in the game but below the headline number by enough margin that the buyer is not paying full price for retention DD could not verify.

The retention holdback escrows a portion of the purchase consideration (typically 5% to 15%) for twelve to twenty-four months post-close, releasable against the actual retention performance of the Existential Dozen or against the concentrated-account thresholds identified in §04. The holdback is structurally distinct from the earnout: the earnout pays IF retention holds, the holdback pays UNLESS retention drops. The asymmetry matters because the holdback gives the buyer a cleaner setoff mechanism for specific identified accounts; the earnout gives the buyer cleaner protection against book-wide degradation.

The non-piracy covenant is the legal-defensibility backstop. It prohibits the seller's producers (and the seller personally, where applicable) from soliciting or servicing the agency's clients for a defined period — typically twenty-four to thirty-six months. The drafting determines enforceability: covenants that list the client base by definition (rather than by name) hold up better in litigation; covenants that include a liquidated-damages mechanic ($X per client per year) give the agency a clean enforcement path; covenants tied to the producer's employment agreement (rather than only to the seller's Purchase Agreement) extend the protection to producers who may not have been parties to the deal documents directly. The deeper legal-DD treatment lives in the legal architecture (Risk Management Provisions) and the legal architecture (Internal & Post-Close Agreements); surfaces the customer-DD requirement.

The pre-LOI buyer checklist

Before you countersign the LOI — before you commit the diligence team and the lender, even — walk through this checklist. If every box is ticked, the book quality has been verified to the standard the structural protections are designed to support.

• Policy retention AND premium retention computed from a three-year AMS export; the gap between the two priced into the offer or the earnout trigger
• Same-store renewal retention recomputed independently (not from the seller's headline report); the gap to net retention quantified as the new-business velocity component
• 15% Rule audit completed; any over-concentrated account identified and the concentration risk priced into the structure (carve-out, account-specific earnout trigger, or holdback)
• Existential Dozen mapped by carrier, line of business, relationship owner, and renewal cycle; defensibility scored per account
• Producer employment agreements reviewed; book ownership confirmed flowing to the agency entity (not to the producers individually); non-compete and non-piracy enforceability validated in the relevant jurisdiction
• File-sample audit, triangulation across AMS / P&L / carrier statements, and direct top-account sample audit completed; attestation drafted as Purchase Agreement exhibit

Getting this list to all-green takes most disciplined buyers the full ninety days of the LOI window. The buyer who skips the producer-ownership overlay is the buyer who discovers the book ownership problem in week six of post-close integration, when the producer who built half the commercial book hands in their resignation. The list is mandatory, and the order matters.