AI-led platforms proliferate the market, all promising better ROI, smarter placement of creative and channel selection, and easy hands-off creative development for instant results.
With these constant changes, it is easy to wonder if as practitioners we will be done out of a job. At a recent talk at SWSX 2018, John Kim spoke on the controversial topic of Love and the Algorithm. While this focussed around the importance of emotions and the human psyche in all decision making, whether this be in the marketing landscape or further afield, the overall takeout resonated strongly; human decision making remains central to everything we do.
Without such human interaction, automation platforms are simply engines spitting out meaningless results. Any algorithm must always have context; a platform to build on, a framework from which to develop and constant refinement and sense checking. This is why humans will always be needed, and context reigns supreme. No more so is this evident in the discussion of ‘data-led’ marketing.
At EightyOne and sister company DOT Loves Data, we constantly get asked “what is data-led marketing?” This is a question that has provided much debate; what do we actually mean by this over-used phrase? In much the same way that ‘social media’ is understood to the younger generations as simply ‘media’, there is a place to describe data-led marketing as simply ‘marketing’.
It is with this lens that EightyOne and DOT Loves Data chose to view this term. While marketing is the action or business of promoting and selling products or services, using available data to inform these actions is the next logical step.
This is what we think of as simply ‘marketing’. Marketing to us, is the act of better understanding our customers and using this information to be smarter about what we do. It was with this lens that the idea to pursue and understand the much-discussed Facebook algorithm was born.
THE RISE OF ENGAGEMENT CLUTTER
A common way of measuring whether the right digital content is served to the right people, is via ‘digital engagement’ – the act of a consumer interacting with digital content in some way. This measure of engagement in the social landscape is a relatively cloudy benchmark that has taken many different definitional forms.
An increasingly cluttered digital media landscape has practitioners demanding a way to quantify the success of digital creative that extends beyond traditional models such as last click and media-mix models.
There is no shortage of definitions describing digital engagement and theories on how to measure this from both an academic and practitioner standpoint. However, a uniform method to quantify or evaluate the effectiveness of digital engagement does not currently exist. This is partially due to the inconsistencies and lack of consensus on what consumer engagement is in a digital paid media context.
The lack of a methodology to measure digital engagement and the increasing redundancy of content success metrics, led EightyOne and DOT Loves Data to develop a framework that effectively quantifies content engagement and identifies a more robust measure of success using digital media data, specifically on Facebook.
This quantifies digital engagement associated with creative content and isolates factors that significantly affect engagement from a statistical and intuitive standpoint.
We believe this measure is crucial.
WHY IS THIS NEEDED?
EightyOne and DOT Loves Data developed this methodology after acknowledging the inconsistency in the industry and understanding that there was a need to better measure, plan and understand how to create authentic social content.
At present, optimisation of content via clicks is a key metric in lieu of an alternative way to measure ‘true’ engagement. Digital success metrics such as click through rates are becoming increasingly irrelevant to advertisers and brands, and marketers are demanding more robust and measurable ways to quantify digital success. Because of this, it is growing increasingly important to better understand the way engagement is quantified and measured.
SO, CAN YOU ‘BEAT’ THE ALGORITHM?
As such a widely used and discussed channel, we wanted to understand whether we could in effect ‘beat’ Facebook’s EdgeRank algorithm, which is Facebook’s newsfeed ranking system.
This algorithm decides which posts appear in a user’s newsfeed, by finding the most ‘authentic’ content.
We aimed to explore how behaviours and reactions to a Facebook post could be combined to derive a more effective, prolonged measure of engagement. This
could then aid today’s marketer in developing authentic social content.
Facebook’s 2018 media release regarding the change in their algorithm to prioritise authentic content has shone a spotlight on this issue. The ability to quantify authentic reactions to content has become increasingly relevant because of this, and the ability to create and shape creative to drive more authentic reactions and engagement is incredibly timely.
For a brand’s content to be prioritised on a user’s feed, it now needs to be deemed ‘meaningful and authentic’. Therefore, it is highly important that engaging content draws on more insightful interactions such as reactions, comments and shares, and not just clicks and likes. With this in mind, there has never been a more relevant time to explore a framework to quantify an emotional reaction to a brand’s post.
The EdgeRank algorithm considers three elements: Affinity, the closeness of the relationship between the user, the content and source. Weight, the importance of this action taken from viewing the content. Time Decay, how recent and current the content is to the user.
The Facebook algorithm prioritises active interactions, i.e. sharing, comments and reactions, over passive interactions such as likes and clicks. Facebook stipulates that actions that require more effort from the user deliver more authentic interactions and therefore are associated with more engaging content. Therefore, a post that delivers a more engaging experience results in a more emotive response, and consequently a greater level of interaction. This content is then prioritised on a user’s feed.
A measurement framework that outlines how to generate more engaging, authentic content is crucial. Enter the new Engagement Framework.
THE ENGAGEMENT FRAMEWORK
This framework considers the consequences of engagement that can be inferred through the types of interactions, passive or active, associated with a Facebook post. Developed from an existing engagement framework that focussed on collating experiences that led to consequences associated with digital engagement, this revised framework adopts the idea that a collection of experiences can be aggregated to produce a single engagement score.
This considers three primary consequences of engagement associated with a Facebook post (refer Figure 1):
In doing this, an engagement score for an individual Facebook post is derived :
Corresponding ‘delta’ metrics are also considered, i.e. change in number of likes, between time A and time B. These delta metrics will introduce a time-component and represent how reactions and behaviours change. The metrics are then classified into three groups: Action, Context and Time. Action metrics refer to usage and attentiveness consequences, context metrics refer to affective responses and reactions to ad consequences, and time metrics refer to change in rate metrics. These are fed into the framework that considers and aggregates the most important metrics that affect post effectiveness.
This produces an engagement score that measures how a user interacts with a post, and how we could quantify this to improve and build from, with future creative testing.
A like, share or comment is no longer robust enough to determine if a user is engaged with social content. This framework allows us to develop content that we know is getting authentic responses and therefore will be prioritised on a user’s feed.
With the myriad of tools, terminology and AI platforms claiming to fix all your data needs, it is easy to get overwhelmed with how we can progress and make positive inroads in a supposedly data-led landscape.
However, with the right practical tools, human intuition and decision making around how to use data more effectively, it could be said that we can, indeed, challenge algorithms such as Facebook’s sans robots. In a landscape where machines leading a marketer’s arsenal is a very real prospect, human emotion, decision making and authenticity cannot be undervalued, both in our consumption of advertising and how we process and understand data-led insights.