I am opposed to that kind of arrangement.  There are three reasons for my opposition:

1. There simply is not enough margin, even with a higher fee for the successful loans considered, to cover the additional work (see point number 2). 

While it is customary for title insurers and their agents to charge only when the loan closes, the premiums, endorsement fees, etc. charged on the successful files should well enough subsidize the washes. 

2. The abstractor would probably see a need to update any unpaid search to verify the truthfulness of his/her customer.  Not knowing if or when a loan may have closed, or how long it might take for the mortgage document to travel through the various channels before arriving at the clerk/recorder's office, one might conduct several updates on a single file before being satisfied that the loan was indeed a wash.

3. It has never been standard practice, to my knowledge, for payment of a title search to be contingent upon closing/funding of a loan.  This sort of practice might start out seeming inoccuous enough, with honest dealers like Seth looking for a way to avoid downwardly mobile income.  Then come the companies working for clients with higher fall-through rates. As the practice gains acceptance, the abstractor will find him/herself devoting far too much time to investigating unpaid files.  Practices unfavorable to the abstractors seem to creep in this business.  They soon enough become demands instead of arrangements.

The way it should work:  Title insurers and their agents who wish to venture beyond the boundaries they are able to handle with their own search staff should price their service in accordance with the reality that bills must be paid for the files that fall through.  Employing search staff across a wide region would still be for more expensive than absorbing the costs of files that don't pay off.  It is their job, also, to keep an eye on their own customers' cancellation rates.

And another thing.  The middlemen/women, the vendor managers, who are not issuing policies, but simply reselling the searches to the insurers or agents, should be resisting this ominous practice along with the abstractors.

That's what I think, anyway.  How about you?