When measuring marketing campaigns, engagement rates count the most. Personalized emails had a 29 percent higher open rate and 41 percent higher click rates than emails without any personalization, according to a recent study. Surprisingly, however, 70 percent of brands fail to use this feature.
Knoxville-based Lirio uses machine learning to reduce the effort behind the daunting task of mass personalization. Their digital automation platform helps financial advisors create highly personalized content and target specific segments for maximum engagement.
“In marketing, you’re always in pursuit of the right message at the right time to the right person. That’s essentially what we’re using technology to assist us in accomplishing,” says CEO Jeremy Floyd.
Floyd saw a need for optimizing communication between financial advisors and clients while working as a chief marketing officer at a financial services firm. After acquiring fintech firm Finworx in December 2016, Floyd and his team incorporated its product into Lirio.
The product highlights relevant communication (think the present needs) of the customer and uses person-based communication — from low- to high-risk tolerance and current life phase — to identify needs. After filling out an investor survey, including product history and digital behavior, machine learning algorithms guide the advisor through each communication to only send what’s most relevant to the client. These are all selected from the startup’s growing library of targeted messaging.
Here, Floyd shares more about how Lirio can improve advisors’ engagement with clients, what they are looking for in their next funding round, and why the perception of the southern startup scene isn’t always accurate.
What problem are you solving with Lirio?
At the core we’re trying to change human behavior by using behavioral economics on principle and applying machine learning to optimize the next best action or the next best communication.
We are the first technology application within utilities to adopt energy-efficiency behavior. For example, a regular utility customer thinks about energy efficiency less than ten minutes a year and so, anything that we can do through our communication technology to get them to stop and maybe buy LED lights or buy a new refrigerator or new windows. By learning what their behaviors are and what the best next communication, we can help the customer react in some way.
How is machine learning helping Lirio become more efficient at learning human behaviors?
The amount of processing of a human to create workflow, essentially A/B testing, requires a lot of pre-planning, a lot of whiteboard time of looking into conditional logic and then constant analysis of the reporting. So, what did they do? What are we learning? What are the trends? What we use machine learning for is to collect that information and optimize that next communication that’s going out based upon a constant drip of not just the email plan time and open time and reaction, but also what were the unique characteristics of that assembly?
In some instances, we’ll have 20 different components of an email that are being composed together at the time of send. So of those 20 components, what was the combination of the email at that particular time? What headline, subject line, what did they find that they react to this time in order to know on the next send, what’s going to attract that behavior of that individual?
What are some features within the platform?
With mail merge, you’ve got some variables that are applied to an email and you can do a mass email that has some personalization with the person’s name.
Our system creates mass personalized emails with more than just names. For example, we are sending out an email to over 100,000 utility customers and there are 17,000 variations of the same email. It’s composed, at a discreet level, where sentences are combined together to form paragraphs in order to drive that behavior. That’s a pretty key feature — a content management system.
It’s also looking at all of the previous behavior — the personality type, the information that we know about an individual and composing it at the individual level.
What’s your revenue model?
Our revenue model with utilities, for example, is on a per customer, per year basis and that’s depending upon the size of their client base. We are their customer success manager, so we handle all of the population of the content and the importing of the list. Then our clients just have access to their DOS words and they can see individual activities, etc. And then for the financial services application, each advisor pays a monthly fee for access.
What’s your current funding status?
We’ve basically had a small group of investors that have invested a seed round. We are currently planning to have an institutional round some time in spring 2018.
What are your thoughts on building your company in the Southeast?
It comes with stigma. The stigma against the South is that we can’t recruit good talent, that we don’t have access to the best minds that there are, that we’re not networking on a regular basis, and that funding is difficult. We found talent and our seed round here.
Secondly, with the networking of knowledgeable entrepreneurs, we found the Southeast to be a great resource for business growth. Atlanta is one of the leading fintech cities in the country and there’s Chattanooga and Knoxville — all three have really growing entrepreneurial communities. Maybe there’s a little bit of a chip on the shoulder. But absolutely, we can make this work in the South and we’re proud of it.