Retention in the 80 to 85 range is normal for an agent doing its own search, exam and clearing, so the underwriter is pricing pure underwriting risk. If you buy search product from the underwriter or an outsourced abstractor, your effective retention drops fast, and abstracting a plain suburban residential file can run $75 to $200 depending on the county and whether the records are digitized. Also check what kind of rate state you're in before you build the model. Some states promulgate a single rate every agent must charge, others use filed rates by underwriter, and a handful of attorney-closing states change the whole labor picture. That one variable moves your $1,800 assumption by hundreds either direction.
On claims, most agency agreements make the agent responsible for losses arising from its own search and exam work, often with a per-file retention and sometimes an obligation to cure defects out of pocket before the underwriter gets involved. Read that section with a lawyer, because it's where a $1,650 file turns into a $40k file. Your E&O is separate from the underwriter's policy and does not usually cover escrow shortages.
The number that closes agencies isn't the split. It's a wire going out to a spoofed payoff instruction. You need crime and cyber coverage, callback verification on every change of instruction, and trust reconciliations done daily, because most states require three-way reconciliation and underwriters audit for it.
And 25 to 30 files a month is a referral assumption, not a market assumption. If four agents or one lender send you most of it, price the month where one of them leaves. RESPA limits sharply what you can pay for referrals, so growth has to come from service, not from buying the pipe.
