A year ago, CMO David Hsieh of Qubole, a big data company, bought into the promise of account-based marketing, or ABM. The idea behind ABM is to tap into large data sets to come up with a list of potential customers in order to focus marketing dollars on the ones with the highest probability to convert.
Hsieh spent a lot of time testing and tweaking the data and algorithm to produce more accurate results that could be trusted by marketing and salespeople. He ran into anomalies and worked backwards to find out why they occurred and how they might be fixed.
Hsieh’s efforts paid off.
With ABM, Qubole acquired bigger customers with bigger budgets, and sales doubled. Not all of this growth can be attributed solely to ABM, of course, but enough to warrant high praise from Hsieh. “This is a transformational technology for marketing.”
“It also has a few pitfalls,” he says.
That’s been the story of ABM, a technology with huge business upside if you’re willing and have the know-how to get it to work right. While ABM took the digital marketing world by storm only a couple of years ago, a series of misconceptions, implementation challenges, cultural hurdles, and elusive returns stymied adoption.
“ABM in 2017 was more pilot than progress, extending the hype and allowing more vendors to attach the ABM moniker to their wares,” says Forrester analyst Lori Wizdo.
The dust should settle next year, as ABM matures and digital marketers get better at using the technology. By 2019, four out of five B2B marketers will have at least one ABM campaign in the field, IDC says. ABM might be difficult, but it’s too disruptive to ignore.
Qubole had a head start in its ABM journey due to the nature of its business. The six-year-old Silicon Valley tech company has a platform running in public clouds like AWS and Microsoft Azure that helps customers process big data. This gives Qubole the advantage of knowing how data comes up with insights.
In fact, Qubole chose Demandbase for ABM because of Demandbase’s broad number of data sources. Using cookies and other methods, Demandbase scours the B2B universe for companies and tracks employee browsing behavior – what they’re looking at by keyword and in what volume – to learn real-time buyer intent signals, Hsieh says.
Then Demandbase aggregates and cross-references this data against Qubole’s customer profile to produce a list of potential customers, scoring each one on the probability of becoming a Qubole customer. “We were able to size our universe, roughly 5,000 companies,” Hsieh says.
Additionally, when an employee from a company on the list goes to Qubole’s website, a Demandbase widget generates content recommendations based on the collected data about the company and employee buyer intent signals.
All of this sounds straightforward, but it’s not.
Problems can arise in two places: an incomplete customer profile or faulty algorithm. Maybe a technology term or use case showing strong intent is left out of the customer profile. Maybe a keyword with multiple meanings is producing anomalous results. Maybe the algorithm is weighing certain terms incorrectly or not at all – that is, there’s a blind spot.
Hsieh often spot checked Demandbase results against real results. Sometimes, he would work backward and run a test with an existing customer. In theory, the existing customer should be putting off strong buyer intent signals and have a high score on the list of potential customers.
“We have found cases where an existing customer wasn’t even on the list,” Hsieh says. “That’s a hugely bad sign, and we have to figure out if this was a problem in the profile we fed the algorithm or a problem in the way the algorithm is processing the data. We’ve found both.”
After a year of fine-tuning, Hsieh believes the Demandbase results are more accurate, more reliable. But it’s hard to know for sure. Constantly shifting product lines and expansion into new markets create variables that impact the customer profile, potential customer list, and data that the algorithm trains on.
Last year, for instance, Quoble didn’t sell to financial services companies. Its product set lacked necessary security and compliance features. When those features were added this year, Quoble sought financial services customers. Hsieh had to tune and test the customer profile and algorithm so that Demandbase results would reflect this new market.
In ABM, continuous testing is the name of the game.
Another hurdle tripping up ABM adoption is the culture of marketing.
ABM changes the way marketers think about their jobs. In the old days, marketers would gather in a room to hash out market segmentation and come up with a set of customer demographic and firmographic attributes. Then they’d convert these attributes into a set of actual companies and market to them.
“It’s kind of guesswork,” Hsieh says.
ABM takes away the guesswork, and marketers pushed back. Some marketers simply relabeled what they were already doing as ABM without making any changes, Hsieh says. Others didn’t commit to making ABM work correctly.
In order to be successful with ABM, though, marketers have to overcome decades of ingrained ‘best practices,’ embrace their new role as data and algorithm testers, and ultimately trust the accuracy of ABM-generated insights.
“That’s what makes ABM a complete game changer,” Hsieh says.
Tom Kaneshige writes the Zero One blog covering digital transformation, AI, marketing tech and the Internet of Things for line-of-business executives. He is based in Silicon Valley. You can reach him at [email protected].