Home

IAB Data-Driven Video Best Practices

Download

Project Overview

As data fuels the digital advertising ecosystem and video consumption continues to rise, data-driven video is emerging as a powerful marketing tactic and strategy that enables brands to deliver innovative storytelling. The goal of this best practices document is to illustrate how marketers can utilize data to inform not only targeting efforts but the actual video assets and creative messaging. This document highlights the state of data-driven video, marketer benefits, tips for how to get started, best practices, and more.

IAB compiled data-driven video case studies to bring to life how data can be used to inform the video creative itself. These case studies showcase a variety of tactics and span across multiple brand verticals. Click here to view the video assets and learn about the campaign goals, data sets that were utilized, and results.

Data-Driven Video Overview

Data-driven video (DDV) is both a strategy and a tactic that allows marketers to use available data to deliver tailored advertising through precise audience targeting and personalized video creative that allows for permutations based on signals about the audience or other external factors. Underlying this strategy are two fundamental questions: who do you want to show this ad to and what messaging will move this individual to take a step towards the next best action?

(Source: Flashtalking)

Growth of Data-Driven Video

IAB Data-Driven Video Best Practices 1
Benefits

  • Consumers like personalization: According to an Epsilon study, 80% of respondents indicated they are more likely to do business with a company if it offers personalized experiences and 90% indicated that they find personalization appealing (Source: Epsilon, The power of me: The impact of personalization on marketing performance, 2018)
  • Personalized video can result in more efficient media planning: By serving more relevant ads, marketers will be able to serve fewer, more impactful impressions.
  • Data-driven video drives engagement: According to Innovid, there was a 78% lift in engagement rate on DDV campaigns compared to traditional pre-roll video ads (Source: Innovid, 2018 Global Video Benchmarks).
  • Marketers pursue DDV to drive sales and lifetime value: According to a Verndale survey, personalization is most important for increasing sales and improving customer satisfaction and retention (Source: Verndale, Jan 2018). In addition, a SundaySky study revealed that personalized video experiences can improve Net Promoter Scores, which measure customer loyalty to a brand, by 48 points or more (Source: SundaySky, A Study of Personalized Videos Deployed by F500 Brands, February 2019).

Data Signals Used to Inform Targeting and Creative

  • Contextual: page-level content via semantic analysis (i.e. headlines, article content) or video-level content
  • Demographics (age/sex)
  • Device level data (device make/model, browser, operating system)
  • First-party (CRM, product feeds, etc.)
  • Location data
  • Prior ad exposure (enables sequential messaging)
  • Psychographics
  • Purchase history
  • Real time data (sports scores, stock tickers, weather, etc.)
  • Site behavior (browsing history, cart information)
  • Viewer Behavior (interactions, engagement, time spent with interactive elements)
  • Viewer Information (first name, birth date)
  • And more

Considerations for Implementing a Data-Driven Video Strategy

IAB Data-Driven Video Best Practices 5

Capabilities Differ Based on Channel and Platform

IAB Data-Driven Video Best Practices 6

Best Practices

  • Data informs both audience and creative: Use data to identify different audiences and apply meaningful creative executions for each audience segment.
  • Always consider where your ad will be seen: Think about how the video ad will fit into the multi-screen ecosystem, take advantage of each screen’s strength, and consider the technology you can leverage to personalize your video.
  • Think about the consumer journey: Messaging should change throughout the consumer journey. Feature the next best action for the viewer based on their engagement with the brand.
  • Ensure data strategy is compliant with privacy regulation (i.e. GDPR, CCPA): First-party data will be most impacted by privacy regulation so marketers should build a strategy that is not 100% reliant on first-party data. One suggestion is to create a cascading logic for your data strategy. For example, first-party data can be the priority data set but if the marketer can’t use it, they can then flow down to another type of data like contextual.

Challenges

  • Data management: Keeping first-party data organized and determining how to segment your own audiences can be challenging.
  • Data quality: It is necessary to vet third-party data sets to ensure that the vendor’s segmentation and modeling methods are sound and data is fresh. For more info, visit datalabel.org.
  • Lack of resources: According to a Sailthru survey of UK and US marketers, more than 4 in 10 respondents said that a lack of time, people and money has inhibited their personalization efforts (Source: Sailthru, October 2017). Note: Vendors that offer managed services can alleviate resource challenges and help get personalization programs off the ground.
  • Regulatory issues: As much as marketers are focused on reaching the right people, be mindful of also avoiding the wrong people (i.e. follow COPPA regulation, Legal Drinking Age (LDA) compliance)

Ad Delivery Considerations

  • VAST: VAST 4.x (i.e. VAST 4.0 and beyond) introduces the concept of “Ad Requests” and “VAST Interactive Templates” – both of which can be used to support some level of dynamic content.
  • VPAID (Note: The spec for interactive ads will transition from VPAID to SIMID): VPAID enables more complex levels of interaction for data-driven video. For example, VPAID ad units can change in response to user interaction.

