A mid-sized VRM who wants humans to do what humans do best
Vacation rental managers are a creative and inspiring group. Working with our VRM clients, we’ve been identifying very powerful ways to help them turn their guests into loyal direct booking customers.
Email marketing and promotion campaigns are widely used across industries – that much is known.
However, specific optimization capabilities can allow marketers to launch more effective email campaigns – and those are generating increased guest engagement and conversions.
A mid-sized vacation rental management organization who Jarvis ML recently started working with is executing more fruitful email campaigns using our personalization engine. We’re here to explain what’s going on behind the scenes and share how they achieved success.
Check out the quick success story here!
Start with the end
Let’s kick it off immediately with the key takeaway: By applying machine learning to their entire database – of previous customers, web analytics, properties, and booking history – this VRM team is now able to convert more guests by providing personalized, contextualized experiences. They can predict and target the guests who are likely to book next, and then individualize the emails to each guest based on the properties and prices that will drive them to engage and reserve.
How did their team get to this capability? We’ll explain.
This VRM’s Marketing team had already been sending their previous guest list a couple different flavors of emails:
- Promotions to audience cohorts presumed as high-potential clients
- Abandoned search & abandoned cart follow-up email campaigns
These types of emails can be sent by anyone and are by themselves an important way to engage your known leads and returning guests. (Hint: make sure you’re doing this!)
We’ll focus on the first type to start. Identifying high-potential, high-value guests is something that can be done manually with some analysis; however, optimizing emails to drive significant revenue uptick requires deeper statistical predictions!
Detecting the right audience cohorts: high value and likely to book
Our VRM client in the south-eastern United States is – like many regions – subject to seasonality peaks and troughs. So some of the optimization around timing can be deduced at the market-level by time of year.
However, this company wanted to go beyond the property and location data, and instead understand each individual guests’ likelihood to book at different times – and then use that to target their emails acutely.
Before investing in machine learning (ML), their data analytics team had been reviewing guests that have previously booked, and generated email recipient groups based on those manual weekly analyses. But again, it was time consuming and they knew they use technology get more thorough results.
To achieve this, their marketing and analytics team started using Jarvis ML’s machine learning platform to crunch through their key data sources and start generating that list of prioritized guests. ML is a self-learning, statistically driven powerhouse that finds correlations that humans simply cannot match at scale.
The output: a specific list of the highest priority people to target on any given day. The platform also calculated the lifetime value (LTV) of these customers to rank-order by a tangible dollar figure. These newly generated data points were pushed back into their CRM.
At this point their marketing team was equipped with the audience lists that would provide the highest expected value return from emails.
Personalized content in each email
The next step was to personalize the email content to each recipient.
These days it’s common knowledge that personalized experiences drive more engagement and conversions from customers. So the strategy was a no-brainer. But the bottleneck was how to automatically execute this effectively at scale?
In addition to the processing done for audience-curation, the Jarvis ML platform ran models on their property inventory data, previous booking transactions, and customer database – applying machine learning to find the best match between guest + property + price.
A machine learning system learns from both historical data and new, real-time data to deduce the best ways to maximize business goals. (To learn more about how Machine Learning works for vacation rentals, you can check out this overview.)
After the calculations took place, the core ingredients for launching the campaigns were ready. The Jarvis ML platform pushed recommendations for the top properties and optimal price-points on each of those properties for each email recipient. These predictions were dynamically inserted in the outbound emails person-by-person, with every recipient having their own unique email content.
So, how did it go?
These email recipients engaged with the content more, spent more time reviewing the properties on the website, and ultimately booked hundreds of thousands of dollars in revenue on the specific property recommendations calculated automatically. In the longer run, the VRM is seeing $1M+ in monthly revenue growth that is being supported by their machine learning initiatives with Jarvis ML
Perhaps obvious, but an important reminder: sending emails to previous guests to drive new bookings is a business standard. But this VRM took the next step by layering in personalization that improves conversions, individualizes recommendations per guest, and does this all without any manual time-consuming analysis work.
Adding machine learning to tried-and-true business practices boosts their effectiveness.
Abandoned Search & Abandoned Cart follow-up email campaigns
This is the second flavor of email campaign that our VRM client was running. Guests in this audience cohort have self-selected as being highly interested in booking by initiating the browse, search, and/or booking flow process.
Since this prioritized audience was known already, the best optimization application is dynamic recommendation of properties within the email content. The same process as above was followed to personalize each email to each customer, just skipping the need to define the high value, prioritized list.
So what happened?
Getting better every day
Before moving forward, here is some context on machine learning.
The unique attribute of ML is its ability to execute self-guided decision making and learn in real time to improve iteratively. That means the ML isn’t coded explicitly to make rule-based decisions; rather, it’s programmed with infrastructure and statistical models to make the best decisions on its own.
Once an ML system makes its predictions, the “real-life” data – about engagement, clicks, page views, conversions, and revenue – is fed back into the ML system as new inputs to learn from to enable ongoing improvement. So each ML model is adjusted continuously based on the new successes and failures, forcing the system to improve on its own automatically based on what actually happens.
Back to our vacation rental business. When Jarvis ML was first turned on for this “abandoned search & cart” email audience, the VRM’s data team saw relatively similar results between their previous methodology for emails and the new machine learning powered solution.
However, week by week the numbers got better and better for the ML-personalized email campaigns. It was learning, iterating, and most importantly continuously improving.
Turns out, this is expected behavior – and is a differentiator of leveraging ML powered-software.
It’s a system that not only directly improves results, but does so while saving time and reducing operating expenses. Humans are freed up to do what they do best – being creative, exhibiting deep industry judgment, and making strategy decisions. Meanwhile, the machine crunches data and makes ever-improving executional decisions for the business.
Takeaways for email campaign optimization: Personalization Works!
There are tons of best practices about optimizing email campaigns. It’s a necessary component for every email send, and at this point it’s a science: from A/B testing your subject lines, CTAs, creative, and everything in between… to selecting the right time-of-day to send… to the right landing pages… to the deliverability best practices… and on and on.
Personalization became a business standard years ago for emails – even just adding a first name token and segmenting your audience into 3 or 4 groups to help high-level relevance.
This vacation rental manager stepped up their email optimization by automatically and dynamically inserting the most relevant recommendations for each customer’s top properties into emails.
Like Google search results, these tailored recommendations are rank ordered in the email; the associated images and property attributes are displayed; and each property leads to the right landing page. This business is even incorporating personalized dynamic pricing to fully curate the booking process.
Many of the optimization efforts applied to running email campaigns take human creativity, manual work, and some strategic planning. Our client has a motto internally: “let humans do what humans do best”. Their leadership team and individual analysts alike use this motto when thinking about what technology they should introduce into their stack.
In the case with Jarvis ML, they found the right technical solution to serve their customer experience personalization needs with automation. In return, they enabled their people to focus their time on the most fulfilling, valuable work.