Continuing with yesterday’s post Machine Learning and the Startup World, a friend asked what machine learning applications might look like for marketing automation platforms (e.g. Pardot). Good question. Considering marketing automation platforms collect so much information about leads through email opens, form submissions, web page visits, ebook downloads, webinar signups, etc., there’s a tremendous amount of training data for machine learning to find insights.
Marketing automation is essentially human-defined rules based on what they think is best (e.g. send email A, wait two days, send email B, etc.) Machine learning “learns” over time and can figure out patterns humans can’t because there’s too many dimensions to analyze. Here are a few ideas on how to use machine learning in marketing automation:
- Email Send Time – Analyze when leads opened emails in the past and automatically schedule future emails to be sent at the same time (time of day, day of week, etc.)
- Email Message to Trigger – Analyze what emails (or other types of content) were most closely associated with pushing the lead through the current phase of the lifecycle (e.g. engaging with a sales rep) and trigger the email automatically (as opposed to human-defined static rules)
- Lead to Opportunity Probability – Analyze every piece of historical data for leads that turned into opportunities (the training data or learning set) and come up with a probability for all other leads that haven’t turned into an opportunity based on how their behavior matches the training data
- Lead to Customer Probability – Like the lead to opportunity probability, do the same thing for leads that became customers (not just those that had an opportunity in the pipeline associated with them) and come up with a probability that any given lead will become a customer
Machine learning has applications in all fields, especially marketing automation. Look for existing vendors to add this type of functionality as well as new vendors to emerge that take advantage of this new technology.
What else? What are some more ways machine learning can be applied to marketing automation?
One of the sales and marketing technologies I’m most excited about (along with account-based sales and marketing) is the whole predictive area. At a simple level, predictive takes existing contacts and opportunities and scores them against a dynamic model based on other contacts and opportunities that became customers. Put another way: find great-fit companies that look like our existing customers so we can target them.
Account-based sales and marketing platforms (see SalesLoft and Terminus) solve the major problem of running programs in a scalable manner against hundreds (or thousands) of target accounts. Traditional marketing, and sales as the opportunity progresses through the funnel, casts a net, sees what is caught, and then works the qualified leads. Now, as a more modern approach, account-based sales and marketing goes spear fishing and proactively seeks out best-fit accounts based on a number of dimensions. Only, there’s often not an easy way to find and refine best-fit accounts — enter predictive technologies.
Here are a few thoughts on predictive sales and marketing technologies:
- As more companies implement account-based sales and marketing platforms, predictive systems become more important to find and analyze best-fit accounts.
- Predictive sales and marketing systems are clearly in the path of revenue.
- Machine learning, a subset of artificial intelligence, is now more accessible with the advent of systems like AWS Machine Learning, making predictive systems more powerful.
- Finding lookalike companies requires technology. Combing through billions of records by hand simply isn’t possible.
Look for the category of predictive sales and marketing systems to grow fast as the technology crosses the chasm and becomes more well known.
What else? What are some more thoughts on predictive sales and marketing technologies?
Earlier today I had the chance to participate on a panel at the new home of the Atlanta Braves: SunTrust Park. After seeing the stadium progress (very cool!), we talked for 30 minutes about the intersection of sports and technology, especially the innovation opportunities that lie ahead.
Here are a few of the ideas discussed:
- Teams want to use technology to enhance the fan experience, especially the downtime when the game isn’t happening (e.g. between innings, pitcher changeovers, etc.) and the community element
- Bandwidth is no longer a concern at modern stadiums as every fan can stream video real-time, opening up new opportunities (drones? more cameras on wires?)
- Startups play an important role in the sports technology world even though most will fail (it only takes one to change the world)
I’m excited about the future of sports and technology.
What else? What are some more thoughts on the intersection of sports and technology?
The Trade Desk, a demand side advertising platform, just filed their S-1 IPO filing to go public. Demand side advertising is where ads are purchased to be then served by supply side advertising platforms. One of the big benefits is having one system to buy ads that then integrates with dozens of other systems.