 

 

Measurement

  • Go beyond the click: Traditional KPIs like video completion rate and clicks should be balanced against more accurate determinations of success, like conversions for performance-based campaigns, and time-spent, for branding and awareness campaigns.
  • Establish benchmarks: Marketers can create benchmarks by conducting A/B testing. For example, run a generic version of the video first to establish benchmark for drop off and conversion rates. Then run the optimized video to understand how it performs against the benchmark.
  • Test and iterate: Check campaign performance on a weekly or bi-weekly basis and adjust your strategy based on performance. Testing can be as simple as an A/B test or as sophisticated as a multivariate test.

Data-Driven Video Case Studies

AccorHotels

Previous Case Study
Next Case Study

CAMPAIGN GOAL

To raise AccorHotels’ awareness across the region (APAC) and promote the brand's new holiday packages. The full-length creative that was 2 minutes and 30 seconds long wasn't performing well and the brand was not seeing desired completion rates. Unruly used Short Fix to cut the creative down to a mobile optimized 10s - opening up new audiences for AccorHotels. Unruly also used their targeting tool UnrulyEQ Custom Audiences to break down the brand's broad target demo into three unique audiences who would be most responsive to the content.

DATA

Unruly used data from proprietary content testing tool UnrulyEQ to identify the most impactful moments from the full length creative (impact was measured via a combination of facial coding and consumer survey) to cut it down to just 10s. Unruly used more emotional data to identify the most receptive audiences, then lookaliked them and targeted them at scale across our network.

RESULTS

4,294,039 completed views of new short form targeted creative across the region.

Previous Case Study
Next Case Study

Cadillac XT4

Previous Case Study
Next Case Study

CAMPAIGN GOAL

Cadillac launched a new vehicle (the XT4) in the market and needed to target a new demographic for its brand: Millennials. Given this new territory, they were challenged to deliver audience insights & learnings in real time and use these insights to adapt planning in-flight.

DATA

Throughout the 10-week campaign, Teads leveraged an agile flighting methodology to improve campaign outcomes and inform further creative decisions in real time. Teads started with 11 existing video assets which they turned into 330 iterations over the lifetime of the campaign. There were five 2-week sprint cycles where Teads tested 66 versions of the creative at a time that were targeted across 3 different audiences and both genders.

RESULTS

There was an increase in CTR of 45% (compared to the baseline, original creative). There was also an increase in VCR by 15%, which was considered an impressive metric when factoring in some natural campaign fatigue with the original creative appearing across many channels during the 10-week period.

Previous Case Study
Next Case Study

Carhartt

Previous Case Study
Next Case Study

CAMPAIGN GOAL

Drive traffic to Carhartt’s web site through interactive features in order to drive sales and raise awareness

DATA

Viewer behavior. Insights were gained regarding what specific Carhartt clothing items consumers were interacting with in the original Carhartt video asset. The viewer was served a sequential ad based on what Carhartt item they interacted with. The personalized video example provided was served to someone who interacted with the Carhartt jacket in the original video.

RESULTS

  • CTR: 2% CTR on personalized video versus 1% on original asset
  • Interaction Rate: 7% on personalized video versus 5% on original asset
    • “Interaction Rate” Definition: Sum of all brand interactions within the video (against impressions). Interactions include scene saves, frame selects, object highlights (which have a qualified timer), object links, button clicks.
  • Average time spent interacting with video: 45 sec with personalized video versus 30 seconds with original asset
    • “Average time spent interacting with video” definition: It is the difference between the first interactive time stamp and the last active time stamp within the player. Examples usually begin with either the Scene Save or Frame Select and the last time stamp is either a complete, a skip or a link out exit.
  • Average time spent interacting with objects: 127 seconds
    • “Average time spent interacting with objects” definition: The first time stamp fires when a user either clicks or taps on the object within the paused frame, when the user exits the text box or hovers off, there is an exit time stamp for this calculation. KERV Interactive then formulates the differences of these time stamps, by object, to determine average object level time spent or time spent interacting with a specific object.

Courtesy of KERV Interactive

Previous Case Study
Next Case Study

Conagra Brands – PAM

Previous Case Study
Next Case Study

CAMPAIGN GOAL

Conagra Brands understands that consumers are different—and thus purchase packaged foods for different reasons. In order to tailor messaging to be more relevant and emotional to target audience groups, they wanted to transform their messaging approach from “one to all” to “one-to-one".

