The spread between what an IMB founder thinks his business is worth and what it will actually trade for is wider right now than at any point in the last fifteen years. The reason isn't that buyers are being unfair. The reason is that the math has shifted, and most sellers are valuing yesterday's earnings against today's market.
I'll walk through how acquirers actually build the number — every haircut, every adjustment, every assumption. If you're an IMB CEO thinking about a process in the next 24 months, this is the math you need to be fluent in before the first banker pitch.
Start with normalized gain-on-sale, not reported
Reported gain-on-sale margin in any given quarter tells you very little. Buyers normalize it across a full cycle — typically a trailing four-to-six quarter window weighted toward the most recent two. They strip out lock-volume timing benefits, hedge gains and losses, and any non-recurring secondary marketing items. They re-cost the loans against the buyer's own warehouse pricing, not yours.
The result is almost always lower than the seller's pitch number. In the deals I've worked, the gap between reported margin and acquirer-normalized margin has run anywhere from 15 to 60 basis points. If your business is producing $1B and the gap is 30bps, that's $3M of pre-tax earnings that doesn't survive the buyer's model. Which means it doesn't survive the multiple. Which means at a 5x multiple, that's $15M off the headline.
The implication for sellers: you should be running this normalization on yourself, twelve months before you go to market. If your normalized number is materially lower than your reported number, you have time to fix the operating story behind it. If you wait until diligence, the buyer is the one telling you the number, and they're the one keeping the spread.
MSR runoff and the recapture question
If you retain servicing, the MSR portfolio is doing real work in your valuation — but only at the level of recapture you can defend. Most sellers mark MSRs against published prepay benchmarks and a recapture assumption pulled from industry data. Buyers re-mark them against the seller's actual recapture history, which is almost always lower than the assumed rate.
The math compounds. A 200bps reduction in assumed recapture flows through the cash-flow projection on the entire portfolio over its expected life. On a $5B servicing book, that's a meaningful adjustment to MSR fair value, and it's an adjustment the seller has very little ground to argue against, because the data is the seller's own.
If your recapture is genuinely strong, prove it pre-process. If it's weaker than your mark assumes, decide whether to fix the operating gap or shorten the assumed life and take the markdown now in your own model. Either way, don't be surprised by it in diligence.
The recruiter-dependent volume haircut
Acquirers will look at your top decile of LOs and ask: how much of this volume walks if we don't get the comp plan exactly right? The answer dictates the haircut.
In businesses where the top 20% of producers generate 60–75% of volume — which is most IMBs — the buyer is, in effect, buying those producers as much as they're buying the platform. They will protect themselves by either negotiating retention bonuses into the deal structure (which comes out of seller proceeds), discounting the projected volume from those LOs by 15–30% in the model, or both. The discount shows up as a lower revenue base, lower contribution, and a lower multiple-applied number.
The defense is producer-level data: tenure on platform, comp parity vs. market, recapture and retention by LO, the recruiter activity in your footprint that the LO has actually fielded. Sellers who can show ten-year median tenure and below-average recruiter pickup get a smaller haircut. Sellers who can't, get a larger one — sometimes large enough that the buyer walks.
Branch-level cost-to-serve
Most IMB CEOs run their business at the consolidated level. Buyers run it at the branch level. The mismatch matters because the buyer is going to ask, "what's the contribution of branches 1 through 30 individually, and which would I close on day 31?" — and the answer changes the cost base.
If the buyer concludes that eight of your thirty branches are negative-contribution after fully-loaded allocations, those branches don't get valued at zero. They get valued at the cost of closing them — severance, lease termination, the production they take with them on the way out. That's a markdown to value, not a neutral adjustment.
If you have negative-contribution branches that you're carrying for strategic reasons, surface them in the process and explain. If you have negative-contribution branches that you didn't realize were negative, the diligence finds them anyway, and you're explaining from a defensive posture.
Tech debt as a value adjustment
The platform you built in 2017 is not the platform a buyer wants to operate in 2027. LOS modernization, POS replacement, integration cleanup, MSR system refresh — these are real dollar requirements that show up in the buyer's post-close capex plan and reduce the price they're willing to pay today.
I've seen tech-debt adjustments range from $2M for a clean platform that just needs an LOS upgrade to $15M+ for a platform that needs a full stack rebuild. The seller is rarely thinking in these terms because the modernization isn't on their three-year plan. The buyer is thinking in these terms because the modernization is on theirs.
The IRR math acquirers actually run
What ties all of this together is the IRR the buyer is solving for. PE buyers in the lender/servicer space are typically targeting mid-to-high-teens IRRs over a four-to-six-year hold, with leverage assumptions that have tightened materially over the past 24 months. Strategic acquirers run different math but apply similar pressure on net adjustments.
Working backward from that IRR, every basis point of gain-on-sale margin matters. Every recaptured loan matters. Every retained branch and every retained LO matters. The valuation isn't a multiple applied to last year's EBITDA. It's a present value of a forward cash-flow projection that the buyer has stress-tested against their target return.
The implication: a seller who can credibly improve any one of these levers in the twelve months before a process — normalize margin, prove recapture, lock retention, rationalize branches, retire tech debt — moves the buyer's IRR math, which moves the price they can afford to pay. That's where the real money in the sale process gets made or left on the table.
The market sees your leaks before you do — in the talent you can't attract, the pricing you can't hold, and the assumptions buyers won't accept. The valuation is the receipt.
What this means for IMB CEOs
If you're 24 months from a process: pick two of the levers above and fix them. Margin normalization and producer retention give you the highest dollar leverage per quarter of work.
If you're 12 months out: build the data room around the buyer's likely adjustments, not around your reported numbers. The seller who frames the story for the buyer's model wins the negotiation.
If you're already in a process and surprised by a markdown: the markdown is real, and arguing with the buyer's math doesn't usually move it. The path to recovery is showing the buyer something they didn't see — usually in producer data, recapture data, or branch-level operating story — that flips one of the assumptions in their model.
None of this is theoretical. It's how the math has actually played out across the IMB deals I've worked. Sale processes leave money on the table because sellers walk in with reported numbers and walk out with normalized ones. The fix is to do the normalization early, on yourself, and use the runway to close the gaps.