Sam Howard

Improving the interface between technology and manual underwriting

I received an extremely nice email over the weekend from a happy client whose project has just completed and sold.


The borrower wrote: “Thank you for your kind assistance on our project, I am very grateful for your help and enjoyed working with you, I hope to bring forward another project later this year for your consideration.”

In my eyes, this is the secret of a successful lending business: you provide a Rolls-Royce service in the fastest possible time, enabling a client to borrow the money they need as flexibly and expediently as required. Thus, ensuring they make as much profit as possible and then become a repeat client.

But, in fact, more importantly, it is the initial underwriting that is undertaken. Many lenders will reject a part-built site on the basis that they are assuming unnecessary risk and will only lend on new schemes or once practical completion has been reached. With many years of development finance experience, we are able to ask the key questions at the outset in order to provide ourselves with the comfort to lend on a part-built site.

This is very much a risk business and you need to know what you are doing.

The client I mentioned above came to us with a part-built, two-house scheme. One of the houses had substructure works completed up to damp-proof course level, whereas the other had reached completion of brick and blockwork. There were issues with footpaths through the site and build over agreements to contend with as well. By ensuring that build regulations were up to date and build warranty was in place, we started to get a picture. The need to check directly with these providers is clear and present.

I often speak to industry colleagues about the role of technology in credit underwriting and it is true a lot of real estate credit underwriting is still manual, as the reality is that very few development finance loans are homogeneous.

In addition, the nuances are such that often a yes/no question is not sufficient. For example, a part-built site yes or no would render an excellent loan unfundable.

One has to dig deep and from the outset know what information you need to obtain and the questions to ask to elicit such information.

Is this a manual process? Yes and no. We can start to create the intricate systems that automatically ask the right questions and then drill down following receipt of the answers to ask the correct follow up questions. It still requires the old-fashioned approach of an expert underwriter to interpret the answers, but it ensures two things. First, the process is speeded up so you are not discounting loans that could be fundable, but you are automatically weeding out the ones that can’t and avoiding a manual process at the outset. Second, you are codifying the questions you need to ask as, dare I say it, even old-fashioned underwriters can miss something.

So, as I reflect on a happy client, for me it begs the questions: how can I continue to improve the interface between technology and manual underwriting to obtain the best result and keep the client happy so that they come back to us for another loan?

Leave a comment