Intent-Driven Automation for B2B Marketing
In this session at GrowthHackers Conference 2018, Guillaume “G” Cabane, VP of Growth at Drift, presented how B2B marketing is evolving into a hyper-personalized experience – and shared Drift’s evil plan for AI to take over the marketing world.
Below is a summary of G’s keynote session. For more #GHConf18 session recaps and growth strategy resources, head over to Rocket Strategy.
“The days of ‘spray and pray’ are over.”
For years, marketers have been using ‘spray and pray’ tactics because they don’t know much about their contacts. Many marketers do everything they can to lower the cost-per-lead, rather than focusing on optimizing the right experience for each lead.
Knowledge is what changed in 2017. We went from our customers being completely unknown to having knowledge about them and being able to make basic predictions. But accurate predictions require data.
Take the sales velocity of a chat campaign in 2017, for example. Chat has a human cost. Some leads are good, and some are bad. So marketers try to lower the cost by automating things. But this automation often leads to bad user experience.
Automation can generate a better B2B marketing experience, but for that to work, you need to know more about your leads so you can predict their sales velocity and future value.
Sales Velocity = (Number of Leads x Average Contract Value x Conversion Rate) divided by Time to Conversion.
“It’s now possible to know the quality of anonymous traffic per source – and take action!”
In 2017, Drift wanted to find ways to get website visitor data other than asking through a lead form. By using IP address tracking and various predictive analysis software platforms, Drift’s marketing team was able to build a model that predicts the intent of visitors, to win 79% of opportunities with just 16% of leads.
How G’s team leveraged these predictive data platforms initially:
- Get visitor’s IP address
- Find their company domain (Clearbit Reveal)
- Get a predictive score (MadKudu)
- Send emails to prospects
- Target high-value prospects with ads
- Engage users actively on the site with a live sales chat for high-value leads only (Drift)
- Change the content that appears on the site based on data you know about the prospect (Intellimize)
Discovering unexpressed intent:
After seeing so much success at accurately predicting lead scores with such limited data, the team at Drift decided to do more digging into other types of data they could find to help predict individual user behavior to optimize their B2B marketing strategy even further.
Here’s what they found:
- Datanyze lets you pull SDK install history to target your competitors’ customers.
- Product Hunt has an API that lets you pull all upvotes on any post and return Twitter handles.
- G2 Crowd, a review site for business software, has a premium subscription that gives you customer IP traffic data every time your logo is displayed on any of their pages (category pages, competitor pages, etc.).
- Bombora has partnerships with thousands of business content publishers. They collect and aggregate data on which content topics are visited by IP addresses. They can even provide a historical analysis on your customer lists to give you an idea of which topics the customers were reading about up to 12 months before converting, to create a topical path model.
- Apptopia offers mobile app data, including SDK installs and revenue figures.
The result? Predicting buyer intent before it’s expressed.
Most people look at the website as the top of the sales funnel, but they’ve made it the bottom of the funnel and moved the customer prediction to the very first touch, automatically.
If you centralize all data on a customer and make sense of that human BEFORE they visit your site, it helps you optimize and personalize content for them. As knowledge of individual sales velocity increases, so does the ability to spend more to close high-value leads without losing too much money on sales.
“In 2017, we observed behavior. In 2018, we understand and predict it.“
Marketing campaigns are usually late to the game. After a user leaves the site, we say “Come back please!”. By intelligently integrating all of the data mining platforms out there, you can now predict product-market fit to estimate user behavior and conversions based on user beliefs and user motivations.
If you can know the sales velocity of a specific user, you can optimize for the perfect timing to inject a promotional offer to reach the conversion tipping point and convert that user; all without giving away too much and without losing them.
What’s the martech stack look like for this?
In 2018, we can automate MARGIN.
The desired B2B marketing outcome shouldn’t only be conversions or leads. It should be max margin. Drift is no longer focused on predicting intent (done); they’re focused on predicting value. They plan to do this by building a system that selects the best action to take with each site visitor and lead based on the cost of that action and its potential (predicted) impact.
MadKudu can qualify and score customer fit based on positive and negative data signals associated with their IP addresses. Then, you can use that data to find more information about specific individuals’ profiles at the company to develop sales strategies tailored to them.
Example: using Crystal to build a profile on the company’s CEO to determine the best way to sell to them.
Drift’s evil plan to automate the world
Drift’s 5-year plan is to build a system that not only computes the next best action, but also its content. Based on all the aggregated signals, one day, content will be perfectly crafted and vastly superior to what a human could write using AI (putting many of us writers out of a job).
Finally, G left us with this speculation about the future of AI and sales to consider:
“We reject automated emails because, today, their value is lower than that of emails written by a human. But when we reach a point that AI-fueled emails are more valuable than handcrafted emails, how will that perception change?”