LendingQB Study Finds 65 Percent of Mortgage Tech Implementations Result in Failure – NMP Skip to main content

LendingQB Study Finds 65 Percent of Mortgage Tech Implementations Result in Failure

Oct 22, 2012

LendingQB has revealed the results from its Enterprise Process Assessment (EPA), engagements with clients and prospects, which is a workflow evaluation model designed that helps lenders make objective decisions on their technology initiatives. LendingQB developed the Enterprise Process Assessment (EPA) as a tool to help lenders fully understand the drivers that motivate technology improvement efforts. "Research shows that upwards of 65 percent of complex software implementations result in failure," said David Colwell, vice president of corporate strategy at LendingQB. "Even for implementations that do succeed, more than a third of these projects go over budget. The goal of the EPA is to help lenders avoid being one of these statistics and achieve an optimal ROI that effectively addresses the technology goals they have in mind." LendingQB unveiled their EPA model earlier this year, and conducted a series of Webinars over the summer to help lenders objectively evaluate mortgage technologies. The EPA model engages lenders through in-depth interviews with lending executives and management, and is followed by detailed mapping of a lender's unique workflow, which then provides a framework for LendingQB to construct a custom survey that gathers productivity assessments from the lender's staff. A statistical analysis is then applied to identify key improvements and correlate data along with recommended technology objectives and a clear path forward roadmap. EPA findings show that: ►Lenders tend to focus on surface-level features instead of addressing solutions to underlying problems. ►There are typically five major productivity bottlenecks that lenders try to address with new technology. ►Prioritization of features is determined using subjective methods instead of relying on objective or empirical data such as productivity improvement. ►Stakeholder input is typically gathered top-down versus bottom-up. ►Vendor evaluations focus primarily on system functionality and give less weight to system utilization. The EPA provides lenders with a balanced and objective perspective on a vendor’s complete technology stack and uncovers issues that are not readily apparent to management. "Even large organizations that can produce detailed RFPs are not immune to what I term 'feature enamored syndrome,' or poor weighting of technology demands," said Colwell. "We designed the EPA to be as an aid to existing evaluation methods. As a company that has successfully implemented systems for hundreds of financial institutions, our value to lenders is more than the systems we build. We want to contribute to a lender's success, whether they use our technology or not."
About the author
Published
Oct 22, 2012
More from
Tech
Pennymac Expands AWS Partnership To Deploy AI Borrower Assistant

Lender says conversational AI will support borrower outreach and application workflows while accelerating cloud modernization of Plaisse

Jun 16, 2026
Ardley Launches Self-Service Platform For Mortgage Portfolio Growth

New platform lets lenders and servicers adjust pricing, borrower journeys and outreach campaigns directly

Jun 15, 2026
Lorien AI Integrates With MeridianLink Mortgage

Integration brings conversational AI, document analysis, and underwriting guidance directly into the loan origination workflow

Jun 15, 2026
More Than Half Of Buyers Say They'd Purchase A Home Without Human Help

Veterans United survey highlights growing consumer trust in AI-powered mortgage guidance, lender shopping, and document management

Jun 12, 2026
Friday Harbor Integrates AI Pre-Underwriting Platform With MeridianLink Mortgage

Integration brings AI-powered pre-underwriting platform into MeridianLink's loan origination system

Jun 09, 2026
Morningstar Adds AI Access To CMBS And CRE Analytics Platform

New capability lets users query commercial real estate and securitized credit data using natural-language prompts

Jun 05, 2026