If you want to understand people’s needs and interests look at their intent. You no longer have to guess what they want, they will tell you. This session will focus on the growing importance of how machine learning is revolutionizing marketing from a demographic-based targeting endeavor to one rooted in the more effective method of mining consumer intent in real-time. And how delivering the most relevant and resonant ads to consumers converts sales at a greater frequency than with traditional methods.
2. Agenda
• The New Digital & Mobile Shopper
• Understanding Consumer Intent & Motivation
• Election Intent Learnings
• Active Intent Drives Performance
3. The Digital Influence Factor
Source: Deloitte, Navigating the New Digital Divide, May 2015
0
10
20
30
40
50
60
70
80
90
100
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
14%
36%
49%
64%
Percentageofin-storeinfluencedbydigital
Digital Influence
Projection
4. Search Users & Mobile Search Users
Source: eMarketer, 2016
Looked for early inspiration and made initial discoveries online on smartphone
Compared choices online on smartphone
Sought advice online on smartphone
Prepared online for immediate offline purchase
157.3
221
220.6
243
0 50 100 150 200 250 300
2015
2020
Search Users Mobile Phone Search Users
5. Mobile Impact on Path to Purchase
Source: eMarketer, June 2015
Ways in which US smartphone users used their smartphone during the path to purchase
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
Looked for early inspiration and made initial discoveries online on smartphone
Compared choices online on smartphone
Sought advice online on smartphone
Prepared online for immediate offline purchase
6. ”Personalization is the underpinning of every
experience that we want to create, within an activation.
I really can’t say enough about how important it is to
deliver a message that makes sense to the right
shopper, in the right context.”
-Anthony Acerra, Director of Digital Strategy
At Digital Ad Agency Rockfish
18. Strategy
Identify BTS Mom’s with promotional messaging to
drive sales at Staples during BTS
Yieldbot targeted relevant intent around shopping,
parenting, and education to serve Staples’ message at
the most relevant moment
Custom Native and Responsive card units were used to
match messaging with relevant intent
Results
2x ROAS
Acquired new customers through intent targeting
Positive results across desktop and mobile
Campaign extended and renewed the next year
Staples E-Commerce Case Study
Results
As mobile usage becomes ubiquitous, the path to purchase is becoming less defined. Consumers are always connected, well-informed and often quick to convert both digitally and in-store.
Our company was built on the belief that if you can identify someone who is raising their hand wanting to talk and you can match that intent with a highly relevant conversation then you can be increase your likelihood to be more successful at point of sale. If you can interpret the signals properly in digital media you can influence consideration in a way that is as meaningful as getting your end cap or in store promotion right.
And the proof is in the results. When you find active intent and message to those shoppers properly they will consider you more and they will buy from you in store.
If you could talk to someone raising their hand right now, wanting more information about your brand, would you ignore them? Of course not
With Yieldbot we enable those premium interactions for brands more than 21B times per quarter across the desktop and mobile web. In fact, 60% of our business today is mobile offering you a huge opportunity to match active intent on the most important device people are using.
<If we know brand KPIs this is where we should translate our ability to solve for that. If we don’t we find them out here before we go any further>
Beyond the performance Yieldbot delivers, the structure of our platform presents several additional benefits to you. The first one is the ability to give you First Look access to inventory. We sit inside the header of nearly one thousand different properties and read every page on those sites that someone visits. This is important because you get the first option to engage with that person if they are the right match for your brand right then.
We do that through our IntentRank engine. This is our algorithm that allows us to determine what a person is seeking in that moment. Every person gets a set of Intent scores based on their current session. That allows us to match the opportunity for a brand to the individual in that moment, based on that moment NOT something they did in the past or what someone who is similar might have done.
When we find that intent you want, the system then decides the best message from the creative elements loaded into the system to serve against that person. Because we understand intent and location we can better target the creative, and when that happens you get results like these (CLICK)
We built the Yieldbot platform to operate without cookies, to only look at the active session of the user and the pages they were visiting right now. For that to work we began by implementing our analytics and ad serving code directly into the header of every publisher. This allows us to look at the point of entry for a person, the content and context of the page they are on and more than 70 other variables including time of day, day of week and device type. This is what forms the basis for our IntentRank system. It’s how we know what has happened before on that publisher network and what the person in that moment will interact with from a brand.
In the days following the presidential election many people have asked “How did nearly all pundits get the outcome so wrong?” From Fivethirtyeight.com to the New York Times the Election Day models showed Hillary Clinton with a projected probability to win from the mid 60% to over 80%. And yet, when Election night became the next day it was Donald Trump that reached the requisite 270 electoral voters.
At Yieldbot, our technology provides the unique ability to analyze data on individual intent on over 1000 publisher properties in real-time from Facebook (and all other referring sources).
We determine intent by understanding an individual’s context (mobile, desktop, time of day, geo) and the words in links, URLs and pages referring people to a URL. Using machine learning, we then measure behavior across this massive data set at the page level (and the ad slot level for our advertising clients). The resulting data is a real-time IntentRank™ unique to every visitor to a page.
This allows us to be uniquely positioned to understand the motivations individuals place on a given topic. We are not beholden to a single source for this data which prevents human errors that can be associated with a specific expectation of outcome or past behavior. The individual expresses a view and the algorithms determine relevancy, human bias be damned.
These results are in stark contrast to mainstream media coverage of Trump that was measured as overwhelmingly negative. This might best exemplify the difference between the content that was created by the media and the content the electorate chose to consume when coming from Facebook. On a percentage basis, people were a lot less interested in the stigmas associated with Trump then the stigmas associated with Clinton.
Conclusions
In retrospect it seems that social media consumption was an excellent indicator of the pulse of the electorate. Much was made of the social media conflicts between Democrats and Republicans. Friends and family members blocking each other over political views or the trolling of individuals on Twitter. This assessment of the social media echo chamber is a common mistake in evaluating consumers. Like-minded people seek out those who look and act the same. Brands seek the next customer who behaves just like the last. Preconceived notions misshape future actions.
If you want to understand people’s needs look at their intent. You no longer have to guess what they want, they will tell you. This should remind us to think about the need of the consumer, not the need of the industry or the brand. There is an opportunity to better understand consumer motivation and intent rather than trying to move product.