“Big data” is helping companies of all kinds tailor their products or services to customer’s needs. For many technology firms, though, big data is often about achieving small gains in efficiency or customer behavior changes, such as improving response rates or time spent on a company’s website.
Many people are familiar with how companies like Amazon and Netflix have been able to customize their product suggestions so that they are tied to a customer’s previous purchases. Some companies, like Pandora in the music streaming industry, have their offerings entirely built around the tracking of customer preferences. A few newer online retailers, like StitchFix or BirchBox , are now creating highly personalized online shopping experiences where the sites choose items for customers based on their preferences and the sites’ algorithms.
Compared to offline businesses, internet-based companies have been able to collect a huge amount of data and process it far more cheaply and quickly. Generally, though, big data is not about massive Holy Grail solutions. More often, it is utilized to achieve incremental gains in productivity across a variety of areas of the business – which, when multiplied over a huge number of instances (or allowed to work over a long time) produce a significant effect. Where massive scale is involved, even small effects can result in large total improvements.
These gains can be achieved in a number of business areas. Tests can be run to see what web page layouts result in users lingering longer, or which design features are more likely to get users to click through to the “buy” button. Delivery firms can perfect their navigation systems so that slightly shorter routes result in significant overall fuel savings. Facebook now can adjust its users’ news feeds so that people will post slightly brighter or gloomier items themselves.
The gains involved in such changes may be small, perhaps involving productivity gains of only 1% or so. But many of such changes can, over time, significantly alter a company’s competitive profile. The process can be compared to evolution; small marginal changes can, over time, make the difference over which companies outlive those that fail to make such adjustments.
How Does This Affect Real Estate?
New sources of real estate capital online lenders, or peer-to-peer (P2P) lenders, have arisen that specialize in raising debt for non-owner occupied real estate deals, such as investor rehab loans. These companies do not simply operate a website. They offer unprecedented speed, efficiency and transparency in the lending process. The automated processes used by these platforms can cut down the time spent on filling out paperwork, in meeting with loan officers, or otherwise waiting for underwriting to be completed. The result is that closings occur faster and with more predictability.
With online lenders that are focused on real estate, a sizable trove of credit data is available to them with respect to real estate loans and borrowers across the country. This data can then be input into algorithms that produce instantaneous determinations of creditworthiness and appropriate pricing. These processes are expected to subsequently lead to reduced loan losses and better overall returns for investors.
The risk modeling technology used by the better online lending firms can also be used for other purposes – for example, to improve upon the FICO credit score that currently underlies most credit decisions. Most such credit scores, for example, don’t include routine payments to landlords, telecom providers, and energy utilities — even though some research indicates that such payments are more accurate at scoring individuals’ creditworthiness. An online lender’s access to its own data creates what has been called a “flywheel effect” that continually increases the defensibility of its business. Data from loan performance feeds back into the marketplace lender’s model, creating an even more accurate model. As the accuracy of the data and model increases, the marketplace lender can offer borrowers lower rates – and as rates decrease, more borrowers flock to the platform, driving more data into the model. The best platforms will thus incorporate proprietary data to improve upon FICO scores in demonstrating correlations with repayment prospects.
Aside from the potential for smarter underwriting, the processing capabilities of online lenders translate into their being able to significantly speed up the entire loan process – a feature that is of direct benefit to borrowers or mortgage brokers. Whereas traditional lenders can take months to approve a mortgage, P2P platforms can use their assembled credit data to determine immediately if a deal fits their guidelines.
All of this trial-and-error testing and data analysis of small things still costs money, but it’s cheaper than it was. In any event, the better internet-based companies have become proficient at examining the fine grains of their various activities for small but useful improvements. Constant experimentation and rapid implementation of demonstrably better business features – even if the gains are tiny – can, in combination with other such improvements, produce significantly improved results.