Now, let’s take a look at the S-1 from their recent filings. Here are a few notes:
- Our platform provides access to approximately 3.2 million ad spots on average every second for our clients to bid on across millions of different scaled media sources—websites, shows, channels, stations and streams. (pg. 1)
- In 2015, approximately $14.2 billion was transacted in the 1 Table of Contents programmatic advertising spot market via real-time marketplaces, according to Magna Global. (pg. 2)
- Financials (pg. 2)
- 2015 – $113.8 million
- 2014 – $44.5 million
- Net income
- 2016 1H – $6.6 million
- 2015 – $15.9 million
- 2014 – $5,000
- Gross billings (pg. 11)
- 2015 – $530 million
- 2014 – $201 million
- Trends (pg. 3)
- Media is Becoming Digital
- Fragmentation of Audience
- Shift to Programmatic Advertising
- Automation of Ad Buying
- Increased Use of Data
- Approximately 389 clients, including the advertising industry’s largest agencies, as of December 31, 2015 (pg. 5)
- Clients can easily buy targeting data from over 80 sources through our platform (pg. 5)
- Average days sales outstanding, or DSO, of 88 days, and average days payable outstanding, or DPO, of 64 days at June 30, 2016. (pg. 18)
- Approximately 7% of our gross spend in 2015 was derived from outside of the United States. (pg. 29)
- Access to borrow up to $125.0 million aggregate principal amount of revolver borrowings (pg. 31)
- Accumulated deficit of $28 million (pg. 48) (Note: it’s really impressive to achieve this scale and growth rate on this relatively small accumulated deficit).
- Between our inception in November 2009 and June 30, 2016, we generated aggregate proceeds of $88 million from the sale of convertible preferred stock (pg. 68)
- Ownership (pg. 124):
- VCs – 34%
- Founder / CEO – 26.7%
- Founder / CTO – 1.5%
Congratulations to The Trade Desk for building a great business in seven years and for heading towards an IPO as their next milestone. This IPO will be well received based on scale, growth rate, and profitability.
What else? What are some other thoughts on The Trade Desk IPO filing?
Continuing with yesterday’s post on Atlanta Companies on the 2016 Inc. 500, there are two tech startups that really standout: Kabbage and Cardlytics. Both are financial tech (FinTech) companies that are growing super fast. Here’s a bit about each and their revenues for the last two years:
Kabbage – Small business lending based on alternative data sources to evaluate credit worthiness (e.g. checks your eBay ratings, Amazon ratings, UPS shipment volume, and QuickBooks statements to determine a loan amount). Here’s Kabbage’s revenue for the last two years as published in the Inc. 500:
- 2016 – $97.4 million
- 2015 – $40.1 million
Cardlytics – Aggregates data from 1,500 financial institutions to run online and mobile banking rewards programs (think anonymized purchase data from consumers that’s used to serve up relevant offers). Here’s Cardlytics revenue for the last two years as published in the Inc. 500:
- 2016 – $77.6 million
- 2015 – $53.4 million
Based on the scale of the business and growth of revenue, both of these companies would be in pre-IPO territory. Often, $100 million in revenue is the magic mark to go public and both should pass that this year. Congrats to both companies on the great growth and here’s to their continued success.
What else? What are some more thoughts on Kabbage and Cardlytics?
Last week I heard a new term: HESaaS. HESaaS stands for Hardware Enabled Software as a Service and the idea is that there are new SaaS opportunities that come from the addition of specialized hardware. Put another way, the Internet of Things (IoT) is going to enable a variety of new HESaaS opportunities.
A local Atlanta Tech Village HeSaaS startup is Gimme Vending (disclosure: I’m an investor). Gimme makes a device that transmits vending machine data to the cloud for more efficient inventory management and product merchandising analytics. Without the hardware to send the data to the cloud, there’s no SaaS business.
Add HeSaaS to the list of reasons to be Bullish on SaaS Growth.
What else? What are some other hardware enabled SaaS opportunities?