Databricks Optimizes Ad Performance With New LinkedIn Message Ads A/B Testing
August 20, 2020
Databricks is a San Francisco-based computer software company that offers a cloud-based platform that unifies data and artificial intelligence. Because of COVID-19, the company needed to shift an in-person conference in London to a virtual format. To help promote the new format, Databricks turned to LinkedIn.
While a virtual setting for its conference provided Databricks with the opportunity to broaden their pool of potential attendees, it also meant that the company had to quickly figure out how to identify and reach out to a broader set of individuals.
Databricks was already using LinkedIn to drive database growth as well as to build a pipeline of high-quality leads for its sales representatives. It made perfect sense to use the platform to promote and drive registration for the event.
“We wanted to find a personalized way to reach the right people,” says Justin Epstein, Databricks’ Senior Manager of Paid Media. The team decided to create a LinkedIn Message Ads campaign and filtered for profiles in Databricks’ target market with director-level titles and higher, but the cost per lead was daunting.
With LinkedIn we have the ability to target specific job titles, seniority, and several B2B attributes that we simply don’t get on other platforms. — Justin Epstein, Senior Manager of Paid Media, Databricks
The Databricks team opted to broaden its sights by expanding the target audience. “We tweaked the campaign to include more of the practitioner-level audience, integrating some member skills around data science and engineering,” Epstein shares.
In addition, the Databricks paid media team took the opportunity to engage in A/B testing with the message copy, a new functionality LinkedIn offers. Testing is a big part of the culture at Databricks. “Whenever we run any campaign, we always integrate some sort of experiment or A/B test and really let the data decide,” says Epstein.
Databricks tested two subject lines and three messaging iterations. In two of the messaging variations, the copy opened with a question. The third variation included a hyperlink in the first sentence of the body copy and stated the event details upfront.
All of the LinkedIn Message Ads variations had average open rates over 70% but click-through rates and conversions for the third version were almost twice as high as for the other two versions.
“The conversions were 2x higher and at half the CPA for the variation opening with the hyperlink and event details,” says Epstein.
Ease and value of testing
Testing is in Databricks’ DNA, and now LinkedIn makes it even easier on its platform. “With any big rock or priority campaign, we always build in A/B or split testing, because why not? You have the opportunity to learn,” says Epstein.
The team quickly realized that the Message Ads campaign “would be a great opportunity for us to test a personalized approach and understand how this format performs for us.”
The ease of testing ads and messaging on LinkedIn resulted in findings that Databricks can use to further refine campaigns, lead generation, and more. “Keep testing, extract insights, and use that to inform future tests so that you’re always evolving,” Epstein recommends.
Databricks didn’t give up after the initial Message Ads campaign targeting executives missed the mark. “If something doesn’t work at first, that doesn’t mean it’s a failure,” Epstein observes.
LinkedIn’s huge user base and its expansive filtering options for audience targeting gave the Databricks team confidence to take another stab at the Message Ads campaign. “We saw a huge difference when we really dug into the data and understood how we could drive our costs down by modifying our audience,” says Epstein.
Top funnel CPL is only the beginning. We take a deep dive into our BI tools and CRM to understand the many layers of lead quality. And overall we’ve seen great results when it comes to quality of leads on LinkedIn.
Data-driven audience insights
The results of the original Message Ads campaign and A/B tests gave the team vital information about the potential attendee profile for its conference—namely that engagement for decision makers wasn’t as high as they had hoped. “Although we were reaching a significant number of decision makers, we hypothesized that a lot of our event content would appeal more to practitioners,” says Epstein.
The Databricks team then shifted their focus from high-level decision makers to practitioners with new targeting, ad creative, and message variations, and saw much better results. “The data helped validate that hypothesis in performance,” Epstein says.
To get started on your own LinkedIn Advertising campaign, visit Campaign Manager today.