In the last post we touched upon an approach to measure the impact of digital media campaigns, this post takes a more detailed look at a few ways to measure this. Two of the terms we will hear more and more frequently this year are ‘People-based Measurement’ and ‘Brand Lift Studies’ (BLS). While neither of these is new – they first emerged in 2013-14 – the increasing clamour for effective measurement of impact and adoption of new technologies will mean you’ll see greater adoption and use of these.

People based measurement

The bane of digital advertising is the anonymity of the online user. While ad serving technologies have vastly improved, a few factors have more than counteracted their effect; primary among this is the fragmentation of media consumption across multiple devices and micro-channels. This fragmentation has led to distortions in online ad campaign measurements like overstating of reach, understating of frequency etc. All leading to wastage of ad spends and lower ROI.

The solution to this fragmentation is people-based measurement. Simply put, people-based measurement is the use of unduplicated, person-level data to plan and measure the impact of marketing activity across devices and platforms.  It typically involves uploading user data from first party databases, particularly email ID’s to an ad server which can then be used to target ads and build audiences on. The terminology was popularised by Facebook which termed the Custom Audiences feature in the Atlas platform as people-based measurement; Google jumped in with its Customer Match, a few months later.

Cookie vs People

Image Credit: Datalicious

The benefits are immense of taking this approach. Optimised media planning in terms of targeting ads to a specific person, frequency capping and retargeting across platforms are the obvious uses. Opportunities open up for creative optimisation as well, where different creatives can be shown sequentially or even basis the micro-channel the user is on. Both these lead to better campaign ROI, but the ultimate goal is better attribution and omni-channel understanding of the consumer.

The underlying catch to using this approach is that brands/companies need to have their first party data (people contacted, sampled, call-ins, purchase, shopper etc.) captured and cleaned. This is something that many companies in India are still lagging in and find challenging.

Brand Lift Studies


Image Credit: Nielsen

Another solution that is gaining traction of late is Brand Lift testing. Marketers are more accustomed to traditional survey based researches that measure higher funnel metrics like brand awareness, ad recall, attribute association etc. Every ad influences these brand measures and sales happen in the context of changes in these. After years of tracking such measures, heuristics are formed about how these translate to lower funnel metrics such as purchase.

One of the ways to measure how online campaigns influence these higher funnel metrics is via Brand Lift Studies (BLS) which are being increasingly employed when large video campaigns are run online. Almost all major networks like Google and Facebook offer them either independently or via a tie-up with research agencies such as Nielsen or Millward Brown.

The premise is simple. Divide your online target group into 2 sets, those who have seen the ad – exposed, and those who haven’t – control. Ask the same set of questions to both and capture responses.  You can then evaluate the percentage point lift that an ad exposure caused to brand awareness, ad recall, consideration, purchase intent etc.

Brand Lift

Image Credit: Google

If the ad is running across media, say online and TV simultaneously, you can use traditional offline brand track to measure brand lift. One such study is described here. Utmost care has to be taken during design (both questionnaire and sampling), administration and analysis of such surveys else the results aren’t going to be reliable. For example, a few questions to think about would be… What exactly do we want to measure lift for? Are we capturing and attributing the recall to the correct media? Does the survey have enough statistical power to detect relationships? Do both exposed and control groups have the same characteristics?

Recently, surveys offered by major online networks have moved beyond assessing online only campaigns to measure cross-device lift across traditional and digital media – for e.g., Google, YouTube and TV and Facebook and TV. The big advantage of doing BLS online is that results can be used to optimise campaigns – for TG, frequency, creative, placement etc. – in real-time.

The obvious use of a BLS study is to understand how the campaign changed perceptions about the brand and to what extent. BLS studies do throw up some other questions though. Does a low lift mean that a campaign plan is sub-optimal or that the creative is not working? What is a good lift benchmark for my brand / industry / platform advertised on?

Unlike people-based measurement where you need to have a good Customer Data Platform, setting up a BLS is fairly simple, it doesn’t cost anything extra (if using the online solution) and uses the standard survey design for measuring impact of advertising that marketers are used to.

In conclusion, while marketers increase their investments into new capabilities for digital marketing, influencing offline sales and hence proving ROI still remains their number one concern. Described in this post are two ways that have emerged to help marketers drastically improve their understanding customer journey across devices and channels, and ultimately of campaign effectiveness.

  • Ravindra Ramavath