Every title operation has a rework rate. Almost none can tell you what it is. Ask a manager how many files came back for a re-search or re-exam last month after being marked complete and you’ll get a shrug, because rework doesn’t appear anywhere on the P&L. It hides inside labor cost, inside turn times, inside the vague sense that everyone is busy. You’re paying this expense line every month. It just doesn’t have a row in the ledger.

The Arithmetic That Makes It Worth Measuring

Say your shop completes 400 searches a month and 8% get reopened for any reason. That’s 32 files touched twice — and a redo isn’t cheaper than the original work. It’s usually more expensive, because it starts with reconstructing what the first pass did before adding what it missed, and it happens under deadline pressure with a closing already on the calendar. Call it 1.25x the original search cost, plus the coordination overhead of pulling an examiner off new work. At loaded examiner cost, that 8% redo rate is quietly consuming something like 10–12% of your total search capacity. At mid-size volumes, that’s a full-time examiner’s worth of output producing nothing new. And that’s before the downstream costs: delayed closings, the fee credits you eat to keep a lender relationship, the occasional cure cost or claim when the redo happens after closing instead of before.

One Number, Then Three Buckets

The measurement is simple enough that the only reason shops skip it is that nobody assigned it. Define a redo as any file that re-enters search or examination after being marked complete, for any reason. Count them. Monthly, trended. That single number puts you ahead of most of the industry.

But the number alone doesn’t tell you what to fix, because “redo” is three different diseases presenting the same symptom. Every reopened file needs a reason code, and nearly all of them land in one of three buckets.

Index gaps. The plant or the county index didn’t surface a document that existed — a missed lien, an unindexed release, a name-variant miss, a gap-period recording. The examiner did the job correctly against an incomplete picture. This bucket is a data problem, and no amount of examiner training touches it. If it dominates, the fix is upstream: plant coverage, index quality, name-variant handling. The July 7 post was about exactly this layer.

Examiner judgment. The documents were all there; the read was wrong. A missed exception, a misconstrued legal, an easement that should have been raised. This is the bucket everyone assumes dominates — and in shops that actually code their redos, it’s usually smaller than expected. It’s the one that responds to training and second-review policies on complex files. Look at concentration before process: three examiners with clean records and one at a 20% redo rate is a management conversation, not a redesign.

Order intake errors. The search was executed perfectly against the wrong question — wrong legal, wrong parcel, misspelled party, missing seller. Everything downstream of a bad intake is waste regardless of quality. This bucket is invisible without coding, because the redo lands on the examiner’s desk and gets mentally filed as a search problem when it’s a front-office problem. It’s often the cheapest of the three to fix: intake validation, legal-description verification against the plant at order entry, required-fields discipline.

Run the decomposition for one quarter and I’d bet the distribution isn’t what you assumed. Most operators expect judgment to lead. Most coded data I’ve seen puts index gaps and intake errors together at well over half — which means the instinctive responses to rework, leaning on examiners and adding a QC layer, are treatments for the smallest bucket. They add cost to every file to catch errors that originate somewhere else entirely.

A Warning About the Scoreboard

The moment you start counting redos, people start negotiating what counts as one. Keep the definition mechanical — file re-entered search or exam after completion — and keep the reason coding blameless. The point of the number is to route the fix to the right layer, not to rank people. A shop where examiners fear the redo metric will stop reopening files that need reopening, and that failure mode costs more than all the rework it hid.

The deeper win is that measurement converts an ambient frustration into an engineering problem with a denominator. Once you know the rate is 8%, that index gaps are half of it, and that half of those trace to two counties with weak indexes, you know exactly what a fix is worth and what you’d rationally pay for it.

Where Scribe Fits

Scribe attacks the middle of this problem: it reads the instruments, orders them, and drafts the report with every document traceable back to source — which shrinks the judgment bucket and, just as valuable, collapses the cost of reconstructing a file when a redo does happen, because the first pass left a trail instead of a mystery. It does nothing for your intake errors, and I’d rather tell you that than let you discover it. Different bucket, different fix.

Measure the rate. Code the reasons. Then spend your money on the bucket you actually have instead of the one you assumed.

See what Scribe does to the judgment bucket →