Exclusive video: The case for Facebook’s conversion lift tool

“It’s not rocket science,” said Smallwood. “[The medical sector] has been doing testing and control for 60 years now but bringing it to online marketing using the assets we have [Facebook] makes it revolutionary.”

According to Smallwood, the biggest difference between Facebook's conversion lift and the tools offered by companies like Google comes down to the nature of the platform and data. (See below for background information on what conversion lift is and how it works.)

“Google uses cookies to say whether it’s the same person or not compared to the control group, but it doesn’t work as well across devices,” said Smallwood. “They also can’t stop the control groups from being biased in how they select those groups to begin with.”

In other words, he believes that Google doesn’t have as much information on individuals to do control group selections. “People regularly clear their cookies anyway,” he added.  

EXCLUSIVE VIDEO: Smallwood states his case for conversion lift (shot at Facebook Hong Kong):

As Facebook aims to extend and fortify its walled garden, be it through its native and embedded video, real-time comments or messenger for business (announced at F8), its tools and measurement systems will evolve in parallel in order to make the world of digital marketing more Facebook-centric.

“Google is trying to do the same thing—they just don’t have the same set of assets as we do,” said Smallwood. “Google serves impressions to cookies, which by definition is very different to an actual individual.”

Smallwood contends that Facebook doesn’t just “cut the numbers” and takes measures to use all of Facebook’s data and dimensions to select random samples. This ensures, for example, that individuals from high-income groups aren’t selected in greater numbers to form a data bias.

The fact that Facebook users remain logged into the platform across devices increases the accuracy of insights, said Smallwood. 

Smallwood states that the other difference between conversion lift and other tools using control and test groups such as “home-scan panel solutions” which measure attitudinal changes, used by  Nielsen, Millward Brown, Alibaba and others, comes down to sample size.

“With these sorts of panels, that sample is small, so you have to run massive campaigns to see any impact,” said Smallwood. “I’m not saying it’s not useful, but it’s designed for big brand advertisers.”

With Facebook conversion lift, advertisers can select and run tests to find samples that are “enough to reach statistical significance”. However, Smallwood does admit there are some limitations.

“The Holy Grail for marketers is that they want to figure out exactly what ad or content led to a conversion,” said Smallwood.

“We won’t be able to know if a particular Facebook ad led a person to convert," he said. "But we’ll be able to figure out whether over a bunch of impressions, clicks and interactions led this person to take action.”

Why is Facebook measuring 'conversion lift'?

Conversion lift is Facebook’s attempt to address several measurement challenges facing marketers while promoting Facebook’s own benefits:

Relying on clicks: Facebook’s argument is that last-click attribution makes sense for search marketing, but is less useful for other digital or display environments such as Facebook. Conversion lift tries to “account for the value created by simply seeing an ad” by measuring the impact of exposure to an ad across devices on the Facebook platform.

Shift to mobile: The technology that supports current measurement systems include cookies to track exposure and tie to behaviour and clicks as a proxy for sales. According to the social network, as people use “multiple devices throughout the day and the majority of purchases still happen in a physical store”, Facebook’s social data allows advertisers to understand individuals rather than chase cookies.

Correlation-based testing methods: Conversion lift testing is based upon "the principles of lift measurement, a scientific approach used in a number of industries—such as direct mail marketing—to determine causation."

How conversion lift works in practical terms:

• When advertisers create a Facebook campaign, a randomised test group (people that see ads) and control group (people that don’t) are established
• Advertisers can then shares conversion data from the campaign with Facebook. This data may come from sources like the Facebook Custom Audiences pixel, conversion pixel or secure point-of-sale (POS) data.
• Facebook determines additional lift generated from the campaign by comparing conversions in the test and control groups
• The results of the study are made available in Facebook’s Ads Manager

Jenny Chan contributed to this report.