DATA

Brand’s first party data (through Salesforce DMP)

RESULTS

  • 200 % lift in brand metrics
  • 190% average lift in engagement rate*
  • 32% average lift in video completion rate*

* Innovid metrics, performance compared to PAM’s same pre-rolls without personalization

Previous Case Study
Next Case Study

Cox

Previous Case Study
Next Case Study

CAMPAIGN GOAL

The goal was to increase online sales while lowering cost per acquisition (CPA). To do this, Cox partnered with SundaySky to deliver 15- and 30-second personalized in-stream, auto-play ads using Web Custom Audiences (WCAs) on Facebook. Each ad is individualized to the consumer based on his or her browsing behavior on Cox’s website, purchase history, and Facebook attributes to drive customer conversion. By leveraging dynamic creative elements and user data, both the video ad content and the creative look and feel are personalized for each individual viewer, delivering the most relevant ad experience possible

DATA

  • Last service / product / package viewed on-site
  • Page reached in registration process before abandoning site
  • Cox.com site visit recency & frequency
  • Facebook profile interests
  • Content viewed on Facebook

RESULTS

Decreased Facebook CPA to a cost that was 193% better than the target CPA

Previous Case Study
Next Case Study

eBay

Previous Case Study
Next Case Study

2nd eBay Personalized Video (Spanish language personalization)

CAMPAIGN GOAL

eBay’s goal was to launch several creative assets during the holiday season. Third-party data was used to optimize the campaign and drive sign-ups and sales on eBay.

DATA

Oracle BlueKai

RESULTS

n/a


3rd eBay Personalized Video (personalized based on user interaction)

CAMPAIGN GOAL

eBay’s goal was to launch several creative assets during the holiday season. KERV’s scene and object interaction data was used to optimize the campaign and drive sign-ups and sales on eBay.

DATA

User behavior. The viewer was served the personalized video ad if they interacted with the “toy” scenes from the original video asset.

RESULTS

n/a

Previous Case Study
Next Case Study

Estée Lauder: La Mer

Previous Case Study
Next Case Study

CAMPAIGN GOAL

One of Estée Lauder’s premier brands, La Mer, needed to stand out in an increasingly saturated beauty market, driving brand recall for their Crème de la Mer Moisturizing Crème in the process. Focused on driving customer engagement, La Mer aimed to create deeper relevance by deploying a personalized, data-driven video strategy across YouTube TrueView.

DATA

Site behavior (i.e. what did user search on YouTube)

RESULTS

  • 47.47% completion rate vs. YouTube CPG Beauty benchmark of 34.45%
  • $0.03 CPV, down from a benchmark of $0.05
  • +43% lift in overall brand recall
  • +72% lift in brand recall vs. YouTube benchmark
Previous Case Study
Next Case Study

F&P

Previous Case Study
Next Case Study

CAMPAIGN GOAL

Qualified clicks to site and time on site

DATA

Viewer behavior, KERV Interactive targeted specific groups of interactors and excluded users who had already clicked out of the video asset.

RESULTS

  • CTR: 2.61% CTR on personalized video versus of 0.14% on original asset
  • Interaction Rate: 20% Interaction Rate versus 0.62% on original asset
    • “Interaction Rate” Definition: Sum of all brand interactions within the video (against impressions). Interactions include scene saves, frame selects, object highlights (which have a qualified timer), object links, button clicks

See screenshots below to see the personalized roll over text:

F&P 2

F&P 3

Previous Case Study
Next Case Study

Samsung Family Hub

Previous Case Study
Next Case Study

CAMPAIGN GOAL

This Italian language campaign’s goal was to generate engagement. Samsung compiled a unique video but it was divided in 3 parts to control the user interaction response – “Controlla” (i.e. Check It Out) and “Compra” (i.e. Buy) were 2 different endings that the viewer could receive.

Note: The video sat in a high-engagement rich media unit (see screen shot and page linked here as well) with different custom events to gather interactions and first party data.

AccorHotels 5

DATA

  • 1st party publisher data about users who are interested in tech products and high spending
  • DMP 1st party data on users who visited the Samsung website to look at fridges and that made some action such as “Find Store” and “More Info”
  • 3rd party data: high spending users

RESULTS

  • Engagement rate 90.81%
  • CTR 15.80%
Previous Case Study
Next Case Study

Download Data-Driven Video Best Practices

Read The Blog Post

Join 400+ attendees from across the digital marketing ecosystem at the inaugural IAB NexGen Marketing Conference and learn how modern marketers are utilizing data, mobile, and video-driven marketing strategies to increase consumer engagement and accelerate business growth.